aus dem Fachbereich Geowissenschaftender Universität Bremen

Nr.135

Schmieder, Frank

MAGNETIC CYCLOSTRATIGRAPHYOF SOUTH ATLANTIC SEDIMENTS

~~~L--___________________ \

Berichte, Fachbereich Geowissenschaften, Universität Bremen, Nr. 135,82 Seiten, Bremen 1999

ISSN 0931-0800

Die "Berichte aus dem Fachbereich Geovvissenschaften" werden in unregelmäßigen Abständen vom

Fachbereich 5, Universität Bremen, herausgegeben.

Sie dienen der Veröffentlichung von Forschungsarbeiten, Doktorarbeiten und wissenschaftlichen Beiträgen. die

im Fachbereich angefertigt vvurden.

Die Berichte können bei:

Frau Gisela BoeIen

Sonderforschungsbereich 26 I

Universität Bremen

Postfach 330 440

D 28334 BREMEN

Telefon: (49) 421218-4124

Fax: (49) 421218-3116

angefordert werden.

Zitat:

Schmieder. F.

Magnetic Cyclostratigraphy of South Atlantic Sediments.

Berichte, Fachbereich Geowissenschaften, Universität Bremen, Nr. 135,82 Seiten, Bremen, 1999.

ISSN 0931-0800

Magnetic Cyclostratigraphy

of South Atlantic Sediments

Dissertation

zur Erlangung des

Doktorgrades der Naturwissenschaften

im Fachbereich Geowissenschaften

der Universität Bremen

vorgelegt von

Frank Schmieder

Bremen 1999

Tag des Kolloquiums:

12.2.1999

Gutachter:

Prof. Dr. U1rich BleilProf. Dr. Gerold Wefer

Prüfer:

Prof. Dr. Volkhard SpießProf. Dr. Heinrich Miller

"So ergibt sich ... ein Bild des Eiszeitalters, welches sich mit den Ergebnissen der geolo­gischen Forschung vollständig deckt. Mit dieser erfreulichen Schlußfolgerung soll nichtdas Bekenntnis unterdrückt werden, daß durch die vorliegende Lösung des Eiszeiten­problems ein neues Problem entstanden ist, nämlich jenes, warum der säkulareBestrahlungsgang, welcher sich durch alle Zeugnisse des Quartärs so unzweideutig zuerkennen gab, während der vorhergehenden geologischen Zeiten in Europa nicht zusolchen Vereisungen geführt habe wie während der letzten 800 Jahrtausende."

Milutin Milankovitch, 1936

Contents

Page

Chapter 1 Introduction

Milankovitch theory and cyclostratigraphy

Main objectives

Individual studies

References

3

5

8

Chapter 2 Using rock magnetic proxy records for orbital tuning and extended 11

time series analyses into the super- and sub-Milankovitch bands

T. von Dobeneck and F. Schmieder

Chapter 3 Mid-Pleistocene climate transition: initiation, interim state and termi- 44

nal event

F. Schmieder, T. von Dobeneck and U. Bleil

Chapter 4 Cycles, trends and events of Pleistocene sedimentation in the oligotro- 49

phic subtropical South Atlantic Ocean

F. Schmieder, T. von Dobeneck and U. Bleil

Chapter 5 Terrigenous flux in the Rio Grande Rise area during the past 1500 ka: 69

evidence of deepwater advection 01' rapid response to continental rain-

fall patterns?

F.X. GingeIe, F. Schmieder, 1'. von Dobeneck, R. Petschick and C. Rühlemann

Chapter 6 Summary and perspectives 81

Introduction

Milankovitch theory and cyclostratigraphy

The biggest breakthrough in understanding the

Pleistocene ice ages was the discovery that climatic

changes are strongly inf1uenced by orbitally induceel

variations in solar insolation received on Earth. AI­

though the seminal idea has been expressed earlier

this theary is inseparably associated with the name

of Milutin Milankovitch (1879-1958). He was the

first to compute in great mathematical detail the ele­

ments of Earth' sorbit and their effect on changes

in insolation (Figure 1). His complete astronomical

theory of Pleistocene ice ages (e.g., Milankovitch,

1936) fundamentally supported the basic idea that

cold summers in northern high latitudes are neces­

sary for the builel up of]arge ice shields (for a his­

torical review ofthe astronomical theory ofpaleo­

climate see Milankovitch, 1936; Berger, 1988 and

Schwarzacher, 1993a).

Variations in insolation mainly result from

changes in obliquity of the Earth's rotational axis

anel in precession ofthe equinoxes (Figure 2). Hays

et a1. (1976) showed that the corresponeling perioels

around 41 (obliquity), 23 and 19 kyr (precession)

can be identifieel in oxygen isotope recorels of ma­

rine sediments tagether with the near 100 kyr icc

age cycle dominating the late Pleistocene. For each

of these periods thc coherency betwecn orbital and

isotopic signals is statistically highly significant

(lmbrie et a1., 1984; Figure 3). The discovery that

paleoclimatic variations can be traced to exactly

preelictable periodical fluctuations of the Earth 's

orbit opened formerly unequaled possibilities far

detailed age modeling. Toelay, the once visionm'y

idea of cyclostratigraphy is realized by quantita­

tively corrclating pa1eoclimate signals or filtered

components to target recards constructed from or­

bital variations. During the last two decades this

'astronomical tuning' has proven to be one of the

most powerful tools for high resolution chrollOstrati-

Jonrlousende

qoc!,,{}(}(J !J5(} $00 tJ:IJ. t5~ T 7(}(J 550 i 550

1, ~ ~~ ~r ~ i ~ I t~ tU\? T 0

J:~ ,~,~.~,'\!~~ ·~~~~tts:~~~II@;i9;-c-·~, ~j~~;~ij;i::~ -j\I~ YI\' 'nv v ',J.. AA,"'\'!l1 f\.... ! ,Vr\.'II'fr-J r W\J'MVr\. i~rvT"AV r-.. V7\

~ 7ßO :'" y~; -y :tt-~~-,v.. --I\-, -t.~ \l T f--·\; -1 ~ ~, :" VI ~ .Y-75° I 1l 11 JY F 1'l J'JJ J1![ JX Ka.t Xl X!a .n .IlIl X1Y

Pliozün? OOll{](J·fiueil wnz-fiszeil Minde/-fisuil ffiß-fiszeif Wiif'fT1-füzeilFig. 16. StrahlungBdiagramm von M. Milsllk{)"it('h

.4bsfiindi' der Ab/agu{j{7gen in f'f!!a/ivtn gwhgisckn J#rfen. 1 2 3 q 5 6 7 8 910 11 I1CL 12 13 N 1$ J6i ~ ~~~H----Hf-~~':(

':'~i[-tl ~1'" ... ---- tt--- :f--------..---.- - a-:-lUY''/----/---Pl' '-MIi-tl~H1':fil4-·...,

'" C' --'11 • • •. '!'.......- ------lll---f-l'------IIII'--Iil!... ,,1,'":::..; 'l ?iJ _.-~ I -------1--~·-·- ~ . ~l , , 1-- ; 11' w

~] 3IJ .--' , .~--+-- -1- -----lil----11h------+---H---I,i+-,-:---l,

~~ qo!--~ : ~ I I ' .. ------:----.;------ ' -0-----r-~lro_:+.~.~ I I ~ 1 I i ( --r-'! T I !I \.~ ",>?f----r ---t-----+--r-r I I -+-----·-·------rw;r;;;;~m ii il~ &7f--1 ~-j-+++---._-~ --1·,-+----··· , I' ::.='-----!--~,~TI~,-l'<:; 70 A !i 01 Oll 0lJ[ 51 5l[ MI MIl 111 IIIlIlla. WI W1l WlI!

~er Sfarrfenbtrg lJonou·Sftldim 6ünz·Sfadien Hindel-SfCldien 1I1f)-J'lodien Würm-J'ladl",SdxJIfer ,>dKJ/Ier

Fig. 17. Stratigraphisches Diagmmm "Oll 13. Eb,,!

Figure 1. Milutin Milankoviteh's radiation estimates far latitude 50,55 and 65°N for the past 1000 kyr (top) eompared to astratigraphie interpretation ofEuropean iee ages (Würm, Riß, Mindel, Günz, Donau) by B. Eber! (bottom). From Köppen &Geiger (1936).

Figure 2. Insolation received on Earth depends on severalelements ofthe Earth's (E) orbit around the sun (S) varyingdue to gravitational effects of the planets and the moon.Eccenh'icitye = .j(;'-!/ / ([ defines the elliptical shape oftheorbit (presently e = 0.016). Obliquity (E, presently 23°27')measures the angle between the Earth's axis ofrotation (SN)and the perpendicular to the ecliptic (SQ) and is mainlyresponsible for the seasonal contrasts and the latitudinalgradient ofinsolation. WW and SS denote winter and summersolstice, respectively, y indicates the vernal equinox. Theinclination E ofthe Earth's axis follows a complex precessionalmovement along the cone shaped surface that produces amigration of the equinoxes scaled to the longitude 01' theperihel ion co. Today, the Earth is about at the perihelion (P,closest approach between Earth and Sun) during the northernhemisphere winter solstice. From Berger (1988).

graph1', especiall1' for the late Pliocene and Pleisto­

cene but b1' no means restricted to these epochs (e.g.,

Schwarzacher, 1993a, b; Shackleton et al. , 1995;

D'Argenio et al., 1998).

The tard1' triumph of the Milankovitch theor1'

over long-standing criticism (see Berger, 1988) is

documented impressivel1' in the delayed acceptance

ofthe orbitall1' tuned age ofthe Brunhes/Matu1'ama

boundary. Although earlier proposed b1' Johnson

(1982), the age of 780-790 ka became widel1' ac­

cepted onl1' after the 'second attempt' ofShackleton

et al. (1990) and Hilgen (1991). Cande and Kent

(1992) for example revised their magnetic polarit1'

time scale in 1995 (Cande and Kent, 1995) and made

it consistent with astrochronolog1', also in older

sections of the record. In the meantime, Wilson

2

N Q

Chapter 1

(1993) showed that astronomical calibration resldts

in a more concordant sea-floor spreading history

when applied to spacings of magnetic anomal1'

pattern in the Pacific Ocean. Still assuming an age

of 730 ka for the Brunhes/Matu1'ama boundar1'­

and therefore somewhat inaccurate in thc older part

- Imbrie et al. (1984) published their epoch-making

SPECMAP time scale for the last about 800 k1'r and

started a s1'stematic documcntation of the Earth' s

paleoclimatic history which since then has been

extended and refined graduall1' by astronomicall1'

calibrated 8 180 time scales (e.g., Ruddiman et al.,

1986; Martinson et al. , 1987; Ruddiman et al. , 1989;

Raymo et al., 1989; Shackleton et al. , 1990;

Tiedemann et al. , 1994; Berger and Jansen, 1994a).

The ratio ofox1'gen isotopes IBO and 1(,0 has been

applied as a paleoclimatic indicator far quite some

time (e.g., Emiliani, 1955). In most cases it is mainly

a measure of global ice volUll1e (e.g., Shackleton,

1967). In the context of cyclostratigraph1' ox1'gen

isotopes are b1' far the most frequentl1' used and

best studied prox1'. But measuring detailed 8 1BO

records is very time consuming and involves an

immense laborator1' work. In low sedimentation

rate, predominantl1' terrigenaus and/or deposits

below the CCD (carbonate campensation depth)

detailed anal1'ses may be seriousl1' hampered as it

is necessar1' to coHect a sufficient number ofcalcare­

ous tests from each sampie. Consequentl1', mainl1'

cores from regions with high (carbonate) sedimen­

tation rates are used to achieve detailed paleo­

climatic records. This strateg1' excludes large parts

ofthe worlds oceans holding valuable paleoenviron­

mental infonnation.

The outlined restrictions ofox1'gen isotope stra­

tigraph1' led marine scientists to explore adequate

alternatives. Since the discover1' that paleoclimatic

patterns are often mirrored b1' ph1'sical properties

like magnetic susceptibilit1' (e.g., Robinson, 1990;

Tm-duno et al. , 1991) and GRAPE (Gamma Ra1'

Introduction 3

-- - - -- 6'1'0 VAR 1I - - - ETP VAAff -+-+-+-+ COHE RENCY

Figure 3. Coherency and variance spectra calculated fromrecords of clil11atic and orbital fluctuations spanning the last780 kyr (Il11brie et a1., 1984). The ETP signal cOl11bines nor­malized variations of~ccentricity, obliquity ([iIt) ancl ~reces­sion. 8 180 is the unsl1100thed SPECMAP stack. From Bergeret a1. (1984).

10

-co

"oci~

.c

100 41 1

,/\ /\ ?' I_~ I g• -" '\ I .'. \ I .... ' ( \ I",' "'-', I,' \1 I,' " \ I 1 ;':" \ ,'J/ \1 1/ \ \ I \ 1 V)

\i I \';;""\f '\V,.... \, I, ,-~ BA"DW!DT>i I ~\1 11 I I '! "1', I cu

o I 11 11 \l, ~ .", " I~z. 11 \( I, "'" "'-.::., ". '"i;! ',' 1,1 '~--' - .... ,,~/~ ~\:- :,~, ~w 0.9 .....

6 "---+-+----:r-+-----t~-;:t_-r -j-",/5"'%--:-SiQ. level (wh) I >U 0.8 r -1

07 I

~~ -.-~~001 0.02 003 004 005 0.06 007 0.08 009 0 I100 50.0 33.3 250 200 167 143 12.5 \ I: \00

Attenuation Porosity Evaluator) density records

(e.g., Herbert and Mayer, 1991; Grützner et a1.,

1997) numerous chronostratigraphies have been es­

tablished on basis ofthese continuously measurable

proxies (e.g., Shackleton et a1., 1995; Chi and

Mienert, 1996; ShackJeton and Crowhurst, 1997;

Bickert et a1., 1997). The measurements are very

rapid and nondestructive and the high resolution

reachable makes these parameters especially

suitable both for the analyses of high frequency

variability and low sedimentation rate deposits.

Main objectives

During recent years Marine Geopbysics in tbe De­

partment ofGeosciences at Bremen University suc­

cessfully establisbed magnetic cyclostratigrapby as

a very efficient dating tool for marine sediment se­

quences in addition to long-proven magnetostrati­

graphic technique (e.g., Bleil and Gard, 1989;

Nowaczyk, 1991; Thießen, 1993; Nowaczyk et a1.,

1994; Frederichs, 1995; Gersondeeta1., 1997). The

age models for tbe sediment series presented in this

thesis are all based on orbitally tuned high-resolu­

tion magnetic susceptibility records, partly C0111­

bined with oxygen isotope data 01' based on Quater­

nary magnetostratigraphies. Advantages, difficulties

and restrictions of using rock magnetic records for

dating purposes are discussed in full detail in Chap­

tel' 2.

Although today Milankovitch's theory ofa link­

age between Earth's climate and orbital variations

is widely accepted and successfully applied chrono­

stratigraphically, the history ofPleistocene climate

variability still holds l1Umerous unresolved puzzles.

To resolve its cyclicity, located within the main but

also in the adjacent frequency bands above and

below the principal Milankovitch frequencies, re­

quires particularly adapted statistical methods.

Extensions of time series analyscs into the super­

and sub-Milankovitch bands are described in Chap­

tel' 2.

The central paleoceanographic topicofthis thesis

(Chapters 2, 3 and 4) is the mid-Pleistocel1e tran­

sition (MPT) of tbe global climate system (Pisias

and Moore, 1981; Prell, 1982; Ruddiman et a1. ,

1989, Berger and Wefer, 1992; Berger and Jansen,

1994b), the change from a predominantly 41 kyr

cyclicity in early Pleistocene to the late Quaternary

100 kyr ice age cycles. Variance in climate indices

at periods of precession of the equinoxes (with

periods in the range of 19 to 24 kyr) and obliquity

01' tilt (with major cycles around 41 kyr) can be

explained in the framework ofMilankovitch theory

as linear responses to changes in solar insolation

(e.g., Imbrie et a1., 1992). In contrast, the origin of

the 100 kyr ice age cycle dominating the late Pleis­

tocene (Figure 3) calls for more complex explana­

tions because the direct influence of eccentricity

(with major periods at about 95 and 124 kyr in ad­

dition to a somewhat stronger 413 kyr component)

on insolation is by far too small to produce the cor­

responding climate style (e.g., Imbrie et a1., 1993).

4 Chapter 1

~:~ ~~=loowMo 100 200 300 400

.------- t ---------..ECCEHffilCrTY BAND

HOOkyr)

]1------------OBUQlJITY BAND

(......11<yr)15

-15

~t~]-~~6180

o 100 200 300 400AGE (ko)

7S

...~

-15

-1

8"-0

Figure 4. The 100 kyr cycle problem. Partitioning radiation (top, data from Berger, 1978) and climate time series (bottom,data from Imbrie et al., 1984) into their dominant periodic cOl11ponents reveals a large discrepancy in the eccentricity band.While the three frequencies contribute to the insolation signal in proportion to 1,0.2 and 0.02, the respective ratios for climatevariations al110unt to 1, 2.5 and 11. From Il11brie et al. (1993).

This discrepancy between driving force and re­

sponse is illustrated in Figure 4.

Besides its pure existence, another enigma is the

more or less abrupt onset of the 100 kyr cycle ap­

proximately 650 kyr aga (e.g., Berger et aL, 1994;

Mudelsee and Schulz, 1997). Build up ofmajor ice

shields on the northem hemisphere already started

at about 2.5 Ma (e.g., Shackleton et al., 1984). While

late Pliocene to early Pleistocene paleoclimate

records reveal a mainly obliquity and precession

related variance (e.g., Raymo et al. , 1989; Ruddiman

et al. , 1989; Bloemendal and deMenocal, 1989),

variations with aperiod near 100 kyr are the primary

rhytbm oflate Pleistocene climatic change although

insolation vm1ations remained almost identical. The

exact timing of the MPT and the question whether

the transition from 41 to 100 kyr cyclicity was gra-

dual ('mid-Pleistocene evolution' ,Ruddiman et al. ,

1989) or abrupt ('mid-Pleistocene revolution', Ber­

ger and Wefer, 1992) were subject ofmultiple stud­

ies and attempts to model this shift with different

statistical tec1miques (e.g., Maasch, 1988; DeBlonde

and Peltier, 1991; Park and Maasch, 1993; Mudelsee

and Schulz, 1997; Mudelsee and Stattegger, 1997;

Clark and Pollard, 1998). Of the diverse explana­

tions proposed for the 100 kyr cycle, models invok­

ing a pC02

controlled insolation threshold for the

melting oflarge ice shields (e.g., Saltzman and Ver­

bitsky, 1993; Berger and Jansen, 1994b) currently

yield the best approximation of8180 signal pattem.

As illustrated in Figure 5, these models accomplish

a reconstruction ofthe MPT timing and the develop­

ment of 100 kyr cycles by introducing a decreasing

atmospheric pC02level (Raymo, 1997; Paillard, 1998).

Introduction 5

~

0)()'1

Q

400 Z

~3-'"

.....c:

450 ..:<I'\l-

, 500 ~2­llJc::',

g

""""",,,j, , I ,:, , , I , ,', , I , ; , ,I ,,' I ,:, , , I ;, , ,I '" I , ,,I ;" I:

Wisconsin lilinoian Kansan NebraakanWürm Riss Mindei Gim

I I IV V VII3

'"Il \

I0 4 I

'" \ (-<0

I I

-.~\ I

5 -'2

-200 o 200 400 600Aga (kyr)

800 1000 1200

Figure 5. Benthie oxygen isotope data from Paeifie ODP Site 849 eompared to insolation at 65°N on Jllly 21, The horizontaldashed line in the alsO reeord indieates the level belO\v whieh 'exeess 100 kyr iee' is observed during glaeials. The vertiealdotted lines mark the midpoint ofterminations. Assllming a long-term drawdown ofthe atmospherie pC0

2level, olltlined by

the sloping line in the insolation reeard, results in a ehanging threshold far the melting of large iee shields. During latePleistoeene a suffieiently high insolation is only attained at times of positive interaetion of obliquity and preeession at the endof the blaek shaded regions, Beeallse the amplitude of preeession is modlliated by eeeentrieity, 100 kyr eycles are produeed,eoineiding with several of the tenninations. From Raymo (1997).

With a few exceptions (e.g., Raymo et al., 1997)

so far all investigations of the MPT were based on

oxygen isotope records. The rock magnetic view

presented in this thesis adds some completely new

aspects for the understanding of this fundamental

enigma of Quaternary climate evolution. Aseries

of sediment sequences from submarine ridges of

the subtropical South Atlantic Ocean located in a

water depth alternately affected by North Atlantic

Deep Water (NADW) and Lower Circumpolar Deep

Water (LCDW) have been studied which recorded

the paleoceanographic history in a region of strate­

gie importance for the global thermohaline circula­

tion (e.g., Berger and Wefer, 1996).

Individual studies

This thesis comprises four manuscripts which are

in press or have been submitted for publication. The

studies were performed in the Marine Geophysics

Section of the Department of Geosciences in the

framework of the Graduiertenkolleg 221 "Stoff­

Flüsse in marinen Geosystemen", associated with

the Sonderforschungsbereich 261 "Der Südatlantik

im Spätquartär: Rekonstmktion von Stoffhaushalt

und Stromsystemen". Both research projects are

funded by the Deutsche Forschungsgemeinschaft.

The work has been supervised by Prof. Dr. Ulrich

Bleil.

A stratigraphie synthesis of twelve sediment

cores covering the deep subtropical South Atlantic

Ocean was established using magnetostratigraphic

and cyclostratigraphic methods. This regional

6 Chapter 1

chronostratigraphic framework fonDs the basis for

various research strategies and scientific co-opera­

tions - most intensely with Dr. Tilo von Dobeneck,

co-author of all four papers.

A first, methodological treatise intro duces

advanced signal analytical techniques enabling

paleoceanographic interpretations in the super- to

sub-Milankovitch ranges. Their immense potential

is exemplified by susceptibility records from the

subtropical and western tropical South Atlantic, the

latter dated by von Dobeneck (1998).

In the course of the mid-Pleistocene climate

transition (MPT) the dominant periodicity of the

climate system response changed from 41 to

100 kyr. Rock magnetic investigations identify this

period as an interim state of reduced carbonate

accumulation in thc subtropical South Atlantic. The

second paper interprets the MPT as aseparate cli­

mate state ofreduced NADW influence, tem1inated

by an ocean-wide event at 530 ka.

The third manuscript compares the temporal evo­

lution ofclimatic cycles in environmentalmagnetic

proxy records to driving orbital variations. Recipro­

cal exchange of spectral energy between the 41 and

100 kyr bands suppOlis models invoking a pC02

­

controlled insolation threshold for the occurrence

of 100 kyr glaciation cycles in late Pleistocene.

The final study is concerned with the tenigenous

input in the Rio Grande Rise area and primarily

based on clay mineral investigations by the first

author, Dr. Franz X. Gingeie. My own contribution

to this study consisted in establishing age models

for the cores and statistical analyses ofclay mineral

data. Additional, previously unpublished results

were provided by Dr. Rainer Petschick (clay miner­

als) and Dr. Carsten Rühlemann (oxygen isotopes).

In the following the four manuscripts are briefly

summarized. All data used in the thesis are archived

in the information system PANGAEA/SEPAN

(www.pangaea.de).

T. von Dobeneck and F. Schmieder

Using rock magnetic proxy records for orbital

tuning and extended time series analyses

into the super- and sub-Milankovitch bands

In: G. Fischer and G. Wefer

Use ofProxies in Paleoceanography: Examples

Fom the South Atlantic

Springer-Verlag, in press.

This paper outlines the methodical background of

the thesis. Two case studies are presented, both

based on a several sediment seq ucnces dated by or­

bital tuning of their high-resolution magnetic sus­

ceptibility records. Extended time series analyses

focus on the statistical evaluation ofdifferent pa1eo­

climatic aspects documented in the frequency bands

above and below the main Milankovitch cycles. The

Subtropical §outh Atlantic §usceptibility (SUSAS)

stack extents back to 1.5 Ma and thus provides ideal

conditions to study various superstructures, e.g.,

amplitude modulations of proxy responses. Com­

paratively elevated sedimentation rates in the Ceara

Rise §usceptibility (CEARIS) stack allow an exami­

nation ofhigh-frequency sub-Milankovitch pheno­

mena. Paleoceanographic implications are only

briefly discussed in this publication.

F. Schmieder, T. von Dobeneck and U. Bleil

Mid-Pleistocene climate transition: initiation~

interim state and terminal event

submitted to Nature

The article addresses a fundamental enigma ofQua­

temary climate evolution which recently has been

subject of several inspiring scientific publications.

So far, most investigations of the mid-Pleistocene

climate transition were based on oxygen isotope

records reflecting global ice volume. The analysis

ofrock magnetic parameters presented here focuses

Introduction 7

on changes in deep water chemistry. The data sug­

gest that the mid-Pleistocene c1imate transition from

about 920 to 640 ka does not represent a gradual

change from a 41 to a 100 kyr world, but rather a

third, quite different c1imate state confined by brief

shifts at both boundaries. Moreover, many of the

sediment series studied document an unusual ter­

minal event at the end of the mid-Pleistocene cli­

mate transition, possibly related to other yet unex­

plained findings in l1umerous paleoclimatic records

at around this time.

F. Schmieder, T. von Dobeneck and U. Bleil

Cycles, trends and events of Pleistocene

sedimentation in the oligotrophie subtropieal

South Atlantic Ocean

to be submitted to Paleoceanography

The submarine ridges ofthe South Atlantic Ocean

provide ideal opportunities to study paleoclimati­

cally driven temporal and spatial altemations of

NADW and LCDW as they intersect the vertically

fluctuating broad transition zone separating these

two deep water masses during glacial as weil as

interglacial times. Sediment series recovered at

twelve sites on these ridges were investigated for

cyclic fluctuations and trends of Quaternary depo­

sition. The results imply a linkage between changes

ofpredominant cyclicity, long-term trends in carbo­

nate accumulation and a paleoceanographic event

at about 530 ka documented at severallocations.

F.X. GingeIe, F. Schmieder, T. von Dobeneck, R.

Petschick and C. Rühlemann

Terrigenous flux in the Rio Grande Rise area

du ring the past 1500 ka:

evidence of deepwater advection or rapid

response to continental rainfall patterns?

Paleoceanography, 14, 84-95, 1999

The main objective of this study is to investigate

the usability of kaolinite as a tracer for NADW in

the Rio Grande Rise area. A comparison of kao­

linite/chlorite ratios in surface sediment sampIes

from the mid-Atlantic Ridge and the Rio Grande

Rise region reveals that in the continental realm the

Rio Doce (Brazil) is a major source of kaolinite in

marine sediments whereas the supply of kaolinite

by NADW is of minor importance. Time series

analyses evince cyclic variations of kaolinite/

chlorite ratios in one ofthe SUSAS cores coherent

with global ice volume in the 41 and 100 kyr bands.

They are interpreted to have recorded fluctuations

in discharge ofthe Rio Doce and to mÜTor humidity

conditions in the South American hinterland.

8 Chapter 1

References

Berger, A, Long-term variations ofcaloric insolation resultingfrom Ealih's orbital elements, Quat. Res., 9, 139-167, 1978.

Berger, A., J. Imbrie, J. Hays, G. Kuk1a and B. Saltz­man, Mi1ankovitch and Climate Part 1, Reide1, Dordrecht,510 pp, 1984.

Berger, A., Milankovitch theory and climate, Rev. Geophys.,26, 624-657, 1988.

Berger, W.B. ancI E. Jansen, Fourier stratigraphy: spectra1ga in adjustment of orbital ice mass models as an aid indating late Neogene deep-sea sediments, Pa/eoceanogra­phy, 9, 693-703, 199421.

Berger, W.B. and E. Jansen, Mid-P1eistocene climate shift­the Nansen connection, in: The po/ar oceans and theil"roje in shaping the global environment, edited by O.M.Johannessen, R.D. Muench and lE. Overland, Geophys.Monogr. Ser., 85, AGU, Washington, pp 295-311, 1994b.

Berger, W.H. and G. Wefer, Klimageschichte aus Tiefsee­sedimenten, Naturwissenschaften 79, 541-550, 1992.

Berger, W.B., M.K. Yasuda, T Bickert, G. Wefer, T. Taka­yama, Quaternary time sca1e for the Ontong Java Plateau:Milankovitch template for Ocean Drilling Program Site806, Ge%gy, 22, 463-467, 1994.

Berger, W.H. and G. Wefer, Central themes ofSouth At1anticcirculation, in The South At/antic: present {[nd past circu/a­tion, edited by G. Wefer, W.H. Berger, G. Siedler andDJ. Webb, Springer, Berlin, pp 1-11, 1996.

Bickert, T, W.B. Curry and G. Wefer, Late Pliocene to Holo­cene (2.6-0 Ma) western equatorial Atlantic deep-watercirculation: inferences from benthic stable isotopes, Proc.

ODP, Sei. Resu/ts, 154,239-254, 1997.B1eil, U. and G. Gard, Chronology and corre1ation ofQuater­

nmy magnetostratigraphy and nannofossil biostratigraphyin Norwegian-Greenland Sea sediments, Geol. Rundschau,78, 1173-1187, 1989.

Bloemendal, land P. deMenocal, Evidence for 21 change inthe periodicity oftropical climate cycles at 2.4 Myr fromwhole-core magnetic susceptibility measurements, Nature,342,897-900, 1989.

Cande, S.C. and D.V. Kent, A new geomagnetic polarity timescale for the Late Cretaceous and Cenozoic, 1. Geophys.Res., 97, 13917-13951, 1992.

Cande, S.c. and D.V. Kent, Revised calibration of the geo­magnetic polarity timescale for the Late Cretaceous andCenozoic,1. Geophys. Res., 100,6093-6095,1995.

Chi, 1. and l Mienert, Linking physical property records ofQuaternary sediments to Heinrich events, Mal". Ge%gy,131,57-73,1996.

Clark, P.u. and D. Pollard, Origin of the middle P1eistocenetransition by ice sheet erosion ofregolith, Pa/eoceanogra­phy, 13, 1-9, 1998.

D'Argenio, B.D., AG. Fischer, G.M. Richter, G. Longo, N.Pe10si, F. Molisso and M.L. Duarte Morais, Orbital cycli­city in the Eocene ofAngola: visua1 and image-time seriesanalysis compared, Earth Planet. Sei. Lett., 160, 147-161,1998.

DeBlonde, G. and W.R. Peltier, A one-dimensional model ofcontinental ice volume fluctlrations through the Pleisto­cene: implications for the origin of the Mid-Pleistoceneclimate transition, 1. Clim., 4, 318-344, 1991.

Emiliani,C., Pleistocene temperature, 1. Geol., 63, 538-578,1955.

Frederichs, T, Regionale und altersabhängige Variation ge·steinsmagnetischer Parameter in marinen Sedimenten derArktis, Bel". Po/arforschung, 164, 212 pp, 1995.

Gersonde, R., FT Kyte, U. Bleil, B. Diekmann, l1\ Flores.K. Gohl, G. Orah1, R. Hagen, G. Kulm, F.J. Sierro, D.Völker, A. Abelmann and J.A Bostwick, Gcologicalrecord and reconstruction of the late Pliocenc irnpact ofthe Eltanin asteroid in the Southern Ocean, Nature. 390,357-363, 1997.

Grützner, l, F.C. Bassinot and l Mienert, High-resolutioncompressional-wave velocity measurements in Pleistoceneseeliments ofthe Ceani Rise (western equatoriaJ Atlantic):implications for orbital driven seelimentary cycles, Proc.ODP, Sci Resu/ts, 154, 135-149, 1997.

Hays, J.D., l Imbrie and NJ. Shackleton, Variations in Earih'sorbit: pace-maker ofthe ice ages, Scienee 194, 1121-1132,1976.

Ilerbert, TD. and L.A Mayer, Long c1imatic time series fromsediment physical property measurements, 1. Sediment.Petrol., 61,1089-1108, 1991.

Hilgen, F..T., Extension of the astronomically calibrated(polarity) time sca1e to the Miocene/Pliocene bounelary,Earth Planet. Sei. Lett., 107, 349-368, 1991.

Imbrie, J., 1D. Hays, D.O. Martinson, A. Mclntyre and A.C.Mix, The orbital theory of Pleistocene c1imate: supportfrom 21 revised chronology of the marine 8 180 recorel, inMilankovitch amI Climate Part 1, edited by ABerger, lImbrie, 1 Hays, G. Kukla and B. Saltzman, Reide1,Dordrecht, pp 269-305, 1984,

Imbrie, l, ABerger, E.A Boyle, S.c. C1emens, A. Duffy,W.R, Howard, G. Kukla, l Kutzbach, D.G. Martinson,A. McIntyre, AC. Mix, B. Molfino, lJ. Morley, L.c.Peterson, N.G. Pisias, W.L. Prell, M.E. Raymo, N.J.Shackleton and lR. Toggwei1er, On the structllre anelorigin of major glaciation cyc1es 1. Linear responses toMilankovitch forcing, Pa/eoceanography, 7, 701-738,1992.

Imbrie, l, ABerger, E.A. Boy1e, S.c. C1emens, A. Duffy,W.R. Howard, G. Kuk1a, 1 Kutzbach, D.G. Martinson,A. McIntyre, A.C. Mix, B. Molfino, lJ, Morley, L.c.Peterson, N.G. Pisias, W.L. Prell, M.E. Raymo, N.1Shackleton and lR. Toggwei1er, On the strucrure andorigin of major glaciation cycles 2. The 100,000-yearcycle, Paleoceanography, 8,699-735,1993.

Johnson, R.G., Brunhes-Maruyama magnetic reversal dateelat 790,000 yr B.P. by marine-astronomica1 cone1ations,Quat. Res., 17, 135-147, 1982.

Köppen, W. and R. Geiger, Handbuch eier Klimatologie,Verlag von Gebrüder Bornträger, Berlin, 1936.

Maasch, K.A., Statistical detection of the miel-Pleistocenetransition, Clim. Dyn., 2,133-143, 1988.

lntroduction 9

Martinson, D.G., N.G. Pisias, 1.D. Hays, J. Imbrie, T.c. MooreJr. and N..T. Shackleton, Age dating and thc orbital theory ofthe ice ages: development of a high-resolution 0 to 300,000­year chronostratigraphy, Quat. Res., 27, 1-29, 1987.

Milankovitch, M., Die astronomische Theorie der Klima­schwankungen, in Handhuch der Klimatologie, edited byKöppen, \\l and R. Geiger, Verlag von Gebrüder Born­träge~Berlin, 1936.

Mudelsee, M. and M. Schulz, The mid-Pleistocene climatetransition: onset of 100 ka cycle lags ice volume build-upby 280 ka, Eorth. Planet. Sei. Lett., 151, 117-123, 1997.

Mudelsee, M. and K. Stattegger, Exploring the structure ofthe mid-P1eistocene revolution with advanced methods oftime-series analysis, Geo1. Rundschau, 86, 499-5 11, 1997.

Nowaczyk, N" I-Iochauf1ösende Magnetostratigraphie spät­quartärer Sedimente arktischer Meeresgebiete, Bel'. Polor­forschung, 78,187 pp, 1991.

Nowaczyk, N.. T.W. Frederichs, A. Eisenhauer and G. Gard,Stratigraphy of late Quaternary sediments from the Yer­mak Plateau, Arctic Ocean: evidence for four geomag­netic polarity events within the last 170 kyr ofthe Brunhesepoch, Geophys. J /nt, 117,453-471,1994.

Paillard, D., The timing of Pleistocene glaciations from asimple multiple-state c1imate model, Nature, 391, 1998.

Park, 1. and K.A. Maasch, P1io-P1eistocene time evolution ofthe 1OO-kyr cycle in marine paleoclimate records, J Geo­phys. Res, 98, 447-461,1993.

Pisias, N.G. and TC. Moore, The evolution of Pleistocenec1imate: a time series approach, Earth Planet Sci. Lett.,52,450-458, 1981.

Prell, W.L., Oxygen and carbon isotope stratigraphy for theQuaternary of Hole 502B: evidence for two nlOdes ofisotopic variability, Initial Rep. DSDP, 68,455-464, 1982.

Raymo, M.E., The timing of major climate transitions,Paleoceanography, 12,577-585, 1997.

Raymo, M.E., W.F. Ruddiman, 1. Backman, B.M. Clementand D.G. Martinson, Late Pliocene variation in northernhemisphere ice sheets and North Atlantic deep \vatercirculation, Paleoceanography, 4, 413-446, 1989.

Raymo, M.E., D.W. Oppo and W. Curry, The mid-Pleistoceneclimate transition: a deep-sea carbon isotopic perspective,Paleoceanography, 12, 546-559, 1997.

Robinson, S.G., Applications for who1e-core magnetic sus­ceptibility measurements of deep-sea sediments, Proc.GDP, Sci Results, 115,737-771,1990.

Ruddiman, W.F., M.E. Raymo and A. McIntyre, Matuyama41,000-year cyc1es: North Atlantic Ocean and northernhemisphere ice sheets, Earth. Planet Sci. Lett., 80, 117­129,1986.

Ruddiman, W.F., M.E. Raymo, D.G. Martinson, B.M.C1ement and l Backman, P1eistocene evolution: northernhemisphere ice sheets and North At1antic ocean, Pa­leoceanography, 4,353-412, 1989.

Sa1tzman, B. and M.Y. Verbitsky, Multiple instabilities andmodes ofglacia1 rhythmicity in the Plio-Pleistocene: a ge­neral theory oflate Cenozoic c1imatic change, Clim. Dyn.,9,1-15,1993.

Schwarzacher, W., Cyc10stratigraphy and the IVlilankovitchtheory, Elsevier, Amsterdam, 225 pp, 1993a.

Schwarzacher, W., Milankovitch cyc1es in the pre-Pkistocenestratigraphie record: a review, in High resolution stmtigm­phy, edited by E.A. Hailwood and R.B. Kidd. Geul Soc.Spec. PuM, 70. 187-194, 1993b.

Shackleton, N..T., Oxygen isotope analyses and P1eistocenetemperatures re-assessed, Nature, 215, J5-17, 1967.

Shack1eton, NJ., J. Backman, H. Zimmermann, D.V. Kent,M.A. Hall, D.G. Roberts, D. Schnitker, lei. Ba1dauf, A.Desprairies, R. Homrighausen, P. Hudcllestun, 1.B. Keene,A.J. Kaltenback, K.A.O. Krumsiek, A.C. Morton, lW.Murray and 1. Westberg-Smith, Oxygen isotope calibra­tion ofthe onset of ice-rafting and history of glaciation inthe North Atlantic region, Nature, 307, 620-623, 1984.

Shackleton, N.J., A. Berger, W.R. Peltier, An alternative astro­nomical ca libration of the lower P1eistocene tirnesca1ebased on ODP Site 677, Trans. Royal Soc. Edinburgh.·Em·th Sci, 81, 251-261,1990.

Shackleton, N.J., S. Crowhurst, T Hagelberg, N.G. Pisias andD.A. Schneider, A new Late Neogene time scak: applica­tion to Leg 138 sites, Proc. GDP, Sci. Re.l'ult.l', 138,73­101,1995.

Shackleton, N.J. and S. Crowhurst, Sediment f1uxes based onan orbital tuned time scale 5 Ma to 14 Ma, site 926, Proc.GDP, Sci. Re.l'ults, 154,69-82, 1997.

Tarduno, lA., L.A. Mayer, R. Musgrave and ShipboardScientific Party, High resolution, whole-core magneticsusceptibility data from Leg 130, Ontong Java Plateau,Proc. GDP, /nit. Rep., 130, 541-548, 1991.

Thießen, W., Magnetische Eigenschaften von Sedimenten desöstlichen Südatlantiks und ihre paläozeanographische Re­levanz, Ph.D. thesis, Bel'. FB Geow. Univ. Brernen, 170 pp,1993.

Tiedemann, R., M. Sarnthein and N..T. Shack1eton, Astronomietimescale for the Pliocene Atlantic 8 1sO and dust f1uxrecords ofOcean Drilling Program site 659, Paleoceano­graph}',9, 619-638,1994.

von Dobeneck, T, Ana1ytical strategies in environmentalmag­netism - a South Atlantic perspective, Habilitation thesis,FB Geowissenschaften, Univ. Bremen, 1998.

Wilson, D.S., Confirmation of the astronomica1 calibrationofthe magnetic po1arity timesca1e from sea-f100r spreadingrates, Nature, 364, 788-790, 1993.

10

Using Rock Magnetic Proxy Records for Orbital Tuning and ExtendedTime Series Analyses into the Super- and Sub-Milankovitch Bands

1. von Dobeneck* and F. Schmieder

Fachbereich Geowissensch{4ten, Universität Bremen, Pos(lach 33 0440,D-28334 Bremen, Germany

*corresponding author (e-mai1):doheneck(ilJuni-bremen.de

Abstl'act: High-resolution rock magnetic pro:\.)' records of marine sediments, in partieular magneticsusceptibility logs, delineate variations of sediment litholol:,TJ' and mirror climatic and oceanographicchanges of different duration. Most commonly, Milankovitch cyclicity resulting from orbital forcingof carbonate dissolution and terrigeno\ls sedimentation prevails. Extracted by bandpass filtering,these signal components can serve for multiple core correlation, cyclostratigraphic analyses andorbital tuning. Phase relations between astronomical obliquity and precession cycles and theirequivalents in rock magnetic records depend on regional sedimentological settings. Two case studiesare developed to demonstrate specific aims, strategies, strengths and restrictions of rock magnetictime series analyses and their extension into the super- and sub-Milankovitch bands. In the firstexample, twelve Pleistocene susceptibility records spanning the oligotrophie subtropical SouthAtlantic (SUSAS stack) were successfeJIly tuned to obliquity and precession despite their lowaverage sedimentation rates< 1cm/kyr. Multiple bandpass filtering and evolutionary spectral analysisreveal two m~jor base line shifts at arOlUld 0.95 and 0.6 Ma ,md various superstmctures (e.g., amplitudemodulations) of orbital pro:\.; response. Compared to analogous analyses of an adapted astronomicaltarget curve, converse residue patterns in the 41 kyr and 100 kyr bands indicate that Pleistocene iceage cycles were mainly triggered by obliquity before 1.25 Ma andfrom 1.05 to 0.7 Ma, while eccentricitymodulation ofprecession predominated between 1.25 and 1.05 Ma and during the last 0.6 Ma. In thesecond example, eight susceptibility records from the western equatorial Atlantic Ceara Rise (CEARISstack) were tuned to a lagged precession index signal. Subsequently, a high resolution core correlationscheme was established on basis of their coherent high-frequency « 15 kyr) signal patterns. Basichannonics and intermodulation frequencies of obliquity, and, predominately, precession resultingfrom nonlinear pro:\."y response were detected by sub-Milankovitch spectral and bispectral analyses.Twin susceptibility peaks corresponding to a tropical double precession cycle appear even in theunfiltered records. Millennial signal variations « 7 kyr) seem to coincide with Bond- and Dansgaard­Oeschger cycles.

Introduction

Many marine rock magnetic records exhibit Milan­

kovitch cyclicities - sometimes in striking agreementwith 8180, 8 13 C or %CaC03 profiles. This phe­

nomenon is generally explained by climatic impact

on the fluxes ofmagnetically enriched terrigenous

and 'non-magnetic' biogenic sedimentm'y compo­

nents. They are increasingly employed as a data

base for high-resolution care correlation, orbital age

modelling and paleoceanographic time series analy­

sis ofmarine sediments. Unlike ice-shield mediated

climate proxies, magnetic records sustain their

periodicity deep into pre-Quaternary times (Park

et a1. 1993; Hilgen et a1. 1995; Shackleton and

Crowhurst 1997). Paleoclimatic signatures ofmag­

netic susceptibility and various remanence parame­

ters have been described and stratigraphically ex­

ploited in paleoceanographic studies from all ma­jor oceans (e.g., Radhakrishnamurty et a1. 1968;Kent 1982; Robinson 1986; Bloemendal et a1. 1988;deMenocal et a1. 1991; Bickelt et a1. 1997).

From FISCHER G, WEFER G (cds), 1999, Use 01 Prox/es /n Paleoceanography. Ex:amples ji'om the SOllth Atlant/c. Springer- VerlagBerlin Heidelberg, pp 601-633.

12 Chapter2

The article on marine 'Environmental Magnet­ism' by Fl-ederichs et al. (this volume) outlines thephysical and sedimentological principles of rockmagnetic climate proxies. Lithological changes insediment series can be traced by a range of speci­fic rock magnetic parameters which indicate con­centrations of para-, ferri- and antiferromagneticminerals, discriminate among remanence-carryingminerals (various iron oxides, oxyhydroxides andsulfides), estimate average ferrimagnetic grainsizes, or quantiry additional magnetic propeliies suchas coercivity or anisotropy.

There is no universal rationale for the linkageofclimate and sedimentary magnetic mineral inven­tories, but detailed rock magnetic studies usua1lyprovide the means to identify prevailing orbital forc­ing mechanisms within a regional context. Pro­nounced climate dependence has been observed foreolian dust load, ice-rafted debris, ocean currenttransport and sea-level related shelf erosion - a1limportant sources or pathways ofmagnetic mineraldeposition. The likewise cyc1ic accumulation ofcalcareous and siliceous microfossils plays a com­plementary and often dominant role in generatingmagnetic signals by modulating the concentrationofthe magnetic mineral fraction. Magnetic signalpatterns can be obscured by too stable or too com­plex sedimentation conditions, discontinuous depo­sition (bottom current erosion, debris ±lows), inter­calated high- or low-magnetic (tephra, fossil) lay­ers, and, most severely, by reductive diagenesis ofprimary felTimagnetic minerals in sub- or anoxicmarine environments (Tarduno 1994; Tarduno andWilkison 1996; Frederichs et a1., this volume).

Isothermal magnetic parameters are determinedby measuring the magnetic moment of artificiallymagnetized, but otherwise untreated bulk sedimentsampIes. They represent averaged volume proper­ties, are precise and weIl reproducible. Because ofthe ubiquity ofmagnetic iron minerals in nature andthe availability ofhighly sensitive magnetometers,virtually every lithology can be investigated andcharacterized by magnetic parameter sets. Theattainable speed and spatial resolution ofwhole-coremeasurements convinced a growing munber ofmarine geoscientists to integrate rock magneticmethods, especially magnetic susceptibility 10gging,into their methodical repertory.

This paper intends to explain and illustrate stand­ard and advanced concepts of Quaternmoy cyclo­stratigraphy with emphasis on marine rock mag­netic records. The three main sections are con­cerned with- stepwise refinement ofmultiple care correlationschemes, phase analysis, and orbital tuning,- evolutionary spectral analysis ofMilankovitch andsuper-Milankovitch signal variations covering long­periodic (> 100 kyr) amplitude modulations andtransitions ofbaslc climate cycles- higher-order spectral analysis ofsub-Milankovitch.sIgnal components relclted to climate variations ofshort duration « 18 kyr).

As necessary restriction and tribute to the mostpopular rock magnetic parameter in paleocea­nography, our case studies will exc1usively focus onmagnetic susceptibility records.

Experimental Methods

Traditional single sampIe measurements in rockmagnetism are increasingly replaced by faster non­destructive techniques. High-resolution whole-core01' half-core logging of magnetic susceptibility K

(Robinson 1990) is now a routine procedure inmarine sediment studies. The lateral sensor char­acteristics ofthe widely used 'Bartington' suscepto­meter have half-widths of 50 to 25 mm for loopsensors ofvarious diameters and of 12 to 4 mm fardifferent spot sensors. These values also representconservative minimum estimates for the attainablespatial resolution.

Modern pass-through SQUID magnetometersequipped with additionalmagnetization coils meas­ure isothermal and anhysteretic remanent magneti­zation at resolutions ranging from 80 mm for wholecore logging down to 30 mm for 'U-channel' (axialsubcore) measurements (Weeks et a1. 1993). A10 mm resolution far surface remanence measure­ments is obtained by a newly developed high-TcSQUID 'Rock Magnetic Micro-Scanner' (vonDobeneck et a1. 1996).

Spatial resolutions down to 1 mm at the ex­pense of a laborious single sampIe preparation areachieved by measuring magnetic hysteresis loops.The concept of the 'Alternating Gradient ForceMagnetometer' (AGFM) allows to detennine high-

Using rock magnetic proxy records for orbital tuning and extended time series analyses 13

quality hysteresis data from miniature 10-20 mgsampIes (Flanders 1988; von Dobeneck 1996).

Typical data sampling densities for high-resolu­tion rock magnetic core-logging are 0.5-2 measure­ments/cm, which is more than the effective spatialresolution ofmost sensors. This over-sampling re­sults in smooth, well-defined and complete records- an important prerequisite for the performance ofsignal analysis techniques. In this paper we presentsusceptibility records acquired with a 'BartingtonF type' spot sensor at a 1 cm spacing.

Orbital Tuning Strategies

Milankovitch' s (1941) visionary idea ofsynchroniz­ing geological records to orbital variations hasevolved into a widely accepted chronostratigraphicmethod (Imbrie and Imbrie 1980; Imbrie et al. 1984;Martinson et al. 1987). Applied to records of iso­topic 01' lithologic characteristics, 'orbital tuning' canyield unmatched precision and consistency in dat­ing Quaternary sediment series. Ideally it takes justthree steps to align a proxy record to its orbitalpacemaker:- establish a rough initial chronology and detectspectral characteristics ofMilankovitch cyclicity,- choose an astronomie target curve and shift it bypostulated 01' empirically determined phase lags,- filter the proxy signal in the appropriate frequencyband and match it with its model record.

In practice this approach bears numerous pit­falls. Rock magnetic records, like most proxy sig­nals, depend more on regional sedimentation condi­tions than on global equilibria. They are not, in prin­ciple, safe bets for orbital tuning pUllJoses and theirpatterns can be anything from a 'SPECMAP (Im­brie et al. 1984) double' to 'random wiggles'. Cy­clostratigraphic age models critically rely on theinitial and thus very consequential age model. Sim­ple peak-to-peak correlation of proxy and targetrecords is highly susceptible to elToneous matches(Shackleton et al. 1995), which have their origin inordinary stratigraphic problems such as undetectedhiati or turbidites, condensed or extended sections,bioturbation, lithologic and diagenetic effects. Al­gorithms such as signal slotting (Thompson andClarke 1993) 01' complex demodulation (Pisias etal. 1990) have been used to find the statistically

most probable proxy-target-correlation. Yet a COl11­

bination ofmagneto- and biostratigraphic, eventu­ally radiometric ages should be thc most reliablebasis to start from. Specific regional settings andscientific objectives may require adaptation of or­bital tuning schemes into flexible, case-wise adapt­ed strategies.

Two contrasting examples illustrate the scopeof possiblc approaches at different interpretationlevels. They are based on two series ofFS METEOR gravity COlTS collected within a long­term Quaternary South Atlantic research project(SFB 261) at the University of Bremen. The cor­ing sitcs are located in the subtropical and in thewestern tropical South Atlantic. The two study ar­eas differ greatly in size as weIl as in prevailingsedimentation conditions and rates. The first exam­pIe outlines the possibilities and limits ofsignal cor­relation and orbital tuning in low accumulation en­vironments. Covering nearly the full Quaternary,these susceptibility records are rewarding targetsfor evolutionary spectral analysis. The particularinterest ofthe second example lies in a much highersedimentation and signal detail opening the possi­bility ofanalyzing periodic sub-Milankovitch signalcomponents. We will primmily focus at the methocli­cal aspects ofthese magnetic time series analyses.

Subtropicctl South Atlantic

The Subtropical ~outh Atlantic Susceptibility(SUSAS) Stack stretches over the total width ofthe pelagic South Atlantic (Fig. 1) at around25-35°S, connects its fourmajor basins (Argentine,Brazil, Angola and Cape Basin) and crosses threesubmarine ridges (Rio Grande Rise, Mid Atlanticand Walvis Ridge). Via two deep water passagesin the western South Atlantic, Vema and HunterChannel, carbonate undersaturated Antarctic Bot­tom Water flows northward into the Brazil Basin,from where only a minor part reaches the AngolaBasin through the Romanche Fracture Zone. 1'0­day, the slightly carbonate supersaturated NorthAtlantic Deep Water (NADW) flows southward atdepths below 2000 m and controls the bottom wa­tel' conditions in the Angola Basin. Therefore, thecarbonate lysocline is much deeper in the AngolaBasin (4700 to 4900 m) than in the Argentine, Bra-

14 Chapter 2

20 0 E0°

35°8

f-l---L.L--'--'--'--L...L-L-'-L-l-.J.-'--'-'---'---'---'-'-''-'-L..L-L..L-~~~.L..L--'--'--'--'--'--L...L-L-'-L~~~I~I~I~I~I~I~I~I~I~1L.L1 I j I I j I L ...L...i.

[m] M '<tMMd> ..o~ ~

1,000 ~ gg~~ ~~-r-

o-1,000

-2,000

-3,000

-4,000

-5,000

-6,000

Fig. 1. GeoB core locations defining the SUSAS transect. The depth profile follows the white dashed line in themap and is plotted against longitude to depict the cores' affinities to the four major pelagic basins and their deepwater bodies. It is not suitab1e to identify the positions and depths of deep water passages.

zil, and Cape Basins (4000 m). This west-eastasymmetry of carbonate preservation diminishesduring glaeial times, when NADW produetion isredueed and the lysoeline rises to about 3800 m inall four basins (Biekert and Wefer 1996).

Due to oligotrophie eonditions along the profilethe average sedimentation rates over the last1.5 Ma range between 0.4 and 1.2 em/kyr (Fig. 5).Suseeptibility eare means range from 18.5 to470x 10-(, SI. They inerease with depth and showa drastie west to east decline originating from amueh higher terrigenous influx to the western partsofthe subtropical South Atlantie (Lisitzin 1996). AllSUSAS logs exhibit distinet periodie oseillationsprimarily in the 100 kyr and 41 kyr band refleeting

carbonate aeeumulation variations, with eoherent10ng-tern1 signal shifts superposed.

Oxygen isotope reeords exist für four SUSASeores, GeoB 1034-3, 1035-4, 1211-3 (Biekeri 1992;Biekert and Wefer 1996) and 1309-2 (W. I-laIe,pers. eomm.). Their stratigraphie interpretability islimited by very low sedimentation rates, partieularlyin the deeper, obliquity-dominated (41 kyr eyeles;e.g., Ruddiman et al. 1986) seetions. Magnetostra­tigraphie dating at 5-10 em spaeing (Fig. 2) pro­vided three 01' four reliable tie points in ten oftwelveeores. By attributing these ages to the respeetivemagnetie suseeptibility logs, some speetraleharaeteristies of Milankoviteh eyelieity beeomeapparent (Figs. 3a, 4a) suggesting elimatie foreing.

Using rock magnctic proxy records for orbital tuning and extended time series analyses 15

ChRM Inclination n2821·1 2820-2 1311-1 1309·2 3814-6 3813-3 3812-1 1729-3 1034-3 1211-3

o 00( 0 m

{

!L.

o 0~ 0 m

ro .C'~ 'e::

cuCf) (5Q) 0-OJ«mCf)'-Q)>Q)

0:::

• I

0.78

1.95

0.991.071.19 CM

cuEcu>.::J

cu:2:

1.77

Fig. 2. Combined magnetostratigraphies (ChRM: Characteristic Remanent Magnetization) ofSUSAS cores com­prising the Brunhes/Matuyama boundary and the Jaramillo event (Jar). The Cobb Mountain (CM) event cannot beclearly identified at all sites. The Olduvai (Old) event is not reached in any of the cores. Reversal ages are givenaccording to Shackleton et a1. (1990). Data ofcores GeoB 1034-3 ane! 1211-3 are from Thießen (1993).

Lomb-Scargle Fourier transfom1 (LSFT; Lomb1976; Scargle 1982, 1989) embedded in the SPEC­TRUM program (Schulz and Stattegger 1997) wasused throughout all spectral analyses unless indi­eated, as it can be direetly applied to unevenlyspaeed times series. The widely applied Blaekman­Tukey method of spectral analysis (Blaekman andTukey 1958) employs Fourier Transform of thetruncated and tapered autoeovariance function andrequires evenly spaced time series. The necessaryinterpolation procedure not only fails at age gaps,but inevitably underestimates high-frequeney eom­ponents and thereby 'reddens' the spectrum (Horo­witz 1974). Moreover, interp01ated data are statis­tically dependent (Mudelsee and Stattegger 1994)eomplicating the assessment ofthe signifieance ofcomputed speetra.

Another important concem is spectralleakage,a mathematieal consequence ofthe finite length oftime series. Due to this effect dominant orbitalcycles generate spectral ripp1es at higher frequen­eies, which may form misleading interference pat­terns. Tapering records with suitable window func-

tions such as a Welch or Hanning window redueesthis leakage effect at the expense of attainableresolution (HaITis 1978). Individual compromisesare found by eomparing results from different ta­pers. A seeond 'spurious peak problem' arises fromrandom fluctuations related to stochastic proe­esses. A relatively simple eounterstrategy is'Welch's overlapped segment average' (WaSA)proeedure (Welch 1967). By dividing a proxy reeordinto overlapping segments and averaging their rawspectra, 'noise' peaks are gradually suppressed, butagain resolution decreases.

Here a combination strategy of tapering andWaSA is applied. Following signal analytica1 no­menelature, speetra by Fourier transform of rawdata are called 'periodograms', tapering yields'modified periodograms' and segment averaging'averaged periodograms' . Solely 'autospeeh'a' givea complete spech'al representation oftime-depend­ent processes. They cannot be determined on ba­sis offinite diserete time series, but only estimatedwithin increasing1y nalTOW eonfidenee limits by theabove defined spectra.

16 Chapter 2

All initial spectra display significant maxima ataround 40 kyr as core GeoB 3814-6 (Fig. 4a), butshow little or no indications 01' a precessional sig­nal component. As a spectrum represents an av­erage over the integral record, it would be overlyoptimistic to inter that a11 obliquity-related oscilla­tions can be extracted at this stage by filtering therecord in the respective frequency band. Instead,all susceptibility records were mutually correlatedto identify missing 01' expanded intervals. This com­parison enables to estimate the extent o1'some shortgaps and two laminated diatom ooze layers (Fig. 5),which had obviously been deposited in very shorttime. Core to COlT correlation revealed that simply

removing these sections 1'rom the sequence resultsin complete records.

A pattern matching to standard 8 180 records.e.g. from equatorial Pacific ODr Site 677 (Shack­leton et al. 1990) was feasible over most signalsections, but remained ambiguous between oxygenisotope stages 16 and 13 between about 0.65 and0.5 Ma. Contll1uously high simdarities, includinglong-term f1uctuations 01' about 500 kyr duration,exist between all SUSAS susceptibility signals andthe benthic 8 13C record at Pacific ODP Site 806(Bickert et al. 1998), based on an orbita]]y tuned8 180 stratigraphy (Berger et al. 1994). The com­pIete and undisturbed susceptibility record of core

Polarity

ISuscepti-bility K i

~~~~.__. .__ "" ~--.J

1.5141.31.21.10.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9Age [Mal

0.1

K

filtered15-47 kYrh,-""T"T'"",,,,--"T'""r"1---.---.-.+--r-,,,-.---r-r-ro"""TT"-.--Tlrr."'--rrT-rT"",,,-r-r-,-,'-,Ho-rr-rrT"""ri"""T""TO""""rT"'-rrT--'r--rr-r'r"""ri""T'T-rl

o

TargetCurve

Fig. 3. Four step refinement ofthe age model for core GeoB 3814-6. (a) Susceptibility record dated by magneto­stratigraphy (3 tie points). (b) Signal pattern conelation to the tuned 0 \lC record ofODP Site 806 (44 tic points). Incritical intervals other SUSAS records were inspccted for supplementary inf0l111ation (c) Susceptibility filtered (35­47 kyr) in the obliquity range (dashed line) and hmed (solid line) to an astronomical obliquity signal shifted by-4.5 kyr (73 tie points). (d) Susceptibility filtered (15-47 kyr) and tuned to an astronomical (obliquity and preces­sion) target curve (95 tie points).

Using rock magnetic proxy records for orbital tuning and extended time series analyses 17

required. 8 180 phase lags have been determmcd onbasis ofradiometric ages (Hays et a!. 1976: 9 kyrtor obliquity, 3 kyr for precession) and assumptionson the coupling of insolation and ice mass (Imbrieet a!. 1984: 7.9 (7.4-8.2) kyr for obliquity, 5.0 (4.8­5.1) kyr and 4.2 (4.1-4.3) kyr for precession). Taestimate orbital phase lags for regional proxy pa­rameters such as susceptibility it is therefore suf­ficient to know thc phase lags relative to a Öl80curve [rom the same COlT.

Here we use a revised 8 180 time scale of coreGeoB 1211-3 (Bickert 1992) based on correlationto thc SPECMAP stack (Imbrie ct a!. 1984). Crossspectral analysis of8 180 and K yields a coherenceofO.99 and 0.96 for the 100 kyr and 41 kyr cyclesand reasonable phase angles of -41 0 and -30 0

. Thisimplies that susceptibility leads8 180 by 3.4 kyr inthe obliquity band and therefore lags the obliquityforcing function by 7.9 kyr-3.4 kyr=4.5 kyr. Thetemporal resolution ofthe 8 180 reeord was tao Jowto determine the precessional phase lag with anacceptable error margin. 'rhe subsequent orbitaltuning process was performed in two stages:

At first the obliquity-related signal componentwas extracted applying a 1Sl order butterworthbandpass (35-47 kyr) filter in forward and reversedirection. This 'zero-phase-filtering' avoids phasedistortion of the resulting sequence (Oppenheimand Schafer 1989) and was thereforc used in allinstances. It effectively doubles the filter order. Alow filter order results in a short filter length andproduces little reverberation. Its soft transition be­tween stopband and passband limits the risk ofsig­nalloss in record sections, which are unduly com­pressed 01' stretched (and hence shifted in fre­quency) by imperfect initial age models.

Each maximum and minimum in the filtered sig­nal was assigned the age (+4.5 kyr) of its postu­lated equivalent in the ash'onomical record (Bergerand Louh'e 1991), stmiing from the most uncriticalsections near magnetostratigraphic 01' other reliabletie points. As illustrated in Fig. 3c, relative changesin the conelation age model are small, typically lessthan a half-cycle of obliquity. In the frequencydomain (Fig. 4c) one notes just a small rise ofpeakamplitudes.

The second tuning step proceeds from the ob­servation that faintly visible precessional peaks in

40

40

50

~~~""iIl'-rl""""'r"M".,..,..,.""':'~30

60

>- >- >- >- >-CL CL >- >- >- CL CL >-.q r-- CL CL CL CL r-- .q CL

0 N <.D .q ::;; m 0'0 N m.q ~ m CD N N NU U u :D :D :D <J.)

~<J.)

u u u 0:: 0::w w w 0 0 0 Cl.

60

50

40

30

@ 60

<J.)0)

«-'(':'t?.8

(i) <J.)c

(ü 0)cu

Ü ::2:(J)'-(1J <J.)(lJ 0)c «2 c

0

"-- ~-- ~

~ CiU

N[;?b

E '.0

2 <J.)

TI 0)

«(lJ

0)0.. c

(J) 'cQj

::JI-

3: :20 ::J

0.. .sr:D

"0 0(lJ

cuE <J.)

0)

~ «0)

W c'c::Jl-

m0)

CoI-

0

GeoB 3814-6 (Fig. 3b) cOITelates particu1arly welland he1ped to substantiate the other eleven matches(Fig. 5). The improved spectral definition ofMilankovitch frequencies owing to this correlationprocedure is obvious (Fig. 4b).

The most widely used orbital tuning procedureassumes a phase lock between climate and met­ronome record (Martinson et a!. 1987). In order toassess the absolute lag of any proxy record withrespect to astronomical obliquity and precession, anindependent and precise absolute chronology IS

~~~~"!""NL,..,..~1-rT-1"rT--,Jl-300.01 0.02 0.03 004 0.05 0.06

Frequency [1/kyr]

Fig. 4a-d. Estimated spech'a (Hanning taper, WOSA: 2segments) of the GeoB 3814-6 susceptibility signal onlinear (solid b1ack) and logarithmic (line) scales at eachage model stage ofFig.3a-d. The cross is re1ated to thelogarithmic scale and delineates 6 dB bandwidth and90 % confidence interval of all estimated spech'a. Thedotted lines depict orbital periods ofeccentricity (Ecc),obliquity (ObI) and precession (Pre).

18

:MM,t4

Chapter 2

CIM MatuyamaI,! r! I!! I! I, I t,! I!!! I,! ! I I! i!!! I! I!! I! I!! I, I!! I !!, I,!!! I!!! I L..l._....l....-~ __.LlJ......L-LLJ

0.2 0.3 0.4 0.5 0.6 0.7 0.8Age [Ma]

0.9 1.0 1.1 1.2 13 1.4

1(+S'l

K-IK-SJ

1.5 16

Fig. 5. Individually tuned SUSAS recards and resulting stack (arithmetic mean with standard deviation band). Tocompensate for different mean amplitudes, each recard was standardized by subtracting its mean and dividing bythe standard deviation s. Dots mark identified paleomagnetic reversals (Jar: Jaramillo; CM: Cobb Mountain). Theyare generally located deeper in the cares than the corresponding reference ages of Shackleton et a1. (1990), de­picted by dotted lines. The resulting average lock-in depth is l2±9 C111. There is an obvious east-west trend (topto bottom) of increasing signal amplitude, but less detailed signal features. The gray vertical bar marks a zone ofpeculiar lithologies in six cores related to a conspicuous increase of averaged sedimentation rate (bottom, indi­vidual sedimentation rate records were previously n0l111alized to the overall 0.5-0 Ma average ofO.75 cm/kyr).

Using rock magnetic proxy records for orbital tuning and extended time selies analyses 19

the logarithmic spectra gradually sharpened at bothrefinements ofthe age mode!. To benefit from thisprecessional signal component, a target curve ofn0l111alized obliquity and precession index (mixingratio of2:3 chosen by visual evaluation) and 4.5 kyrlag was calculated, comprising not just more, butalso more prominent signal features than either sin­gle orbital parameter. The obliquity-tuned primarysusceptibility record was filtered using a widebandpass (15-47 kyr) to include obliquity and pre­cession cycles. As many extrema as possible werematched (Fig. 3d), admittedly at the resolution limitofthese records. The resulting final age models areagainjust slight modifications ofthe previous mod­els and possibly not even more precise 111 absoluteages as the precession time lag is undetermined.Nevertheless, this higher-frequency tuning leads toa better mutual signal correlation of the SUSASrecords and therefore to an improved conservationof signal details in the SUSAS stack (Fig. 5). Be­cause of the very different amplitude ranges, thestack was determined as an arithmetic mean ofnormalized (subtraction ofcore mean and divisionby standard deviation), interpolated (Lit=2 kyr)records.

Bioturbation should be responsible for the com­paratively low spectral power in the precessionband. Its effect can be assessed from the fact thatseetions ofhigher sedimentation rate (e.g. interval0.6-0.2 Ma) clearly coincide better with precessionpatterns. A precessional cycle at a typical sedimen­tation rate ofr=0.5 cm/kyr corresponds to roughly10 cm core depth, twice the minimum estimate forthe pelagic mixed layer thickness L (Peng et a!.1977). Mathematical models of vertical mixing(Berger and Heath 1968; Dalfes et a!. 1984) per­mit to quantify the frequency selective amplitudedamping. Under the stated sedimentation conditionsthe 'gain factor' G at an averaged precession fre­quency of[= 1/21 kyc l is

Less intense damping of the obliquity relatedpaleosignal (G=0.30) and nearly negligible damp­ing of 100 kyr (G=O.72) and 400 kyr (G=0.98)cycles account for a relative loss of spectral inten-

sity in the precession band. Bioturbation also in­fluences phase relations to some extent (Dalfes eta!. 1984) and can bias orbital tuning ages.

Fig. 5 summarizes the combined resll1ts 01'magnetic age modelling for the complete transecLA 1'ew question marks remain in core sections,where individual signal features seem to be missing01' incomplete (e.g., GeoB 1311-1 at 0.42Ma,GeoB 2820-2 at 0.8 Ma). The SUSAS stack is freeof such local effects and therefore representativefor the whole Quaternary oligotrophie deep SouthAtlantic Ocean.

Cearä Rise

The Ceara Rise .§.usceptiblity (CEAR1S) Stackcomprises two short tTansects on the northern flankof the Ceara Rise, a submarine elevation 700 kmnorth-east offthe Amazon Delta (Fig. 6). A steady,but highly variable deposition ofterrigenous parti­eIes strongly modulates the susceptibility signa!.Several pathways are involved. The sea-Ievel fallin glacial periods causes erosion ofthe inner con­tinental shelf, channeling ofthe Amazon dischargedirectly into the Amazon Canyon (Milliman et a!.1975; Damuth 1977) and intensification ofmass­flow events on the slopes of the Amazon Fan(Maslin and Mikkelsen 1997). These mechanismsenhance the suspended particle load at differentwater depths reaching the Ceara Rise either bysurface transport via the retrofleetion ofthe NorthBrazil Current into the North Equatorial Counter­current (Johns et a!. 1990) or by intensified nephe­loid transport via the southeastward directedNADW flow (Kumar and Embley 1977; Francoisand Bacon 1991). Increased eolian input from Af­rica in glacial periods (Sarnthein et a!. 1981;Matthewson et a!. 1995) and possibly enhancedsediment yield of the Amazon in glacials are fur­ther factors which contribute to the susceptibilityvariations.

This rather complex scenmio ofcompeting sedi­ment fluxes might be suspected ofproducing com­plex composite susceptibility signals. Yet allCEARIS records are perfectly suitable for orbitaltuning and exhibit very consistent patterns not onlyin the Milankovitch band, but also down to periodsof2-5 kyr.

20

52°W 50 0 W 48°W

Chapter 2

46°W 44°W 42°W 40 0 W

6°N

5°N

4°N

3°N

2°N

1°N

0°

1°S

2°S

Fig. 6. Simplifieel hyelrography ofthe Ceani Rise region anel GeoB core locations inclueleel in the CEARIS stack.The retrofleetion ofthe NOlih Brazil Current (NBC) into the NOlih-Equatorial Countercurrent (NECC) is activateelfrom June to January (Muller-Karger et al. 1988). The North Atlantic Deep Water (NADW) flows e1irectly over theCeara Rise, while the Antarctic Bottom Water (AABW) fully circumvents this bathymetric high.

Although no CEARIS eore was believed toreach the Matuyama chron, a11 were submitted tosystematic paleomagnetic analysis (Bleil andvon Dobeneck, this volume). Several reversedintervals, some with multiple rebounds, were de­teeted in eaeh polarity record. Dating these fea­tures, whichjust vaguely coineide with previouslyreported Brunhes polarity events (e.g., Nowaczyket a1. 1994) seems diffieult.

Oxygen isotope stratigraphies are available fortwo of eight CEARIS eores (Mulitza 1994; Rühle­mann et a1. 1996). These ehronologies (Fig. 7)were based on a SPECMAP (Imbrie et a1. 1984)eorrelation and represent eonvineing initial agemodels. They were adopted for the two respeetivesuseeptibility reeords (Fig. 7) and transferred to theother six cores (Fig 10a) by graphie eorrelation.

The pronouneed similarity of8 180 and suseep­tibility records and their obvious agreement withSPECMAP testify to the elimatie dependenee ofboth parameters during late Quaternary glaeial/interglaeial eycles. It is interesting to note that thesynehrony of oxygen isotope and susceptibilityvariations is disturbed by varying peak lags andincoherent sections (Fig. 7). These effects are toolarge to be simply explained by the higher spatialresolution of the magnetie (1 cm) versus the iso­topie (5 cm) record. Potential eauses are- proxy- and eycle-speeific gain and lag va1ues re­sulting from different signal forn1ation meehanisms(Martinson et a1. 1987),- hatmonie and intermodulation frequencies gen­erated by (eoupled) nonlinear response to orbitalforeing (e.g. saw-toothed eurves),

Using rock magnetic proxy records for orbital tuning and extended time series analyses 21

GeoB 1515-1

8180 [%0] K [10-6 SI]

-2 -1.5 -1 -0.5 0 50 100150200250u,,-c'J..u.~.LUJ..u.JW-.LLli-- 0

GeoB 1523-1

8180 [%0] K[10-6 SI]

-2 -15 -1 -0.5 0 50 100150200250

SPECMAP

6180 [%0]

-3 -2 -1 0 1 2

o 1.1

2.23 1

50 334.2

100 53254

5.56.2

150 - 643

65 0m 66 (l).Y 7.1 -0'Q)' 200 72 s:Ol 7.3 4~«

74 2,7.5

2505

300

6

350111 7

400

Fig. 7. Initial age model for CEARIS cores. Globigerinoides sacculij'er 8180 records 01' cores GeoB 1523-1 (Mulitza1994) and GeoB 1515-1 (Rühlemann et al. 1996) were graphically correlated to SPECMAP (Imbrie et al. 1984) sub­stages (dotted lines). Their ages provide 27 (19) tie points for the respective susceptibility records. Note inconsist­ent lead (stages 4.2,5.4,6.2,8.4) and lag (stage 5.1) OfK to 8180.

- the influence of extraneous (non-Milankovitch)forcing sif,'11als.

All possibilities are realistic and have impOliantimplications for the tuning sh-ategy, pmiicularly forchoosing the most suitable target record.

The spectral representations of Fig. 8 revealsome information in this respect. Botho l80-basedspech-a exhibit well-defined and similar, albeit notidentical orbital characteristics. The susceptibilityspectrum has considerably more precessionalpower and some additional peaks. Reasons for thediffering distribution ofspech-al power will be givenlater. In the context of orbital tuning, it is sufficientto state that magnetic susceptibility chiefly followsprecession and retains a rather stable phase lag(Fig. 9a, c). A shifted astronomical precessionrecord is therefore the most evident tuning target.

SPECMAP determines the phase lag of ourinitial age model. According to cross-spectral analy­ses of both proxy parameters and the precessionindex (PI), the lags at the three main precessionfrequencies amount to 3-8 kyr foro '80 and 0-5 kyrfor K (Fig. 9c). Since there are more than threeorbital precession frequencies (the peliods 23.7 kyr,22.4 kyr and 19 kyr are only the three 1m-gest am­plitude terms in the h'igonometric expansion ofas­tronomical precession; Berger and Loutre 1991),the applicable lag should be represented by aweighted average combining these phase angles,period lengths and spectral power. The time-domainequivalent of cross-spectral density, the cross­covariance function c ,supplies this weighted av-

xy

erage lag. As both compared records are band-lim-ited (precession by nature, susceptibility by band-

22 Chapter 2

Fig. 8. Spectra (0-377 ka, Welch taper) ofSPECMAP andGeoB 1523-1 8 180 and K records according to the corre­lation age model ofFig. 7 on linear (solid black) and loga­rithmic (line) scales. The cross is related to the logarith­mic scale and delineates 6 dB bandwidth and 90 % con­fidence interval ofal! estimated spectra. The dotted linesmark basic orbital periods. Note the various influenceof precession forcing in each spectrum and additionalpeaks in the K record. The lower spectral resolution incomparison to Fig. 4 is a mathematical consequence ofthe much shorter time series.

pass filtering), the horizontal shift of the centralmaximum of the cross-covariance function(Fig. 9b, d) corresponds to the requested lag(Fig. 9c). According to this analysis, the suscepti­bility record ofcore GeoB 1523-llags precessionalforcing by 1.77 kyr and leads 8 180 by 3.90 kyrwhich adds up to a plausible total delay of 5.67 kyrbetween PI and 8 180.

Starting out from the 8 180 based correlationage (Fig. 10a), precession tuning is a straightfor­ward, but typica11y iterative procedure (Fig. lOb).Each step starts with filtering the time series in an

extended precession band (15-28 kyr to include er­roneously compressed or stretched signal sections),proceeds with identification and correlation ofclearly developed maxima and minima to the shiftedprecession signal and ends with releasing the pre­ceding and adopting the improved age model. Theseage modifications averaged 1.3-2.2 kyr in the firstand 0.4-0.9 kyr in the sec(md tuning step.

While the initial pattern cOlTelation scheme wasessentia11y based on matching individual peaks, theorbital tuning procedure relies on thc maxima andminima of a filtered signal, to which, due to thenecessary filter length, entire record seetions nu­merically contribute. Consequential1y, a tuncd agemodel relies more on the continuous evolution thanon singularities of a paleoccanographic record. Onthe other hand, nalTOW peaks and steep slopes havea strong impact on filter-extTacted signals and hencealso on their dating. The origin ofthese signal com·poncnts is apriori uncertain and provides no ties toexternal time sca1es unless convincing links toknown high-frequency climate variations such asHeinrich events (Grousset et a1. 1993; Robinson eta1. 1995; Chi and Mienert 1996), Bond (Bond et a1.1993) or Daansgard-Oeschger cycles (Dansgaardet a1. 1993; Moros et a1. 1997) can be established.

Spike and slope features, which are we11 repro­ducible within a core co11ection, can neverthelesshelp to improve the internal coherence ofcombinedage models. Exh-acted and emphasized by highpassfilteling in the sub-Milankovitch band (here <15 kyr)the high-frequency patterns of the CEARIS corecollection was graphica11y conelated at an averagespacing on kyr between tie points (Fig. 1Oc). Eachtie point was set to a common mean age (CMA)assuming that relative delay times in the study areaare negligible. The graphic cone1ation ofhigh-fre­quency signal components results in mean ageshifts ofO.8-1.5 kyr. From a statistical viewpoint,this procedure corrects for orbital tuning errorsrelated to core-specific features and thereforeimproves the precision ofeach age model. Becausea11 conelated cores were previous1y dated by thesame proxy and principles, the cumulative age shiftequals zero for the total core co11ection and, forstatistical reasons, also for each core.

At this refinement stage, it is justified to demand,how precise this tuned age model might be. A major

20

10

o

40

30

(j)

0::

>->--'" -'"t'-'<;f"

MNN N(j) (j)

0::0::

""""''''l'''!''"'I''"I'''I''"'1'-:r'-H.J,-,-"'I''"I''''l''\L....,...,~'-rl- -1 0 S"20 '§.,

s:10 :3

o'o g>

~CD

.....,..~..,..,."!--10CL

50 ~

~~""!"l-20

0.01 0.02 0.03 0.04 0.05 0.06Frequency [1/kyr]

);, );,);, );, );,-'" -'"

'<;f" t'- -'" -'" -'"0 N (0 '<;f" ~

'<;f" ~ rn LI) '<;f"0 0 0 Li .D0 0 0W W W 0 0

o

Using rock magnetic proxy records for orbital tuning and extended time series analyses 23

€I..••·.'~." ':':" <':' :.'X.·':':\.. " ...>:..... :,.'...:. ' .... ~.. :.. ~>. ::•. ,: •• '.:~:" ...:,,' •• '. :C>:.:':. '.. J.""".':':'. '; ',.,/1\.,. ').'~~.'.~·:>:-I'\§J' •..••• ". • • •. . ••••••. ••••. k- \J

/ ..'+"" -

;L ~I\/V~-'I

~1:iili!:jl·jl:{-T~ij":lli'il

o 20 40 60 80 100 120 140 160 180 200 220 240 260 280 300 320 340 360 380Age [ka]

·50 o 50 100 150 200 250 300 350 400

~ ~ ~ \~ Signal Lag [kyr]

~ ~ ~ ------------C'J 0J ~ __

g:. g: CL 0180 ---,. ~;",.<,":rl\ • I 5.67kyr .l.LL...L.L....t:'...L.l.•....LL......L~

• '·I·g·,;rI·} \. TI '\. JI·I·l

• 1

-300 -250 -200 -150 -100-350

.·h

-400

o

Lag[kyr]

10

-10 .L...... _10bJ .... .. KI_1:~~:~r77kY'

3.9 kyr

1,77 kyr

5.67 kyr

0,03 0.035 0.04 0.045 0,05 0.055 0,06Frequency [1/kyr]

-10 -8 -6 -4 -2 0 2 4 6 8 10Signal Lag [kyr]

Fig. 9. Quantification ofthe 0'80 and K signallags in relation to precession forcing (example: GeoB 1523-1). (a)Based on the SPECMAP correlation age model (Fig. 7),0 '80 and K were zero-phase filtered in the precession band(3rd order butterworth 15-28 kyr) and compared to the precession index (PI) ofBerger and Loutre (1991). All signalsexhibit similar amplitude modulation and fairly stable phase relations. (b) Comparison of cross-covariance func­tions ofPI and precession-filtered proxies. The centralmaximum (shown magnified in (d)) indicates the best matchofdelayed metronome record and proxy response. Estimated lag values are 1.77 kyr for susceptibility and 5.67 kyrfor 0'80. (c) The phase spectra of PI and unfiltered proxy records (derived from cross-spectral analysis) were con­verted to 'lag spectra' to show their frequency-dependence and good coincidence with lag values obtained from(b). Enor bars represent 80% confidence intervals, corresponding auto-spectrum amplitudes are shaded gray. (d)Close-up at cross-covariance functions of (b).

error source is age-depth-nonlinearity (Schiffelbeinand Dorman 1986). The applied (linear) age inter­polation does not consider fluctuations ofthe sedi­mentation rate between tie points. The relative agediscrepancies between CEARIS cores due to thiseffect should on average be less than 1 kyr, as thematched signal features are densely spaced andtheir high resemblance implies regionally uniformsedimentation conditions. A quasi-continuous pat-

tern correlation (Martinson et a1. 1982) may fur­ther reduce this margin. The synchronism with theexternal forcing signal is potentially more affectedby interpolation elTors. Its precision can be roughlyestimated from cross spectral analyses and is in theorder of 2 kyr. The absolute enor margins of or­bital tuning are even larger, but difficult to quan­tify. They result from uncertain and unstable phaserelations (Martinson et a1. 1987), vertical mixing by

24 Chapter 2

E:.ciliQ)ü({)

:::J(J)

Ü

~COl<U 1521-1::2:

1523-1

PI

"0 1515-1~

~li=

~ 1517-1-'"coN

1519-1J:,~

"0C<U.0 1521-1c0'(ij 1522-2(()Q)ü~ 1523-1

0..

"0Q)

~ 1515-1li=

'C' 1516-2>.-'"L{)

1517-1~

Y...-"0 1519-1c<U.0

1520-2.c.B.:;

1521-10-'"c~ 1522-2~..6 1523-1:::J

(J)

rso

r1SOI150

EW 250~ , ' " ",'--'"a:ü ••.•..•..•...•..•....••.•...••......•.•.••..• \. / ..•.•.••. /< / .«~a1::.,',,",,'., " ',. , ',',,,'.. ,,/ ... (. ,.,.. ,... ,... ,." ,',.,•,

o 20 40 60 80 100 120 140 160 180 200 220 240 260 280 300 320 340 360 380Age [ka]

Fig.l0. Tlu'ee step refinement ofage models for CEARlS eores. (a) Graphie eorrelation to 8180 age model of eareGeoB 1523-1 (48 tie points). (b) Initial (dashed) and subsequent stage (solid) of synehronizing all bandpass filteredsuseeptibility signals to the astronomieal preeession index (lag 1.77 kyr, 33 tie points). (e) Graphie eorrelation ofsub-Milankoviteh band « 15 kyr) features to their eonm10n mean age (90 tie points). (d) CEARIS stack.

Using rock magnetic proxy records for orbital tuning and extended time series analyses 25

bioturbation (Berger and Heath 1968) and an in­creasing error ofastronomical calculations with age(Berger and Loutre 1992).

Specijic Aspects ofTuning Rock Magnetic

Records

Orbital tuning ofproxy records implies the assump­tion that 21 parameter is almost exclusively respond­ing to primary orbital forcing and that the responsefunction is not misleading the correlation ofproxyand target signals. Saturation effects, competinganticorrelated orbital response mechanisms andsecondary overprinting will bias the tuning ap­proach and eventually make it pointless. It may bepossible to master such problems with specificallydesigned target curves (e.g., Berger ami Jansen1994(1), but this strategy requires 21 deep under­standing 01'2111 processes involved. Obviously thereis 21 need for simpler criteria defining whichsediments are appropriate for orbital tuning. A fewgeneral rules 1'01' rock magnetic age modelling invarious depositional environments can be outlined.

01' 2111 rock magnetic records those from sub­or anoxic sediments are the most precarious as theyare usually affected by magnetite reduction. Thesecomi example in the paper by Frederichs ct 211.

(this volume) illustrates, how multi-parametricmagnetic methods can be used to reconstruct 21

susceptibility record which has been diageneticallyoverprinted during temporary suboxic conditions.

An abrupt coarsening 01' fining ofthe magneticmineral fraction as indicated by magnetogranulo­metric proxies such as M IM (ibid.) reveals

ar Ir

changes in the sedimentation modus, that probablyalso mask paleoceanographic inf01mations 01'otherproxy parameters. Whatever may cause this distur­bance - erosion, winnowing, reductive diagenesis,slumping 01' intercalated turbidite 01' tephra layers ­the respective sections should be carefully studiedand eventually dissected. Tbe remaining record canstill be reassembled and tuned, if 21 pattel11 corre­lation with undisturbed records permits 21 convinc­ing quantification 01' resulting age gaps.

We found the simultaneous orbital tuning of5­20 regionally related cores much more conclusivethan 21 treatment of individual records, because itgives 21 better impression 01' pattel11 variability in

problematic sections and c1ues 1'01' conelation un­certainties. Parallel core processing is suppen'tcd bythe speed 01' experimental data acquisition in envi­nmmental magnetism.

Partial 01' total carbonate dissolution does notaffect the stability ofthe magnetic mineral fractJon.However, without carbonate or opal sedimentation(pelaglc c1ay) the most important mechamsm ofsusceptibility modulation, the mutual dilution ofterrigenous and biogenic sediment fractions (Robin­son 1990; Diester-Haass 1991; Mienert and Chi1995) is missing. Unless orbital variations inducecompositional changes within the lithogenic frac­tion, concentration-dependent magnetic parameterswill merely reflect consolidation \vith depth. In thisas well as other cases ofpurely terrigenous sedi­mentation, it is advisable to base orbital tuning uponmagnetomineralogic (e.g., hematite-magnetite ra­tio) 01' -granulornetric parameters (e.g., SOH 01'

M IM). These records often carry 21 climatic sig-;}l Ir

nal related to varying somce regions 01' transport(oscillating eolian 01' glaciomarine fluxes, bottomcurrent activity).

In oceanic regions with very low, inegular 01'

high sedimentation rates, rock magnetic records de­viate more or less severely from standard pattel11s(as do other types 01' records). A promising agemodelling strategy is to start out with establishingsusceptibility-based core-to-core conelation frame­works. When 2111 inconsistencies in this multiplecorrelation scheme are resolved, the most convinc­ing and complete record is dated by the best avail­able method and may then serve as initial age model1'01' subsequent tuning ofthe entire collection.

In principle, magnetostratigraphy provides idealinitial age models 1'01' orbital tuning of rock mag­netic parameters. Paleo- and rock magnetic inves­tigations can share the same sampIe set which isadvantageous for cross-validation. At least onecore reaching Matuyama age within 21 set 01'

magnetically paralleled cores is needed to assign 21

rough age frame to 2111 others. Although numerousShmi inverse polarity events ofthe Brunhes Chron« 0.78 Ma) have been detected (Nowaczyk et 211.

1994) their chronostratigraphic use is severely ham­pered by lithology-dependent lock-in effects (Bleiland von Dobeneck, this volume).

26 Chapter 2

Extended Time Series Analyses

Having tuned a rock magnetic record to basic as­tronomical cycles in the Milankovitch range givesaccess to time series analyses into the adjaccntfrequency bands. These are essential to investigatethe temporal evolution and physical principles 01'proxy response to orbital forcing. Suitable methodsto analyze super- (> 100 kyr) and sub-Milankovitch« 18 kyr) signal variations with and without thevie'Vvpoint 01' orbital forcing are presented in thefollowing.

Evolutionary spectral analysis requires long timeseries and is therefore demonstrated for the SUSASstack covering the last 1.5 m.y. The high temporalresolution ofthe CEARIS stack makes it a suitableexercise for spectral analyses in the sub-Milanko­vitch frequency range.

Super-lvfilankovitch Signal Variations 0/theSUSAS Stack

Linear orbital forcing through insolation variationscan only accOlll1t 1'01' proxy oscillations in the pre­cession and obliquity band (Imbrie et al. 1992). The96 kyr, 127 kyr and 404 kyr eccentricity cyclesmodulate the precession amplitude, but producenegligible insolation changes in their own frequencyrange (Fig. 12a). Yet, the SUSAS stack like otherPleistocene climate records exhibits strong 100 kyrvariance (Fig. 12c).

Many attempts have been made to explain this, 100 kyr cycle problem', frequently invoking non­linear processes channeling energy into the 100 kyrband (1'01' a summary see Imbrie et al., 1993). Inrejecting the assumption that variations in insola­tion alone are responsible for the observed climatic

changes, MuHer and MacDonald (1995) explain the100 kyr cycle by inclination changes ofthe Earth' sorbital plane relative to the sym111etry plane ofthesolar system. Liu (1992; 1995) concludcs tha1 1're­quency variations o1'the obliquity cycle can give riseto strang 100 kyr forcing 01' climate. Like variousothers, Raymo (1997) fa vors the idea that eccen··trici1y modulation ofthe precession index amplitudegenerates quasi -periodic 100 kyr ice age eyc1esThe climate changes from iee shield growth todecay, when summer insolation exeeeds a certainthreshold value. A 10ng-term reduction 01' the a1­mospheric pC0

2level (Saltzman and Verbitsky

1993), eventually caused by tectonic processes(Raymo et al. 1988; Raymo ami Ruddiman 1992),is thought to have induced a s1eady rise of the in­solation 1hreshold tor 1arge-sca1e melting in Quater­nary times. This may exp1ain, why late Plioeene joearly Pleistocene paleoclimate records exhibitmainly obliquity and precession related variance(e.g., Raymo et al. 1989; Bloemendal anddeMenocal 1989; deMenocal 1995), while latePleistocene climate f1uctuations, in spite ofsimilarinsolation variations, are dominated by 100 kyrcycles (e.g., Hays ct al. 1976; Ruddiman et al.1989). The exact timing ofthis mid-Pleistocene cli­mate transition and the question whether it was agradual Cmid-Pleistocene evolution', Ruddiman etal. 1989) or a sudden change Cmid-Pleistocene re­volution', Berger et al. 1994) have been subject 01'many studies (Shackleton and Opdyke 1976; Pisiasand Moore 1981; Prell 1982; Ruddiman et al. 1989;Berger and Jansen 1994b; Mudelsee andStattegger 1997) and attempts to model or char­acterize the transition with different statistical tech­niques (e.g., DeBlonde and Peltier 1991; Park andMaasch 1993; Mudelsee and Schulz 1997).

Fig. 11. Temporal analysis ofvarying frequency content ofthe SUSAS stack. From bottom to top: Lowpass filter­ing of the normalized susceptibility stack reveals three distinct periods with different base levels and transitionscentered at around 0.95 and 0.6 Ma. Extracting the basic Milankovitch cycles by bandpass filtering depicts fre­quency shifts within the eccentricity band and fading of the obliquity related variance in early Pleistocene.Precessional amplitudes are generally low due to bioturbation damping, particularly in periods of lower sedimen­tation rates. The envelopes were derived by Hilbert transformation (Bendat and PiersoI1986). For the evolutionaryspectral analysis (see text) 500 kyr seetions ofthe highpass filtered SUSAS signal (purpie) progressing in 10 kyrsteps are tapered and transformed into the frequency domain. Related to the center of each section, the resultingspectra are color-coded and form a so-called 'spectrogram'. The colored horizontal bars represent the limits usedin the bandpass analyses below.

Using rock magnetic proxy records for orbital tuning and extended time series analyses 27

""""

1.6 1.71.2 1.3 1.4 1.51.0 1.10.8 0.9Age [Mal

0.4 0.5 0.6 0.70.0 0.1 0.2 0.3

0---1---""""""'"

i\ n A n1 1\ A 1\ 1\ 1/\

0.5 \JI !VI 1\ I~\ MI0:· llAi· \vv \~ V\0.5

0~_.Y1A

g -0.5

0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0 1.1 1.2 1.3 1.4 1.5 1.6 1.70.07--j--'-'-...L.L-l--'--'--'-'-~_

0.00

1

0.02

}~ 0.04>.CJcQ)

:::J 0.030-

~LL

0.05

0.06

:0~~ 0.5

~ 0~ -0.5~

~i.L

g:0~Q)CJlFl:::J

Cf)

goZ

28 Chapter 2

0.07 -00(1)...,

::::!. 2':0 .....c.. DltJl

0.06 ~~

19.00.05

.-.::' 22.4>- 23.7

..l<::-- 0.04.--

~ 29c<I>

5- 0.03<I>.....

LL41

0.0254

0.01 96127

0.000.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0 1.1 1.2 1.3 1.4

0.07 -00(1)...,

:::::!. sr.0 .....c.. DltJl

0.06~

~~

19.00.05

.-.::'22.4

~ 23.72. 0.04

>-<.> 29c<I>

5- 0.03<I>.....

LL41

0.0254

0.01 96127

0.000.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0 1.1 1.2 1.3 1.4

Age [Ma]

Fig. 12. Evolutionary spectral analysis of(a) insolation and (b) adapted ETP signal (see text for explanations).

Using rock magnetic proxy records for orbital tuning and extended time series analyses 29

0.07 -00CD ...,...,0-o' ;::;:c..~(J)

0.06~.=.

19.00.05

'i::'22.4

~ 23.7;:: 0.04

>-<.l 29cQ)

5- 0.03Q)....

u..41

0.0254

0.01 96127

0.000.3 004 0.5 0.6 0.7 0.8 0.9 1.0 1.1 1.2 1.3 104

0.07 -00CD...,:::l.2:0 .....c..~(J)

0.06~

~.=.

19.00.05

'i::'2204

~ 23.7;:: 0.04

>-<.l 29cQ)

5- 0.03Q)....u..

41

0.0254

0.01 96127

0.000.3 004 0.5 0.6 0.7 0.8 0.9 1.0 1.1 1.2 1.3 1.4

Age [Mal

Fig. 12. Evolutionary spectral analysis of(c) SUSAS stack and(d) SUSAS-ETP residue (see text for explana-tions).

30 Chapter2

These controversies of 100 kyr cyclicity andPleistocene climate transitions essentially addressthe issue, whether long-term paleoceanographicvariations are related to basic Milankovitch cyclesand their superstructures (e.g. envelopes), or ratherto extraneous (forced by non-Milankovitch mecha­nisms) or free oscillations. A simple, frequentlyapplied method to analyze proxy signals in this con­text is multiple bandpass filtering. In the lower halfofFig. 11 five signal components ofparticular in­terest are extracted from the SUSAS record.

The prominent long-term trend obtained fromlowpass filtering (1 sI order Butterworth >350 kyr)is a remarkable feature not found in oxygen iso­tope signals and therefore not directly related toglobal ice volume. It is mirrored by the sedimenta­tion rate stack (Fig. 5) and also by CaC0

3concen­

tration (P. Müller, unpublished data), hence by car­bonate accumulation. Resembling more a doublebase line shift than a cycle, this signal marks twotransitions:- a first from 1.0 to 0.9 Ma corresponds to the ageofthe mid-Pleistocene (r)evolution,- a second from 0.65 to 0.5 Ma culminates in thelarge-scale sedimentation event outlined in Fig. 5.

Their duration may have been shorter, since low­pass filtering inevitably broadens step functions.Thethree age intervals delimited by these two transi­tions show different spectral characteristics in theMilankovitch bands. The precession-related signal(16-26 kyr) is poorly documented due to bioturba­tion damping with the exception of sections ofhigher sedimentation rates and enhanced forcing.Apart from superimposed amplitude fluctuations,the influence ofobliquity (35-47 kyr) on the SUSASstack diminishes continuously with a major de­crease near 1.2 Ma. The analysis in the 100 kyrrange was subdivided into two bands from 85 kyrto 110 kyr and from 110 kyr to 135 kyr to distinguishthe 96 kyr and 127 kyr eccentricity cycles. Theseperiods clearly dominate the signal in late Pleisto­cene, but also give important contributions to oldersections. As the amplitude variations in the 41 kyrand 100 kyr bands resemble in great detail the pat­terns found in 8180 records of Pacific and Atlan­tic sites (Park and Maasch 1993), we conclude thatthe observed features are of a global nature andclosely related to changes in ice volume.

Evolutionary spectral analysis is an excellentmethod to visualize the variability ofcyclic climatecharacteristics in arecord through time (e.g.,Pestiaux and Berger 1984; Bloemendal anddeMenocal1989; Mwenifumbo andBlangy 1991).Spectra generated within a moving window aredepicted in three-dimensional 'spectrograms' com­bining frequency and time domain (top ofFig. 11).These calculations were made using the'SPECGRAM' algorithm embedded in the'MATLAB Signal Processing Toolbox'. A 500 kyrframe was advanced at 10 kyr steps and, in orderto minimize cut-offeffects, tapered with a Hanningwindow, thereby focusing to the central section ateach step. Long-term trends (periods >350 kyr)were previously removed by highpass filtering(purple curve) to avoid disturbance by these unre­solved signal components. The resulting data ma­trix was normalized to an average spectral densityof 1by dividing all values by the total matrix mean.This procedure merely scales all spectrograms uni­formly and enables their comparison, but does notalter relative variations in the time or frequency do­malll.

In Figs. l2a-d spectrograms of forcing andresponding variables are related. The July mid­month insolation signal at 65°N (Berger and Loutre1991) and a normalized ETP curve are shown asreference. ETP curves are artificial, but often em­ployed target records composed by calibrating andadding the time series ofEccentricity, Tilt (obliq­uity) and Precession (Imbrie et al. 1984). Here, thecumulative spectral intensity in each band of theETP spectrogram was calibrated to equal that ofthe SUSAS spectrogram. Being based on the sameorbital variations, both reference signals exhibitidentical amplitude modulation patterns, but verydifferent spectral power in each band. In the ec­centricity band this difference amounts to severalorders ofmagnitude - an expression ofthe 100 kyrcycle problem. Characteristic Milankovitch super­structures as the cyclic 400 kyr modulation of the100 kyr, 54 kyr and 23 kyr cycles and the synchro­nous decay of the 100 kyr and 19 kyr cycles areclearly illustrated in Fig. 12b. The faint 29 kyr cy­cle is a subordinate component ofthe obliquity sig­nal and shows similarmodulation as the main peri­ods near 41 kyr.

Using rock magnetic proxy records for orbital tuning and extended time series analyses 31

The numerieal differenee between the speetro­grams ofthe SUSAS record and the ETP curve isdisplayed in Fig. 12d. For the two precessiona1bands at 19 kyr and 23 kyr, the assumption of aconstant proxy response appears justified, as bothrecords largely compensate. In contrast, the resi­dues in the 41 kyr band show a variable responseto obliquity forcing. Two marked spectra1 intensitydeclines at 1.2-1.1 and 0.7-0.6 Ma (Fig. 12e) can­not be exp1ained as a result of steady response toorbital forcing (Fig. 12b) and suggest that the po­tential of obliquity to drive the observed carbonatedissolution eycles decreased stepwise during PIeis­tocene times. This finding is in accordance withanalyses made for 8 180 records (e.g., Ruddimanet al. 1989; Joyce et al. 1990) and implies that theclimatic system was less sensitive to obliquity forc­ing during the late Pleistoeene. The coincidence oftarget and proxy records in the 100 kyr band iseven lower, as indicated by larger residues andsteeper gradients. While the ETP model looses100 kyr power (Fig. 12b), the SUSAS stack, likemost climate records, documents an intensificationof 100 kyr oscillations in late Pleistoeene.

This trend is interrupted by an early Pleistoeene100 kyr maximum appearing between 1.2 and1.1 Ma. This 'premature 100 kyr ineident', alsoobserved in 8 180 records (e.g., Mudelsee and Statt­egger 1997), merges with a phase of strong 70 kyreyelieity (Fig. 12c) which has also been found inother paleoeeanographic reeords (e.g., Ruddimanet al. 1989; Robinson 1990; Bassinot et al. 1994;Bolton et al. 1995). Two low-frequeney positivemaxima eentered at 0.95 and 0.6 Ma in the SUSASand the residual spectrogram (Fig. 12c, d) resultfrom remainders ofthe earlier diseussed base linetransitions of the long-term susceptibility trend(Fig. 11 bottom).

The most striking result ofthe residual spectro­gram is the reciprocity ofspectral intensities in the41 and 100 kyr bands (dashed outlines in Fig. 12d).In relation to the stationary ETP model, the SUSAS100 kyr cyclicity develops over-proportiona11y inseetions with reduced response to obliquity (blueoutline) and retreats, where the response to obliq­uity is strong (yellow outline). These findings im­ply an exchange of spectral energy between ob­liquity and eccentrieity in overall aeeordanee with

insolation threshold models (e.g., Saltzman andVerbitsky 1993; Raymo 1997). Intensified northernhemisphere summer insolation eapable oftrigger­ing major deglaeiations henee should result ti'om aninterferenee of obliquity maxima with preceSSlOnindex minima. At times, when the insolation thresh­old is relatively low, obliquity maxima alone willtrigger the withdrawal ofeontinental iee shields and41 kyr eyelieity prevails. When the insolationthreshold is higher, the required peak insolation 15

only reaehed by optimum interaetion ofobliquity andpreeession. As eeeentrieity modulates the preees­sion amplitude, suffieiently large preeessional peaksoccur only during 100 kyr eeeentrieity maximaeausing the observed 100 kyr elimate eycles.

Raymo (1997) assumes a linear, pCOlcontrol­

led rise of the insolation threshold over the past1.2 m.y. and obtained a single transition from 41 kyrto 100 kyr predominanee at about 0.7 Ma. Our evo­lutionary speetTal analyses (Figs. 12e, d) imply, thatthis tumover was predated by a 100 kyr episodearound 1.2 Ma.

Sub-Milankovitch Signal Variation5' ofthe

CEARIS Stack

A most interesting feature of a11 CEARIS reeordsare eonsistent high-frequeney variations in the sub­Milankoviteh to millennial range. As shown inFig. 1Oe, the < 15 kyr signal eomponents extraetedby high-pass filtering ean be matehed in a11 detail.Again, speetral analysis is a promising strategy todetect the environmental faetors eausing this unu­sua11y regular pattem. Stoehastie proeesses wouldyield smeared 'eolored noise' speetra with poorlydefined maxima, while periodie signals are repre­sented by mueh sharper peaks appearing at ehar­aeteristie frequeneies - under the premise that arecord is undisturbed, accurately dated, and cov­ering enough time to provide ample spectral reso­lution. Because these eonditions are never fully metin practiee, elaborate statistiea1 teehniques havebeen deve10ped to distinguish harmonie and randomsignal oseillations (e.g., Thomson 1982; for appIi­eations see Yiou et al. 1991; Nobes et al. 1991;Cortijo et al. 1995).

Sinee all fundamental Milankoviteh periods ex­eeed 18 kyr (e.g., Berger and Loutre 1992;

32 Chapter 2

rn,m'..-:r-:N:N:+, +'t.....:v:C'0:N

~~~E!~:

,",,'

'N:~...-, :

r..... : +;"<t"n.r..... :0.i:NC0:N"

N~:~

:t:~::t

o

Period g g 0 0 0 lD N 0 co <D lD '7 (0 N ~ 0

[kyr] _lD",1~~I~_<D'-,I..-'"--,>7--+-,(0,,,,1P+~~~''c-~LL,N+I_I~I---',._~-,I--!-+--,-I-,-L-+I ~...JI--!~I,--~_Iy-~-+-(J)'-I +1~-'--,L-l'f_+.---+-,-L-,-',--!~~~'~

:-:;r:~ fj::~: :

!.....:rt,:...,.:m: "'<(0:"1~: : 0!;[::I:l:::'+: m: ~:+:~

~. ~: ~:~:~::s :~: c: c' c

.c (f)o (j)

'5 :-eo 0

-,,<::"0C 0cu oe:,- (j)

20..

(j)

~ -20Cf)

l-::~v~~ ~::j (9

j ~:J' "0.. .

Cf) -80

~ -100o

0.. -60"0(j)

cu -80Et; -100'w

-120

-140o

-20

-40

-60"""rrrrtTJ'rn+"'fn""'H-rr<+ri1-n-.rrrtnirrn'>b'ncion'i,'rrl~Ti-rnci-rrr~""'ci-n'r~rrior,+rn+rn'rr+n;,..rt,,:,..,.,;,,;,.lJ..i,.rt-rI',.;'f,-,rln'rrn-~

o 0.01 0.02 0.03 0.04 0.05 0.06 0.07 0.08 009 01 011 0.12 0.13 0.14 015 016Frequency [1/kyr]

Fig. 13. Reference spectra (Blackman-Harris taper, WOSA: 3 segments) offundamental orbital frequencies andtheir low-order (:::;3) combinations based on astronomical time series (2500-0 ka; Berger and Loutre 1991). (Sub-)Milankovitch periodicities (vertical dorted lines) are classified in groups (gray bars) of common order (index num­bers) and origin (intermodulations (Int) and harmonics (Har». Crosses depict 6dB bandwidths and 90 % confi­dence intervals. (a) A simple nonlinear transform of (b) the 65°N July insolation signal shows spectral characteris­tics in common with (b), (e) obliquity and (d) precession signal, even in the sub-Milankovitch 8-18 kyrrange. Otherhigher-order cycles of (a) coincide with (e) squared precession and (f) coupled obliquity and precession.

Using rock magnetic proxy records for orbital tuning and extended time series analyses 33

co~

Q)

(IJu

(f)

uE:S'e:(IJDJo-l

:!:::N

b

(IJ....tlQ)Q.

(f)....Q)

:s:o0--0Q)

roE.~w

20 • •• •

101c((w"ioJm ... CD .

: :::. : : :::: :: :::: ::: :: :: ::"1""1""1"';"("':1":'1':""1"'''1';':1'':'i""'i",ii"i'I:"il"i'I""I"';I,""I""I""I'"

0.01 002 0.03 0.04 005 0.06 0.07 0.08 0.09 0.1Frequency [1/kyr]

roJe

: :: :

/M

Fig. 14. Extended spectral analysis ofCEARIS stack. References are (a) spectrull1 of Fig. 13a, (b) raw and (e) aver­aged periodograll1 ofsignal section spanning the CEARIS age range (377-6 ka). Ana10gous1y, (c) and (f) depict theraw and averaged periodograll1 ofCEARIS stack. In (d) the eight individual CEARIS periodograll1s are stacked. Theinitially narrow standard deviation band widens towards 11igher frequencies. The coherency spectrull1 (g) ofrefer­ence (e) and proxy signal (f) yields high squared coherencies not on1y far the basic Milankovitch periods but alsofar spectralll1axima near 30 kyr, 15 kyr, 13 kyr, 11.5 kyr, 10.5 kyr, and 9 kyr. COll1bined with the overall coincidence ofspectra1 patterns this identifies these CEARIS sub-Mi1ankovitch peaks as the hypothesized cOll1bination tones.

34 Chapter 2

Schwarzacher 1993) and known solar activity cy­cles Ce .g. sunspot, Haie ami Gleissberg eyeles) fre­quently deteeted in varved sediments have perIOds<200 years (Glenn and Kelts 1991), whieh penodi­eities may actua11y be expeeted in the speetralrange from 1 to 18 kyr? Besides fi'ee, quasi-pen­odie elimate t1uetuations, the most Iikely candidatcsare multiples ami eombination tones ofMilankovitehfrequeneies originating from nonlinear transforma­tions in the response ehain 01' insolation, elimate,proxy physics and sedimentation.

'I-Iarmonies' result from nonlinear response toa single orbital eyeIe, while 'intermodulation fre­queneies' arise from nonlinear eoupling 01' two ormore basIc orbital eyeles. The frequeneies ofthcseadditionalmodes are simply integer (k,l, 111, 11, ... )

eombinations ofthc fundamental foreing frcquen­eiesf;,j~.f;.f~, '" (Le Treut and Ghil 1983):

f (k ,I, /11. 11, ) = kir +I/i +/IIj] +IIj4 +

k,l,/II,II E{ ... ,-I,O, 1,2, .. };

oreler = Ikl +1II +1/111 +1111+..The number ofpotential combination tones withinan investigateel frequeney band can be boosteel aelIib by admitting high order eombinations involvingnegative coefficients. Coneerning orbital foreing,only the lowest order eombinations within a givenfrequeney range are bound to represent substan­tial speetral power (e.g., Yiou et al. 1991). Alllow­oreler (:S: 3) eombination tones calculated from obli­quity anel preeession and their realization and signi­fieance in the spectra ofvarious astronomical timeseries are shown in Fig. 13. The five most impor­tant eccentricity frequencies can be obtained as li­near combinations ofthe basic precession frequen­cies (e.g., 11404 kyc l = 1122.4 kyc l

- 1123.7 kyc l;

Berger and Loutre 1992). It is therefore unneces­sary to include them separately in the calculationofpotential orbital combination tones. I-Iowever, itis essential to distinguish not just two, but a11 threemajor precession frequencies.

Even pure obliquity and precession signals(Fig. 13c, d) and their transformation into latitudedependent insolation (Berger and Loutre 1994) in­volve (slightly) nonlinear process (Berger andPestiaux 1984) and display sub-Milankovitch peaksabove the noise floor of numerieal precision

(Fig. 13b). From the viewpoint ofphysieal ciimatemodels, non-linear response is rather the rulc thanthe exeeption. Not just iee-sheet dynamles CLeTreut and Ghil 1983), but also ü1ctors controllingmonsoon intensity (Short et al. 1991:. Crmvlcy d

al. 1992) favor the generation of orbital harmon­ics. Periods 01' 10-12 kyr, corresponding to a dou­bled preeesslOn cyele (simulated here by squaringthe precession index; Fig. 13e), appear in insola­tion (Fig. 13b: Berger and Loutre 1997) and proxyspeetra (Pestiaux et a1. 1988: Park et a1. 1993;Hagelberg et al. 1994), partieularly at low latitudes.The coupling ofobliquity ami precession (Fig. 131)generatcs charactenstie peaks in the 13-15 kyrband, whieh are also found in sediments (e,g ..Berger et al. 1991) and ice core reeords (Yiou etal. 1991). A simple non-linear mathematical trans­form oLm astronomical insolation signal (Bergeranel Loutre 1991) yields a nearly eomplcte spee­trum 01' a11 theoretiea11y derived orbital eombinationtones (Fig. 13a) and may be used as unbiased ref­erence to identify significant sub .. Milankovitchpeaks ofproxy spectra. (Fig. 14a). In order to pro­vide equivalent frequency resolution with respectto the CEARIS spcctra, the underlying time serieswas reduced to the same time range (377 - Cl ka:Fig. 14b,e).

The spectral amplitudes 01' insolation (Fig. 13b)fade steeply in the sub-Milankovitch range, whilethose 01' its nonlinear transform (Fig.13a/14a) andthe CEARIS record (Fig. 14e) decrease much less.This discrepancy implies spectl'al power channelinginto the sub-Milankoviteh band. By visual eompari­son ofthe synthetie spectra (Fig. 14b,e) with those01' the CEARIS stack (Fig. 14c,f) and the stack 01'the eight individual CEARIS speetra (Fig. 14d)many analogous peaks ean be deteeted. The mostprominent sub-Milankovitch peaks appear neal' 9,10.5, 11.5, 13 and 15 kyr. They eoineide with theearlier mentioned 100v-order eombination tones andharmonies ofobliquity and precession. A combinedevaluation ofmodel and proxy spectra and their co­herence speetrum (Fig. 14g) indicates, that the co­herence at these four sub-Milankovitch frequen­eies is significant at an 80% level and thus nearlyas high as at the basic Milankoviteh frequeneies.This result is statistically valid, but is based upon asynthetic and somewhat arbitrary reference eurve.

Using rock magnetic proxy records for orbital tuning and extended time series analyses 35

Int 1123-1/41

Int 1/41+1123-1/19

Int 1119-1123=Ecc 1/109

Period 500 100 60 40 30 25 20 18 16 15 14 13 12 11 10 91//1 I" , I, , ! 1 1 I,!, ! 1 ! , 1 ! 1 ! 1 1 1 , 1 , 1 , 1 1 1 , ! ! I I[kyr] cn cn cn cn

~~ ~ ~-- -- -.

'+ ~

(')';; M~ (') cn M (') (') cn

~~~ ~ ~~

~~

~ ~

~o --~W '+ '+ '+ ""+ (') '+0,11 M~

d, (') cn d,~ ~ (') N~

~ ~ ~

~(')

~ ~~ ",,"N

~ -- ~ ~ ~ ~-. -. :;:::c::J N~ :c ~

ffi:E :E :E ~ ~~ :E ~ ~ :E ~ ~

0 E <:: <:: n. n. E E I <::

0.05

'C'g 0.04--.......

N.......~ 0.03cQ):::s0-Q) 0.02....u..

0.01

0.00

I 'o

, , I

0.01I '

0.02I ' I " I"" I " I'

O.ro O.M O.~ QOO QWFrequency /1 [1/kyr]

I ' I ' ,0.08 0.09

, I

0.1, I

0.11

Fig. 15. Bispectral analysis of CEARIS stack (see text for explanation).

A more direct approach is bispectral analysis(e.g., Rao and Gabr 1984; Nikias and Raghuveer1987). This higher-order statistical method wasemployed by Muller and MacDonald (1997) toanalyze 100 kyr cyclicity and by Hagelberg et al.(1991) for detection of sub-Milankovitch climatevariations. In contrast to power and cross spectra,which regard individual or compare equal frequen­eies, bispectra are capable ofdetecting non-linearcoupling ofdifferent frequencies. A phase-lock be­tween ahypothetical combination frequencyJ; andits assumed generating frequenciesJ; and}; is con­dition for a peak in the mirror-symmetric, two-di­mensional diagram. Noise components and ran­dom oscillations are efficiently suppressed.

The bispectrum ofthe CEARIS stack (Fig. 15)displays a considerable number of peaks, mostlyatthe intersections ofgrid lines representing a sim­plified set of basic Milankovitch frequencies andtheir low-order combinations. A combination fre­quencyJ; can result from different pairs (fJ:). Forevident mathematical reasons these realizations arepositioned along diagonallines. Besides basicobliquity and precession peaks resulting from com­binations oftheir own intennodulation frequencies,varlous low-order hannonics (as 2/23 kyr- I ) and in­tennodulations (as (1/41+1/23) kyrl

) are detected.The four m~or sub-Milankovitch peaks ofthe cohe­rence spectrum (Fig. l4g) are validated as phase­locked combination tones of obliquity and preces-

36 Chapter2

o 20 40 60 80 100 120 140 160 180 200 220 240 260 280 300 320 340 360 380250 --!-'-T-'-~',-'-l-'-'--'-r'-'-'-'-.J,-L..L..L1+L-L.l---+-,---'--.L-'-+-~'-'-+--'--'---'--TL.Li--Y-..L.L.,+,c...L.L-"-T--'-'-L..y-'-'--'-'-..L.L.L.L.T'--'--'---'-f-..L.L.-',--'-'

200

150

100

::J500 U>o

§I450 g'~

400 320 40 60 80 100 120 140 160 180 200 220 240 260 280 300 320 340 360 380 ....!:l

Age [kaI

Fig. 16. (a) Susceptibility record ofGeoB 1523-1 (anows mark twin peaks conesponding to precession halIDonics).(b) idem, lowpass filtered (2nd order ButterwOlih) to exc1ude (> 18 kyr, dotted line) and inc1ude (> 7 kyr, solid line)sub-Milankovitch hmIDonics. (c) The residue ofboth filtered signals (7-18 kyr) frequently shows deep minima inphase with insolation maxima (dotted verticallines), followed by twin maxima (shaded as in b) before and afterinsolation minima. (d) The millennial band « 7 kyr) signal component appears to be aperiodic, although in someseetions, its envelope (shaded) resembles curve c (see also Fig. 16) (e) Precession index (solid line), shifted by-1.77 kyr (dashed line), and 15°N July insolation curve (shaded) by Berger and Louu'e (1991).

sion. Two further peak ehains near 65 kyr and30 kyr are identified as third-order intermodulations((1/41+1123-1/19) kyr J and(1I41-1I23+1I19) kyr 1

)

within the Milankoviteh band. Their important eli­matie implieation will be discussed below.

In essence these results suggest that large partsof the observed sub-Milankoviteh signal compo­nents down to periods of at least 10 kyr can beexplained by non-linear c1imate response to orbitalvariations. Below 10 kyr 'subtone dispersion'comes into play (Fig. 14). Preeession and obliquityeyc1es are given as trigonometrie expansions ofseveral sinusoidal signals, most ofthem with verysimilar frequeneies (Berger and Loutre 1992;Sehwarzaeher 1993). The frequeneies oftheir see­ond and third order harmonies are still c10se enoughfor their peaks to merge. With inereasing order, the

differenees beeome larger and eaeh intermodula­tion peak splits up into a peak ehain, eventually in­terfering with other peak ehains ... so that evenprominent and signifieant spectral peaks are nomore safely identified. Considering that for theCEARIS eores the estimated age uneertainty is inthe order of 2 kyr, there is no sense in searehingfor orbital effeets mueh belowperiods on kyr. Therelative increase of the noise eomponent withhigher frequeney, resulting from c1imatic and depo­sitional random proeesses and stratigraphie elTor,is doeumented by the widening of the standarddeviation band in Fig. 14d.

In view of paleoeeanographie interpretation,signal phase and regularity are almost as importantas frequeney and amplitude. A band seleetive timedomain representation ofthe suseeptibility reeord

Using rock magnetic proxy records for orbital tuning and extended time series analyses 37

! I, ! f I ! ! ! ! I ! I ! ! I! '! I~~t ! ! ! ! I ! ! , I I ! , , ! j , !...J.....LJ.! I , I ! ! ! ! ! I I ! ! , ! , I LLLLJ...J-L1..L..L..LLLLLLLjI! ! I I!

50- Gi)(j)

<0

b,...- 100

'0QJt 150QJ

>c=~ 200

>,

'}0~

r---..vD~

~4=

~ 20

0'~

0ro

ce

0...er:l')

0 10 20 30 40 50 60Age [kaI

70 80 90 100 110

Fig. 17. Tentative conelation of 110-0 ka section of (a) susceptibility record of GeoB 1523-1, (b) idem, highpassfiltered (2ml order Butterworth < 7 kyr) to extract millennial scale variations, to (c) GRIP 8180 record on Dansgaardet al. (1993) age model. Labels mark Dansgaard-Oeschger cycles 1-24, YOlll1ger Dryas, and Heinrich events HI-H6.As considerable disagreement over details of the GRIP timescale persists (e.g., Hammer et al. 1997; .lohnsen et al.1997; Adkins et al. 1997), the conelation ages are not sufficiently reliable to evahiate the precision oforbital tuning.

of core GcoB 1523-1 is shown in Fig. 16 for theMilankovitch (> 18 kyr), sub-Mi1ankovitch range(18 to 7 kyr) and millennial range « 7 kyr). The7 kyr boundary is clearly a statistical and not apaleoclimatic limit.

The coherent orbital periodicities in the Milan­kovitch band (Figs.14b, c, 16b) suggest that globalglacial interglacial sea-level changes modulate theflux of(Fe-rich) Amazon sediments from the shelfregion into the adjacent pelagic realm. Only tropi­cal insolation variations can explain the much higher23 kyr spectral power of K compared to 8 180(Fig. 8) and precessional susceptibility variationsprior to Northern hemisphere glaciation at theCeani Rise (HalTis et al. 1997). At low latitudesprecession large1y controls insolation (Fig. 16e,Berger and Pestiaux 1984) and modulates sea sur­face temperature, monsoon intensity and nutriclinedepth (e.g., McIntyre et al. 1989). Periods of el­evated summer insolation result in increased car-

bonate accumulation (Rühlemann et al. 1996) andreduced Saharan dust deposition (Sarnthein ct al1981; Balsam et al. 1995) in the western equato­rial Atlantic. Both effects lower the precessionalsusceptibility signal (Fig. 16c,e). In periods of re­duced insolation, 'twin peaks' of susceptibility arefrequently observed - not just in the filtered(Fig. l6c), but also in the raw signal (anows inFig. 16a). They are therefore authentic paleocea­nographic features and not artefacts of signal dis­tortion by varying sedimentation rates (Martinsonet al. 1982; Schiffelbein and Dorman 1986). De­tailed rock magnetic analyses (T. von Dobeneck,unpublished data) indicate that the earlier ofbothpeaks carries more Saharan, the second moreAmazonian characteristics. Together they reflectthe double precession cycle ofmonsoon intensity.

Further support to pC02

threshold models(' 100 kyr cyc1e problem'), discussed in the contextof SUSAS time series analyses, comes from the

38 Chapter 2

extraordinarily high bispectral intensities ofthe in­termodulation frequencies (1/41 + 1/23-1/19) kyc l

and (1/41-1/23+1/19) kyc l. Both combination

tones express nonlinear interaction of obliquitywith the envelope ofthe composed precession sig­nal, which has a frequency of(-1/23+1/19) kyc l

,

corresponding to an (averaged) eccentncity fre­quency of 1/109 kyc l

. This interaction provides thetrigger mechanism 10r the late Quaternary 100 kyrice age cycles.

Millennial climate cycles are believed to resultfrom free, self-sustained glaciomarine oscillationsas proposed by many coupled climate models (e.g.,Birchfield et a1. 1994; Paillard 1995). Large millen­nia1 temperature oscillations documented by icecore 8 180 studies from both hemispheres (Dans­gaard et al. 1993; Bender et a1. 1994) have recent1yalso been 10und in various marine high-resolutionproxy records (Behl and Kennett 1996; Charles etal. 1996; Moros et a1. 1997; Adkins et a1. 1997),with some reservations also at the Cearä Rise(Curry and Oppo 1997). It is therefore not overJyambitious to tentatively cOlTe1ate the millennia1signal component ofGeoB 1523-1 (Fig. 16d) withthe GRIP 8 180 record (Fig. 17). Although bothpattems are not exactly minor images, the simi­1arity is quite reasonable given the fact that a 2.8 mand a 2800 m record from two climate extremesare compared.

Conclusions

The SU SAS and CEARIS studies demonstratehow the methods ofcyclostratigraphy and environ­mental magnetism can be fruitfully combined.Other than most physical properties parameters,magnetic susceptibility depends linearly on theconcentrations of a few (iron-bearing) minerals,predominate1y (titano-) magnetite. In most marineenvironments its mean value and climatic modu­lation primarily mÜTors the ratio oftenigenous andbiogenic sediment accumulation. Susceptibilityvariations can be elosely related to carbonate dis­solution cycles as in case ofthe SUSAS cores, butmayaiso primarily respond to climatic cyeles ofterrigenous supply as in the CEARlS example. Thebasic mechanisms of these two cycles are weIlstudied and largely understood. They are effects

of deep water formation, sea level change, aridityand wind intensity, all intimately linked to orbitalcycles. Although the phase relations 01'8 130 and K

differ regionally, the orbital response of suscepti­bility to climatlc change is o1'ten adequate forcyc10stratigraphy and orbital tuning. PiJot spectralstudies on selected cores yield applicab1e phase lagsand indicate best tuning targets.

Particularly in low-accumulation environmentsphase settll1gs are not as important as pattern re­producibility. Because susceptibility Jogging gener­ates a complete low-noise image based on thesediment volume, subtle signal characteristics areresolved enabling pattem identification even in prob­lematic sections. Aperiodic trends, shifts and am­plitude fiuctuations of the susceptibility recorddocument modifications o1'the depositiona1 system.Extracted by lowpass filtering and evolutionm'yspectral analysis, these features may have moreinteresting climatic implications than a perfect or­bital correlation scheme. In the SUSAS example,analogue long term trends 01' subtropical SouthAtlantic K and %CaCO, and global 8 l3C (but not(

180) records indicate marked changes at around0.95 and 0.6 Ma. Repeated converse shi1'ts 01'100 kyr and 41 kyr spectral power relative to anETP model substantiate pC0

2threshold concepts

(e.g., Sa]tzman and Verbitsky 1993; Raymo 1997)ofthe mid-P1eistocene transition.

In addition, most high-resolution susceptibilityrecords calTY important signal components in thesub-Milankovitch and millennial bands. In case ofthe eight CEARIS records these high-frequencyvariations are largely coherent and can be used toimprove the intemal precision of the combinedcyclostratigraphic age model. As shown by spec­tral and bispectral analysis, the signal modulationin the 7-18 kyr band is mainly due to periodic har­monics and combination tones of orbital frequen­cies such as the doubled precession cyele predictedfor tropical climate by models (Short et al. 1991).Similar spech'al interpretations in the < 7 kyr bandare inhibited by the complexity ofcombination tonesand uncertainty ofthe age model. A tentative cor­relation of millennial CEARIS variations toDansgaard-Oeschger cycles (Dansgaard et al.1993) yields quite convincing results within the limi­tations ofboth age models. Both categories ofhigh-

Using rock magnetic proxy records for orbital tuning and extended time series analyses 39

frequency variations open attractive perspectivesto improve chronostratigraphic precision beyondthe present state of the art.

Acknowledgments

We thank U. Bleil for many helpful disCllssions amiconstructive criticism. Reviews and many valuabJesuggestions by W.H. Berger and M. Schulz aregratefully acknowledged. T. Bickert kindly pro­vided unpublished 8 11C data ofODP Site 806. Tbisstudy was funded by the Deutsche Forschungs­gemeinschaft (Sonderforschungsbereich 261 atBremen University, contribution No. 219). F. S. wassupported by the Deutsche Forschungsgemein­schaft in the framewor1c ofGraduiertenkolleg 221.

Data are available under www.pangaea.de/Projects/SFB26l.

References

Aelkins JF, Boyle EA, Keigwin L, Cortijo E (1997) Vari­ability of the North Atlantie thermohaline eireula­tion eluring the last interglaeial period. Nature390:154-156

Balsam WL, Otto-Bliesner BL, Deaton BC (1995) Mod­ern and last glaeial maximum eolian sedimentationpatterns in the Atlantie Oeean interpreteel from seeli­ment iran oxide content. Paleoeeanography 10:493-507

Bassinot FC, Beaufort L, Vineent E, Labeyrie LD,Rostek F, Müller PJ, Quielelleur X, Laneelot Y(1994) Coarse fraetion fluetuations in pelagie car­bonate sediments from the tropical Indian Oeean:a 1500-kyr reeorel of carbonate dissolution.Paleoceanography 9:579-600

Behl RJ, Kennett JP (1996) Briefinterstadial events inSanta Barbara basin, NE Paeifie, during the past 60kyr. Nature 379:243-246

Bendat JS, Piersol AG (1986) Random elata. Wiley, NewYork,pp 1-566

BeneleI' M, Sowers T, Diekson ML, Orehardo J,Grootes P, Mayewski PA, Meese DA (1994) Climatecorrelations between Greenland anel Antaretiea dur­ing the past 100,000 years. Nature 372:663-666

Berger A, Pestiaux P (1984) Accuracy and stability oftheQuaternary terrestrial insolation. In: Berger A,Imbrie J, Hays J, Kukla G, Saltzman B (eds)Milankovitch anel Climate Pali 1, Reidel, Dorelreeht,pp83-ll1

Berger A, Loutre MF (1991) Insolation values for the

climate of the last 10 million years. Quat Sei Rev10:297-317

Berger A, Meliee JL, Hinnov L (1991) A strategy for ti'e­queney speetra ofQuaternary climate recorels. ClimDyn 5:227-240

Berger A Loutr'e MF (1992) Astronomieal solutions forpaleoelimate studies over the last 3 million years.Earth Planet Sei Lett 111 :369-382

Berger A Loutre MF (1994) Preeession, eceentricity,obliquity, insolation anel paleoclimates.ln: DuplessyJC, Spyrielakis MT (eds) Long-term climatie varia­tions. Springer, Berlin, pp 107-151

Berger WH, Heath GR (1968) Vertieal mixing in pelagicseeliments. J Mar Res 26: 134-142

Berger WH, Yasuda MK, Biekert T, Wefer G, TakayamaT (1994) Quaternary time seale for the Ontong JavaPlateau: Milallkoviteh template for Oeean DrillingPragram Site 806. Geology 22:463-467

Berger WH, Jansen E (1994a) Fourier stratigraphy: spec­tral gain aeljustment of orbital iee mass models as anaid in elating late Neogene deep-sea seeliments.Paleoeeanography 9:693-703

Berger WH, Jansen E (1994b) Miel-Pleistoeene climateshift - the Nansen conneetion.ln: JoharUlessen OM,Muench RD, Overlanel JE (eels) The polar oceans andtheir role in shaping the global environment.Geophys Monogr Ser 85, AGU, Washington, pp 295­311

Biekert T (1992) Rekonstruktion der spätquartären Bo­elenwasserzirkulation im östlichen Süelatlantik überstabile Isotope benthischer Foraminiferen. PhD the­sis, Bel' FB Geo Univ Bremen 27,pp 1-205

Bickert T, Wefer G (1996) Late Quaternal'y eleep watercireulation in the South Atlantic: reeonstruetion fromcarbonate elissolution anel benthie stable isotopes.In: Wefer G, Berger WH, Siedler G, Webb DJ (eels)The South Atlantie: present anel past eireulation,Springer Verlag, Berlin, pp 599-620

Biekert T, Curry WB, Wefer G (1997) Late Plioeene toHolocene (2.6-0 Ma) western equatorial Atlantieeleep-water eireulation: inferenees from benthie sta­ble isotopes. Proe ODP, Sei Results 154:239-254

Biekeri T, Berger WH, Wefer G (1998) TIle eleep westernequatorial Pacifie in Plio-Pleistoeene times: Resultsfrom Leg 130 (Ontong Java Plateau). Paleoceano­graphy (subm)

Birehfield GE, Wang H, Rieh JJ (1994) Century/rnillenililllinternaI clinlate oseillations in an oeean-atrnosphere­continental iee sheetmoelel. J Geophys Res 99: 12459­12470

Blaekman RB, Tukey JW (1958) The measurement ofpower speetr·a. Dover Publ Ine, New York, pp 1-190

40 Chapter 2

Bloemendal J, Lamb B, King J (1988) Paleoenvironmentalimplicalions ofrock-magnetic properties oflate Qua­ternary sediment cores from the Eastern EquatorialAtlanlic. Paleoceanography 3:61-87

Bloemendal J, deMenocal P (1989) Evidence for a changein the periodicity oftropical climate cycles at 2.4 Myrfrom whole-core magnetic susceptibility measure­menls. Nature 342:897-900

Bolton EW. Maasch KA, Lilly JM (1995) A wavelet analy­sis of Plio-Pleistocene climate indicalors: A newview of periodicity evolution. Geophys Res Letl22:2753-2756

Bond G, Broecker W, Johnsen S, McManus J, LabeyrieL, Jouzel J, Bonani G (1993) Correlations betweenclimate records from North Atlantic sediments andGreenland ice. Nature 365: 143-145

Charles CD, Lynch-Stieglitz J, Ninnemann US, FairbanksRG (1996) Climate connections between the hemi­sphere revealed by deep sea corehce core correla­tions. Earth Planet Sci Lett 142: 19-27

Chi J, Mienert J (1996) Linking physical property recordsof Quaternary sediments to Heinrich events. MarGeology 131:57-73

COliijo E, YiouP, Labeyrie L, Cremer M (1995) Sedimen­tary record ofrapid elimatic variability in the NorthAtlantic Ocean during the last glacial cycle.Paleoceanography 10:911-926

Crowley Tl, Kim KY, Mengel JG, Short DA (1992)Modeling 100,000 year climate fluctuations in pre­Pleistocene time series. Science 255:705-707

Curry WB, 01'1'0 DW (1997) Synchronous, high-fre­quency oscillations in tropical sea surface tem­peratures and North Atlantic Deep Water produc­tion during the last glacial cyele. Paleoceano­graphy 12:1-14

Dalfes HN, Schneider SH, Thompson SL (1984) Effectsofbioturbation on climatie spectra inferred from deepsea cores. In: Berger A, Imbrie J, Hays J, KuklaG, Saltzman B (eds) Milankovitch and Climate Pali 1,Reidel, Dordrecht, pp 481-492

Damuth JE (1977) Late Quaten1a1'y sedimentation in thewestern equatorial Atlantic, Geol Soc America BuH88:695-710

Dansgaard W, Johnsen SJ, Clausen HB, Dahl-Jensen D,Gundestrup NS, Hammer CU, Hvidberg CS,Steffensen JP, Sveinbjornsdotiir AE, Bond G (1993)Evidence for general instability ofpast climate froma 250-kyr ice-cOl'e record. Nature 364:218-220

DeBlonde G, Peltier WR (1991) A one-dimensional modelof continental ice volume fluctuations through thePleistocene: implications for the origin ofthe Mid­Pleistocene climate transition. J Clim4:318-344

deMenocal P (1995) Plio-Pleistocene African climate.Science 270:53-59

deMenocal P, Bloemendal J, King .I (1991) A rock­magnetic record ofmonsoonal dusl deposition tothe arabü1l1 sea: evidence for a shirt in the modeof deposition at 2.4 Ma. Proc ODP. Sci Res\llts117:389-401

Diester-I-Iaass L (1991) Rhythmic carbonate contclltvariations in Neogene sediments above lhe oceaniclysocline. In: Einseie G, Ricken W. Seilacher A (eds.)Cycles and events in stratigraphy, Springer Verbg,Berlin, pp 94-109

Flanders PJ (1988) An allernating gradient force mag­netometer,.1 Appl Phys 63:3940,3945

Franeois R, Baeon MP( I991) Variations in terrigenousinput into the deep Equatorial Atlantic during the past24,000 years, Science 25 1: 1473- I475

Glenn CR, Kelts K (1991) Sedimentary rhythms in lakedeposits. In: Einseie G, Ricken W, Seilacher A (eds.)Cycles and events in stratigraphy, Springer Verlag.Berlin, pp 188-22 I

Grousset FE, Labeyrie L, Sinko JA, Cremer M, Bond G,Duprat.l, Cortijo E, Huon S (1993) Patterns ofice­rafted detritus in the glacial North Atlantie (40-55°N).Paleoceanography 8: 175-192

Hagelberg TK, Pisias N (1991) Linear amI nonlinear cou­plings between orbital forcing and the marine BIROrecord during the late neogene. Paleoeeanography6:729-746

Hagelberg TK, Bond G, deMenocal P (1994) Milanko­vitch band foreing ofsub-Milankovitch climate vari­ability during the Pleistocene, Paleoceanography9:545-558

Hammer CU, Andersen KK, Clausen HB, Dahl-JensenD, Hvidberg CS ,Iversen P (1997) The stratigraphicdating of the GRIP ice core, Special Report of theGeophysical Department, Niels Bohr Institute forAstronomy, Physics and Geophysics, University ofCopenhagen

Harris, FJ (1978) On the use ofwindows for harmonicanalysis with the Discrete Fourier Transform. ProcIEEE 66:5 1-83

HaITis SE, Mix AC, King T (1997) Biogenic andterrigenous sedimentation at Ceara Rise, westerntropical Atlantic, supports Pliocene-Pleistocenedeep-water linkage between hemispheres. Proc ODP,SciResults 154:331-345

Hays JD, Imbrie J, Shackleton NJ (1976) Variations inEarth's orbit: pace-maker of the ice ages. Science194:1121-1132

Hilgen FJ, Krijgsman W, Langereis CG, Lourens LJ,Santar'elli A, Zachar'iasse WJ (1995) Extending the

Using rock magnetic proxy records for orbital tuning and extended time series analyses 41

astronomical (polarity) time scale into the Mioccnc.Earth Planet Sci Lett 136:495-510

Horowitz LL (1974) The effects ofspline interpolationon power spectral elensity. IEEE Trans ASSP 22:22­27

Imbrie J, Imbrie JZ (1980) Moeleling the climatic re­sponse to orbital variations, Science 207 :943-953

Imbrie J, Hays JD, Mal"!inson DG, Mclntyre A, Mix AC(1984) The orbital theory ofPleistocene climate: sup­port from a reviseel chronology ofthe marine 8 180recorel. 111: Berger A, Imbrie J, Hays J, Kukla G, Saltz­man B (eels) Milankovitch anel Climate Part 1, Reielel,Dorelrecht, pp 269-305

Imbrie J, Berger A, Boyle EA, Clcmens SC, Duffy A,Howarel WR, Kukla G, Kutzbaeh.J, Martinson DG,MeIntyre A, Mix AC, Molfino B, Morley JJ, PetersonLC, Pisias NG, Prell WL, Raymo ME, Shackleton N.J,Toggweiler JR (1992) On the strueture anel origin ofmajor glaciation cyc1es 1. Linear responses to Milan­kovitch forcing. Paleoceanography 7:701-738

Imbrie J, Berger A, Boyle EA, Clemens SC, Duffy A,Howarel WR, Kukla G, Kutzbach J, Martinson DG,McIntyre A, Mix AC, MolfillO B, Morley JJ, PetersonLC, Pisias NG, Prell WL, Raymo ME, Shackleton NI,Toggweiler .JR (1993) On the structure anel origin ofmajor glaeiation eyc1es 2. The 1OO,OOO-year eyc1e.Paleoeeanography 8:699-735

IolU1S WE, Schott FA, Zantopp RJ, Evans RH (1990) -D1eN Ol"!h Brazil Current retrofleetion: seasonal stmctureanel eelely variability, I Geophys Res 95 :221 03-22120

Iolmsen SJ, Clausen HB, Dansgaarel W, Gunelestrup NS,Hanmler CU, Anelersen U, Anelersen KK, HvielbergCS, Dahl-Jensen D, Steffensen IP, Shoji H,Sveinbjörnsel6ttir AE, White IWC, Jouzel I, FisherD (1997) The 8 180 recorel along the Greenlanel IceCore Project eleep iee eore anel the problem ofpossi­ble Eemian elimatie instability. I Geophys Res102:26397-26410

Ioyee JE, Tjalsma LRC, Pmtzman IM (1990) High reso­lution planktie stable isotope reeorel anel speetralanalysis for the last 5.35 m.y.: Oeean Drilling ProgramSite 625, northeast gulfofMexieo. Paleoceanography5:507-529

Kumar N, Embley RW (1977) Evolution anel origin ofCeara Rise: an aseismie rise in the westem equato­rial Atlantie, Geol Soe Ameliea BuIl88:683-694

Kent DV (1982) Apparent eorrelation ofpaleomagneticintensity anel c1imatic reeorels in eleep-sea seeliments.Nature 229:538-539

Le Treut H, Ghil M (1983) Orbital forcing, c1imatie inter­aetions, anel glaeiation eyc1es. I Geophys Res88:5167-5190

Lisitzin AP (1996) Oeeanie seelimentation: lithologyanclgeochemistry. AGU, Washington D.C.:400 pp

Liu HS (1992) Frequeney variations ofthe Earth 's obliq­uity anel the 1OO-kyr ice-age cyeles. Nature 358:397­399

Liu HS (1995) A new view on the elriving mechanism ofMilankovitch glaciation cyc1es. Eal"!h Planet Sci Lett131: 17-26

Lomb NR (1976) Least-squares frequency analysis ufunequally spaeed elata. Astrophys Spaee Sci 39:447­462

Martinson DG, Menke W, Stoffa P (1982) An inverseapproach to signal eorrelation. I Gcophys Res 87:4807-4818

Martinson DG, Pisias NG, Hays .JD, Imbrie J, Moore TC,Shackleton N.J (1987) Age elating anel the orbitaltheory ofthe iee ages: elevelopment of a high-reso­lution 0 to 300,000-year chronostratigraphy. Quat Res27:1-29

Maslin M, Mikkelsen N (1997) Amazon fan mass-trans­port eleposits anel unelerlying interglacial eleposits:age estimates and fan elynamics. Proc GDP, Sci Re­sults 155:353-365

MatthewsonAP, Shimmielel GB, KroonD (1995) A 300kyr high-resolution arielity reeorel ofthe North Afri­can continent. Paleoceanography 10:677-692

Mclntyre A, Rueleliman WF, Karlin K, Mix AC (1989)Surface water response of the equatorial AtlanticOcean to orbital forcing. Paleoceanography 4: 19-55

Mienel"! J, Chi J (1995) Astronomical time-scale for physi-cal property recorels from QuatemalY seeliments ofthenorthem North Atlantic. Geol Rundschau 84:67-88

Milankoviteh M (1941) Kanon eier Erelbestrahlung unelseine Anwenelung auf elas Eiszeitenproblem. RoySerb Acael SpecPubl 133:1-633

Milliman JD, Slmmlerhayes CP, BalTetto HT (1975) Qua­ternary seelimentation on the Amazon continentalmargi.l1: a moelel, Geol Soc America Bu1l86:61 0-614

Moros M, EnellerR, LackschewitzKS, Wallrabe-AelamsHJ, Micnel"! J., Lemke W (1997) Physical propertiesof Reykjanes Rielge seeliments anel their linkage tohigh-resolution Greenlanel Ice Sheet Project 2 icecare elata. Paleoceanography 12:687-695

Muelelsee M, Schulz M (1997) Tbe Miel-Pleistocene cli­mate transition: onset of 100 ka cyc1e lags ice vol­ume builel-up by 280 ka. Earth Planet Sci Lett151:117-123

Muelelsee M, Stattegger, K (1994). Plio-lPleistocene cli­mate mOelelling baseel on oxygen isotope time seriesfrom eleap-sea seeliment cores: The Grassberger­Procaccia algorithm anel chaotic c1imatic systems.Math GeoI26:799-815

42 Chapter2

Mudelsee M, Stattegger, K (1997) Exploring the struc­ture ofthe mid-Pleistocene revolution with advanced111ethods of time-series analysis. Geol Rundschau86:499-511

Mulitza S (1994) Spätquartäre Variationen der ober­flächennahen Hydrographie im westlichen äquato­rialen Atlantik. Berichte, Fachbereich Geowissen­schaften, Universität Bremen 57, pp 1-97

Muller RA, MacDonald GJ (1995) Glacial cycles andorbital inclination. Nature 377: 107-108

Muller RA, MacDonald GJ (1997) Simultaneous pres­ence oforbital inclination and eccentricity in proxyclimate records from Ocean Drilling Program site 806.Geology 25:3-6

Muller-Karger F, McClain CR, Richardson PL (1988) '111edispersal ofthe Amazon's water. Nature 333:56-59

Mwenifumbo CJ, Blangy JP (1991) Short-term spectralanalysis of downhole logging measurements fromSite 704. Proc ODP, Sci Results 114:577-585

Nikias CL, Raghuveer MR (1987) Bispec1TUll1 estimation:a digital signal processing framework. Proc. IEEE75:869-891

Nobes DC, Blo0111er SF, Mienert J, Westall F (1991)Milankovitch cycles and non linear response in theQuaternary record in the Atlantic sec tor ofthe south­ern oceans. Proc ODP, Sci Results 114:551-576

Nowaczyk NR, Frederichs T, Eisenhauer A, Gard G (1994)Magnetostratigraphic data from late Quaternarysediments from the Yermak Plateau, Arctic Ocean:evidence for four geomagnetic polarity events withinthe last 170 ka ofthe Brnnhes Chron. Geophys J Int117:453-471

Oppenheim AV, Schafer RW (1989) DisCl'ete time signalprocessing. Englewood Cliffs, NJ, Prentice Hall, pp311-312

Park J, D'Hondt SL, King JW, Gibson C (1993) LateCretaceous precessional cycles in double time: awarm-Earth Milankovitch response. Science261:1431-1434

Park J, Maasch KA (1993) Plio-Pleistocene time evolu­tion of the 100-kyr cycle in marine paleoclimaterecords. J Geophys Res 98:447-461

Paillard D (1995) The hierarchical struchue of glacialclimatic oscillations: interactions between ice-sheetdynamics and climate. Clim Dyn 11: 162-177

Peng TH, Broecker WS, Kipphut G, Shackleton N (1977)Benthic mixing in deep sea cores as determined by14C dating and its implications regarding climatestratigraphy and the fate of fossil fuel CO}.In:Andersen NR, Malahoff A (eds) The fate of fossilfuel CO

2in the oceans. Plenum Press, New York, pp

355-373

Pestiaux P, Berger A (1984) An optimal approach to thespectral characteristics ofdeep-sea c1imatic recordsIn: Berger A, Imbrie J, Hays J, Kukla Cl, Saltzman 13(eds) Milankovitch and Climate Part], Reidel..Dordrecht, pp 4] 7-445

Pestiaux PI, van der Mersch I, Berger A, Duplessy Je(1988) Paleoclimatic variability at frequencies rang­ing from 1 cycle per 10,000 years to 1 cycle per 1000years: evidence for nonlinear behaviour of the cli­mate system. Clim Change 12:9-37

Pisias NG, Moore TC (1981) The evolution ofPleistoceneclimate: a time series approach. Emih Planet Sci Lett52:450-458

Pisias NG, Mix AC, Zahn R (1990) Nonlinear response inthe global climate system: Evidence üom benthicoxygen isotopic record in core RC 13··11 O.Paleoceanography 5:147-160

Prell WL (1982) Oxygen and carbon isotope stratigraphyfor the Quaternary ofHole 50213: evidence for twonlOdes ofisotopic variability. Initial Rep DSDP 68:455­464

Radhakrishnamurty C, Likhite SD, Amin BS, SomayajuluBLK (1968) Magnetic susceptibility stratigraphy inocean sediment cores, Earth Planet Sci Lett 4:464­468

Rao TS, Gabr M (1984) An introduction to bispectralanalysis and bilinear time-series models. Springer,NewYork

Raymo ME (1997) The timing of major climate transi­tions. Paleoceanography 12:577-585

Raymo ME, Ruddiman WF, Froehlich PN (1988) Influ­ence of late Cenozoic mountain building on oceangeochemical cycles. Geology 16:649-653

Raymo ME, Ruddiman WF, Backman J, ClementBM,Martinson DG (1989) Late Pliocene variation innorthern hemisphere ice sheets and North Atlanticdeep water cireulation. Paleoceanography 4:413­446

Raymo ME, Ruddiman WF (1992) Tectonic forcing oflate Cenozoic climate. NahIre 359: 117-122

Robinson SG (1986) The late Pleistocene paleomagneticrecord of North Atlantic deep-sea sediments re­vealed by mineral-magnetic measurements. PhysEarth Planet Inter 42:22-47

Robinson SG (1990) Applications for whole-core mag­netic susceptibility measurements ofdeep-sea sedi­ments. Proc ODP, Sci Results 115:737-771

Robinson SG, Maslin MA, McCave IN (1995) Magneticsusceptibility variations in Upper Pleistocene deep­sea sediments ofthe NE Atlantic: implications for icerafting and paleocirculation at the last glacial maxi­mum. Paleoceanography 10:221-250

Using rock magnetic proxy records for orbital tuning and extended time seIles analyses 43

Ruddiman WF, Raymo ME, McIntyre A (1986) Matu­yama 41 ,OOO-year cycles: North Atlantic Ocean andnOlihern hemisphere ice sheets. Earth Planet Sci Lett80: 117-129

Ruddiman \VF, Raymo ME, Mmtinson DG, Clement 13M,Backman J (1989) Pleistocene evolution: narthernhemisphere ice sheets and North Atlantie acean.Paleoeeanography 4: 353-412

Rühlemann C, Frank M, HaIe W, Mangini A, Mulitza S,Müller PJ, Wefer G (1996) Late Quaternary produe­tivity changes in the western equatorial Atlantie:evidence from 2.'oTh-normalized carbonate and or­ganic carbon accumulation. Mal' Geology 135: 127­152

Saltzman13, Verbitsky MY (1993) Multiple instabilitiesand nlOdes of glacial rhythl11icity in the Plio-Pleisto­cene: a general theory of late Cenozoie climatiechange. Clil11 Dyn 9: 1-15

Sarnthein MG, Tetzlaff G, Koopmann 13, Wolter K,Pflaumann U (1981) Glacial and interglacial wind re­gimes over the eastern subtropical Atlantic andnOlih-west Africa. Nature 293: 193-196

Scargle JD (1982) Studies in astronol11ieal time seriesanalysis. II: statistical aspects of spectral analysisofunevenly spaced data. Astrophys J 263:835-853

Scargle JD (1989) Studies in astronomical time seriesanalysis. III: Fourier transforl11s, autocorrelationfunctions, and cross-correlation functions of un­evenly spaced data. Astrophys J 343:847-887

Schiffelbein P, D0I111an LR (1986) Spectral effeets oftime­depth nonlinearities in deep sea sediment records: ademodulation technique for realigning time anddepth scales. J Geophys Res 91:3821-3835

Schulz M, Stattegger K (1997) SPECTRUM: spectralanalysis ofunevenly spaced paleoclimatic time se­ries. Camp Geosci 23:929-945

Schwarzacher W (1993) Cyclostratigraphy and theMilankovitch theory. Elsevier, Amsterdam, 225 pp

Shackleton NJ, Opdyke ND (1976) Oxygen-isotope andpaleomagnetic stratigraphy ofpacific eore V28-239late Pliocene to latest Pleistocene. In: Cline RM,Hays JD (eds) Investigations of Later Quaternarypaleoceanography and paleoclimatology. Mem 145Geol Soc Amer, Washington, pp 449-464

Shackleton NJ, Berger A, Peltier WR (1990) An altema­tive astronomical calibration ofthe lower Pleistocenetimescale based on ODP Site 677. Trans Royal SocEdinburgh: Ealih Sci 81 :251-261

Shackleton NJ, Hagelberg TK, Crowhurst SJ (1995)Evaluating the success of astronomical tuning: pit-

falls ofusing coherence as a criterion far assessingpre-Pleistocene timeseales. Paleoceanography10:693-697

Shackleton NJ, Crowhurst S (1997) Sediment fluxesbased on an orbital tuned time scale 5 Ma to 14 Ma,Site 926. Proc ODP, Sci Results 154:69-82

ShartDA, Mengel JG, CrowleyTJ, Hyde WT, North GR(1991) Filtering ofMilankoviteh cycles by Earth 'sgeography. Quat Res 35: 157-173

Tarduno JA (1994) Temporal trends ofmagnetic disso··lution in the pelagic realm: gauging paleoproduc­tivity? Earth Planet Sci Lett J23:39-48

Tarduno JA, Wi!kison SL (1996) Non-steady state mag­netic mineral reduction, chemicalloek-in, and delayedremanence acquisition in pelagic sediments. EarthPlanet Sei Lett 144:315-326

Thiessen W (1993) Magnetische Eigenschaften desöstlichen Südatlantiks und ihre paläozeanographi­sche Relevanz. PhD thesis, Bel' FB Geo Univ Bremen,41,pp 1-170

Thompson R, Clarke RM (1993) Quantitative marine sedi­ment core matching using a modified sequence-slot­ting algorithm. In: Hailwood EA, Kidd RB (eds) Highresolution stratigraphy. Spec Publ Geol Soc London70, pp 39-49

Thomson, DJ (1982) Spectmm estimation and harmonicanalysis. hoc IEEE 70: 1055-1 096

von Dobeneck T (J 996) A systematic analysis ofnatu­ral mineral assemblages based on modelling hyster­esis loops with coercivity-related hyperbolie basisfunctions. Geophys .I Int 124:675-694

von Dobeneck T, Freelerichs Th, Fabian K, Blei! U (1996)Rock Magnetic Micro-Scanning mit HTSL-SQUID­Graeliometer, VDI-Technologiezentrum (eels) Supra­leitung und Tieftemperaturtechnik 1996. VDI-VerlagDüsselelarf(Germany), pp 51-54

Weeks R, Laj C, Endignoux L, Fuller M, Roberts A,Mangamle R, Blancharel E, Goree W (1993) Improve­ments in long-core measurement techniques: appli­cations in paleomagnetism anel paleoceanography.Geophys .I Int 114:651-662

Welch PD (1967) The use offast Fourier transfOllli forthe estimation ofpower spectra: a method based ontime averaging over short, modified perioelograms.IEEE Trans Audio, Electroacustics, AU-15 :70-73

Yiou P, Genthon C, Ghi! M, .Iouzel J, Le Treut H, Bamola.IM, Lorius C, Korotkevitch YN (1991) High-fre­quency paleovariability in climate anel C02 levelsfrom Vostock ice core recoreIs. J Geophys Res96:20365-20378

44

Mid-Pleistocene climate transition:initiation, interim state andterminal event (submittecl to Nature)

F. Schmieder, T. von Dobeneck & U. Blei!

Fachbereich Geowissenschaften, Universität Bremen,

P.O. Box 330 440, 28334 Bremen, Germany

The mid-Pleistocene transition (MPT) of the globalcIimate system l -", initiated by a shift towards muchlarger northern hemisphere ice shields at around920 ka5 ami ending with predominance of 100 kyr iceage cyclicity since about 640 kaG-8, is one of the funda­mental enigmas in Quaternary cIimate evolution9• Ofthe diverse explanations proposed for the 100 kyr cycIe,models invoking a pCOrcontrolled insolation thresholdfor the meIting of large ice shields lO- 14 currently yieldthe best reproduction of 8 180 signal pattern, aIthoughinconsistencies remain between oxygen isotope stages 15ami 13. Climate proxy records not excIusively linked toglobal ice volume are necessary to advance understan­ding of the MPTI5. Here we present high-resolutionPleistocene magnetic susceptibility time series of twelvesediment cores from the subtropical South Atlanticessentially ret1ecting variations in carbonate accumu­lation. In addition to characteristics known from 8180records, they reveal three remarkable features intima­tely related to the MPT chronology, (1) aprematureoccurrence of a near-100 kyr cycle around 1150 ka, (2) aMPT interim state of reduced carbonate depositionbound by transitions from 1000 to 920 ka and 640 to500 ka, and (3) a terminal MPT event at around 530 kadocumented in various unusual lithologies.

The twehre pelagic gravity cores recovered with R.V.METEOR in the framework of a long-term paleoceano­graphic research program (SFB 261 at the University ofBremen) form an oceanwide transeet spanning the subtro­pical South Atlantie between 22°S and 34°S (Fig. 1).Located on the Danks of major submarine elevations (RioGrande Rise, Mid-Atlantic Ridge, Walvis Ridge) in theelepth zone alternately inDuenceel by North Atlantie DeepWater (NADW) and Circumpolar Deep Water (CDW), theseeliments provide detailed records of carbonate elissolutioneyeles related to changes in deep water ehemistry. Oligo­trophic open ocean eonditions result in low average sedi­mentation rates of 0.5 to I cm/kyr at these latitudes. In suchenvironments, a elose inverse eorrelation of magnetic sus­eeptibility to carbonate content is typieally observedlG. Thedata were acquireel on split core halves using a Bartingtonspot sensor at a 1 cm spacing allowing a signal definitionand temporal resolution commonly not attained in 8180 or%CaC03 analyses.

The recoreIs were founel particularly suitable to refineinitial magnetostratigraphic age models by means of astro­nomical tuning l7 to orbital variations ealculateel by Bergerand Loutre l8 . A phase lag applied (4.5 kyr to obliquity) was

Figure 1 Loeations 01' Geoß eores defining the SUSAS transect.The depth profile follows the dashed line in the Illap and is plottedagainst longitllde to depict affinities to the major basins 01' theSOllth Atlantie anel theil' deep water boelies. It is not sllitable toielentify positions anci depths 01' deep water passages.

determined by cross spectral analysis of 8180 and magneticsusceptibility of one of the cores (GeoB 1211-1, 8180 datafrom Bickert & Wefer I9). The perfect overall eonelation ofall twelve inelivielually tuneel records (Fig. 2) leel us toelefine the SUSAS (subtropical §outh 6tlantic ~uscepti­

bility) stack.A prominent feature of this stack and all inelividual rec­

orels is a baseline shift to 40% lügher average suseeptibili­lies during the MPT interim state (920 to 640 ka). Confinedby transitional intervals lasting about 80 anel 140 kyr at itsonset and termination, this period exhibits the most distinctdissimilarities between individual core logs. The precedingand following elimate states show similar mean suscepti­bilities modulateel by elearly eleveloped ancI coherenteyelieities of 41 and 100 kyr, respeetively. An exception is a'premature' near 100 kyr cyele around 1150 ka whieh wasalso identified in 8180 reeords and has been interpreted byMudelsee and Stattegger7 as an 'unsueeessful attempt of theelimate system to attain a non-linear Late Pleistocene ieeage state'. The MPT suseeptibility shift is notably mirroredin lowered average sedimentation rates (Fig. 2).

Evolutionary speetral analysis of the SUSAS stackexhibits a continuous Pleistoeene eleeline of 41 kyr eyeli­city, but a eomparatively abrupt intensification of 100 kyrcyelieity after about 640 ka 17. Oxygen isotope reeordsdoeument larger iee shields sinee the beginning of the MPTinterim state at 920 ka. Why did 100 kyr eyelieity not startuntil 640 ka and henee lag the initial iee volume inerease byapproximately 280 kyr8? The observed baseline shift in theSUSAS reeords preeisely fills this time lag and should shedlight on this question.

The cause for the overall rise in magnetie susceptibilityeluring the MPT interim state is revealeel by a temporalanalysis of west-east trends. Plotted against longituele, eoremean suseeptibilities for the postulateel three elimate states(Fig. 3 top) reDect the well-known west-east asymmetry of

Mid-Pleistocene climate transition: initiation, interim state and terminal event 45

I41 kyr StateTran·sition

900

100 kyr State

100o 1000 1100 1200 1300 1400 1500 1600I, ,li !.I, i!."I!! ,I ! • .LL..u..Ll..u..u.I'!I,I,!! 1,1 ,r·! ,I,!,.l" !:"'I.J_U..L..LLd~~

150 IEarly Pleisloce~e 100 kyr Cycle I I I100 ':UJ I

50 GeOB1211-311o "--'__ +_,---c-, -II-_+-__~ ----- WO !.084 m

~~~~d"~A;tl GeoB 1034-J~II'0- 0,90 : :0,63 __--'W",O 3772 m~ ~ I

~~lMJ' ... .\mAAi T '! ~~:;~o Shift i ~:::i'~::~::::O~~:;O GooB 10354 1~ I01,18 ~ " WO 44_53 m

~!!t~J~~~ C;;~B4;~;~3 "

(J)~ ~~ooo~ ~ C;;~~;:~~6 ~L.:..C'-=-------- i-- --+-- , +-~'_--'__ __=.:=__=__=_=_'_'_'___I Q2

~ 400t_Ac-' AI A_ i iL,.r~"" i ~c 200k =VVV'''V-=VV'Ur : T Hi~ GeoB 3812-1 ~

1;::jO:~~1~~~G~::::::n3~~ J~.'~ : :0,57 ::0,68 WO 4331 m'~ 750-" ::: : ~C .O'l 500· UJ~ 250 eoB 814-6

o WO 4340 m

250

100

(I)

O--'-'-'c.:....:...-------------t --F-'---------,!\----;-:-:.:..:..... -r;--- -,;-----;------'-'-.=.c"-"-""--IF-'-I cE~. ~

c:, GeoB 1311-1 ~

o 0,39 ':0,34 ~iO,~ WO 2901 m ~

::~t~"~.. , '" ~,~~,,~ "~ ",i<"G {\" 0:: I ; \j.::J"IJc\)\J·V~OB 2820-2o 0.49 ; 10,37 ',0,74 WO 3615 m

:1 ~ ", : 18 ::14

:16" : 20 22: : 30 3,6 4~ 4,8l ... . " :'IJw-2 ,:" " SUSAS Stack

t 0 :J~~I~~=,~~~~:, ",!i~"",""" "',i"",i,~!;,;~«~:~;~i~;i~~!;~,~:;"'~~:,"~o 100 200 300 400 500 600 700 800 900 1000 1100 1200 1300 1400 1500 1600

Age [ka]

500

Figure 2 Individually tuned SUSAS records (WO = Water Oepth) and resulting susceptibility (arithmetic mean with standard deviationband) and sedimentation rate stack. A diamagnetic susceptibility of -15,10.6 SI units reprcscnting pure carbonate sediment was chosen as axisminimum To compensate for different signal levels in the stacking, each record was normalised by subtracting its mean and dividing by itsstandard deviation, Labels at the SUSAS stack indicatc even oxygen isotope stages, Horizontal dotted lines mark baseline averages for pre-,syn- and post-MPT states exclllding the bounding transitions, Ouring the interim statc, all cores display increascd susceptibilities and reducedsedimentation rates, Thc timing of the three intervals approximatcly corresponds to thc Laplacc (1800-1200 ka), Croll (1200-600 ka) andMilankovitch chron (600-0 ka) dcfincd by BergeT and Wefer5 Numbers in each section denote mean sedimcntation rates; grey vertical bars apremature 100 kyr cycle centred near 1150 ka and a terminal MPT evcnt fcatllring unusual1ithologies at about 530 ka,

46 Chapter 3

[!1 kyr State (> 1000 ka) ]

40'W 30'W 20'W 10'W 0' 10'E

2820-22.0 ,

2821-1

1.0

0.110'E 40'W 30'W 20'W 10'W 0'

1.21.11.00.90.8 2820-2

0.74&

1034-3• 0.6 1309-22821-1 ••0.5 •Ü 3812-1

1311-1

• 04 4&1211-3 3814-6

0.310'E 40'W 30'W 20'W 10'W 0' 10'E

051211;3

4&0.3

1034-3

• 0.2

•1034-3Enhanced SusceptibilityConstant W·E Gradient

Constant MagnetiteAccumulationConstant W·E Gradient

3814-62821-1 • 4&

4& • 3813-31309-2

•2820-2 4&3812-1

50

302010

0.6 381~ 3~33

0.5 2~21-"3~_2

DA 3812-1• •2820-2 Ü1311-1

300

200

100

0-'--,----,---,--,--,---,1-

10'E 40'W 30'W 20'W 10'W 0'

0"

3813-3

3'12-1•

•3814-6

[ 100 kyr State (500·0 k!J] [Interim State (920· 64~

30'W 20'W 10'W 0' 10'E 40'W 30'W 20'W 10'W 0' 10'E-,-'----'----'-----'--'-----"'-ri800-,-'--".,---"-L-".,---="-".,--".,-'--,-,-".,-'-----,,.,-"'-riJOO-r----'------'-----''----"-----'--,800

500

40"W 30"W 20"W 10"W

10'E 40'W 30'W 20'W 10'W 0'2 T"------'------'------'r,--------"---r'T-'-r 1.2 ,...L--'-------'--'---'---J,

1.1 Reduced Sedimentation10 Collapsed W.E Gradient0.9

0.8

07

•1.0 2820-2

0.5

0.3

0.1 -'--,---,---,--,-------,--,-'-040'W 30'W 20'W 10'W

(J)

co~::::J

E::::Joo«2 0.2~cOlco

::2:

~.Y. 3.0 3814-6

E 2.0 2821-1 : 38~-3OJ • 1309-2

O. 3J,-----,----,----,----,---,-'-0.3-'-,--,---,----,--,--,-1-

10"E 40"W 30"W 20"W 10"W 0"Longitude

Figure 3 Combined spatial and temporal trend analyses of magnetic susceptibility, magnetite aeeumulation (ealculated by splittingsuseeptibility into a diamagnetie background and ferrimagnetic, grain-size independent magnetite signal) and sedimentation rate. MajoraSy111111etries ancl shifts are summarised by white regression lines, dark-grey shading delimits mean data range, light-grey shadingwestern and castern South Atlantic eore sets. Three eores were exeluded from the analysis: eores GeoB 1035-4 and 3801-6, as they donot reaeh beyond the 530 ka event, and core GeoB 131 I-I recovered from 290 I m water depth clearly above the CDW/NADWtransition zone and therefore showing a somewhat different evolution.

terrigenous partic1e flux and accumulation in the SouthAtlantic20 . While the gradients are almost identical through­out Pleistocene, a shift to lügher susceptibilities is evidentduring the MPT interim state. This relative increase ofmag­netic mineral concentration cannot be explained in terms oftemporal changes in magnetite accumulation (Fig. 3centre), as its west-east decline from 2.5 to 0.3 g/m2kyrremains fairly constant over time. A time-varying dilutionof the terrigenous fraction by non-magnetic carbonate musttherefore be responsible far the observed shift.

Mean sedimentation rates (Fig. 3 bottom) deduced fromour age models are mainly controlled by carbonate accu­mulation. Even the westernmost core GeoB 2821-1 near theSouth American continent consists to 70 ± 9% of CaC03

(P. Mueller, pers. conun.). Both for the 41 and 100 kyrc1imate states, sedimentation rates vary from about

0.5 cm/kyr in the western to around 0.9 cm/kyr in the east­ern South Atlantic. During MPT interim state sedimentationrates on either side of the mid-Atlantic Ridge are restrictedto between 0.4 and 0.6 cm/kyr. The dec1ine primarilyaffects the eastern part and brings the west-east asynunetryto collapse.

In the working area cyc1ic variations of the sedimentCaC03 content at Milankovitch frequencies are mainly dueto orbitally driven lysocline shifts resulting from an inter­play of NADW and more corrosive CDW19. In the AngolaBasin (cores GeoB 1034-3, 1035-4 and 1729-3) the glacial­interglacial contrast is particularly manifest as the sur­rounding bathymetric highs restrict the access of CDWduring interglacials. The same reasoning should apply for10ng-term changes in carbonate accumu1ation. We thereforeassume that the influence of southern-source deep water

Mid-Pleistocene climate transition: initiation, interim state and terminal event 47

eurrents and intense produetivity blooms. Results from thcJoint Global Oeean Flux Study (JGOFS), based on photo­graphs and measurcments from satellites, aireraft, ships, andthe Space Shuttle ATLANTIS, emphasise the potentiallyimportant role of oceanie frontal zones for the rapidaecumulation of diatom biomass28 They indicate dramaticbiological responses to eirculation and mixing processesassoeiated with open-oeean frontal systems separating coldfrom warm waters and gave rise to the idea that thiek dia­tom layers may be deposited in such environments.

The diversified terminal MPT events at around 530 kadoeumented in South Atlantic Oeean sediments might weHbe related to other paleoelimatie 'puzzles' like unusuallyhigh and low 8 180 values during isotope stages 13.2 and13.3, respeetively, found in the equatorial Indian Ocean29 ,an anomalous sapropellayer in the Mediterranean Sea dated528-525 ka30 and highest l11agnetic susceptibilitics of thepast 2500 kyr at about 500 ka in Chinese loess deposits31 ,il11plying an extreme1y warm and humid clil11ate.

The South Atlantic plays a key role for the global ther­1110haline circulation. We have reasons, therefore, to assumethat the MPT phenomena documented in the SUSAS coresrespond to major global paleoceanographic shifts andsubstantially extend constraints on mid-Pleistocene e1imatereconstructions.

Acknowledgements. We are grateful to W.H. Berger forhelpful suggestions. This study was funded by the DeutscheForsehungsgemeinsehaft (Sonderforschungsbereich 261 atBremen University, contribution No. XXX). F.S. was sup­ported by the Deutsche Forschungsgemeinschaft in theti'amework of Graduiertenkolleg 221.

I. Pisias, N.G. & Moore, T.C. The evolution of Pleisto­cene climate: a time series approach. Earth Planet. Sei.Lett. 52,450-458 (1981).

2. Prell, W.L. Oxygen and earbon isotope stratigraphy forthe Quaternary 01' hole 502B: evidence for two modes ofisotopic variability. [nil. Rep. DSDP 68, 455-464 (1982).

3. Maaseh, K.A. Statistical detection of the mid-P1eisto­cene transition. Clim. Dyn. 2, 133-143 (1988).

4. Ruddiman, W.F., Raymo, M.E., Martinson, D.G.,C1ement, B.M. & Backman, J. Pleistocene evolution:northern hemisphere ice sheets and North At1anticocean. Paleoceanography 4,353-412 (1989).

5. Berger, W.H. & Wefer, G. Klimageschichte aus Tief­seesedimenten. Naturwissenschaften 79,541-550 CI 992).

6. Berger, W.H., Yasuda, M.K., Bickert, T., Wefer & G.,Takayal11a, T. Quaternary time scale for the Ontong JavaPlateau: Milankovitch template for Ocean DrillingProgram Site 806. Geology 22, 463-467 (1994).

7. Mude1see, M. & Stattegger, K. Exploring the struetureof the mid-Pleistocene revolution with advanced meth­ods of time-series analysis. Geol. Rundschau 86, 499­511 (1997).

8. Mude1see, M. & Schulz, M. The mid-Pleistocene cli­mate transition: onset of 100 ka eycle lags ice vo1umebuild-up by 280 ka. Earth Planet. Sei. Lett. 151, 117­123 (1997).

544 546 548 550 552 554 556528

306 307 308 309 310 311 312 313 314 315 316

Core Position [em]

Figure 4 Terminal MPT event documented in a sharp colourtransition at unbioturbated finely laminated layers in core GeoE3812-1.

conelusion is supported by a simultaneous decrease ofglacial and interglacial kaolinite/chlorite ratios in coreGeoB 2821-1"1, interpreted to be induced by ehlorite­enriched deep southern waters. Consequently, NADWshould have been reduced during that interval. Indeed, Öl3Crecords from Atlantic and Pacific ODP sites indicatesignificantly weaker NADW between 900 and 400 kaiS, Inview of this global evidence we suggest, that the MPTshould not be regarded as a gradual transition from a'41 kyr world' to a '100 kyr world', but rather as a third,contrasting climate state.

A conspicuous finding in several of our cores is the coin­cident occurrence of unusual sediment facies at around530 ka e10se to the end of the terminal MPT transition.Granulometric analyses of cores GeoB 2820-2 ancl 2821-]show a sharp grain-size shift with an almost total loss of the> 63 pm fraction 22, possibly related to the mid-Brunhesdissolution cycle23 . In core GeoB 2820-2 this episode cor­responds to a short hiatus. The continuous Late Pleistocenerecord of core GeoB 1035-4 is interrupted by a thick tur­bidite dating to the same age. Core GeoB 3812-1 displays asharp colour change at a de1icately laminated horizon(Fig.4). Most impressive are thick interca1ated laminateddiatom ooze layers in cores GeoB 3801-6 (124 cm) and3813-3 (38 cm). Omitting these intervals in the susceptibil­ity time series results in continuous Milankovitch signalsequences. Hence, the laminations must have been depos­ited in a very short time and should ret1ect an extremelyshort-term climate variability. Very high sedimentation ratesin the overlying carbonate sediments may have contributedto preserve the siliceous sections in both cores. The almostmonospeeifie layers eonsist of the giant diatomEthmodiscus rex (Rattray) Wiseman & Hendey (C.B.Lange, pers. cOlmn.) and are entirely uncommon at theselatitudes. Thick ooze deposits of this diatom were mainly

. I . f I ld 24-26reported from equatona regIOns 0 t 1e wor oceans .Even there, their occurrenee is puzzling, because Ethmodis­cus rex very rarely oeeurs in plankton sampIes. However,enigmatie deep populations have been observed in thePaeifie Oeean27 . Several hypothesis have been proposed toclarify the paradoxiea1 'Etlunodiscus rex prob1em'26,ineluding differential dissolution, foeusing by bottom

48 Chapter 3

9. Imbrie et al. On the structure and origin of major gla­ciation cycles: 2. The 100,000 year cycle. Paleoceano­graphy 8, 699-735 (1993).

10. Saltzman, B. & Verbitsky, M.Y. Multiple instabilitiesand modes of glacial rhythmicity in the Plio-Pleistocene:a general theory of late Cenozoic climatic change. Clim.Dyn. 9,1-15 (1993).

11. Berger, W.B. & .Iansen, E. Mid-Pleistocene climate shift- the Nansen connection. In The Polar Oceans and TheirRole in Shaping the Global Environment (eds..Iohannessen, M., Muench, R.D. & Overland, 1.E.) 295­311 (Geophys. Monogr. 85, AGU, Washington, 1994)

12. Raymo, M.E. The timing of major climate transitions.Paleoceanography 12, 577-585 (1997).

13. Paillard, D. The timing of Pleistocene glaciations from asimple multiple-state c1imate model. Nature 391, 378­381 (1998).

14. Raymo, M.B, Glacial puzzles. Science 281, 1467-1468(1998).

15. Raymo, ME, Oppo, D.W. & Curry, W. The mid­Pleistocene climate transition: a deep sea carbon isoto­pic perspective. Paleoceanography 12, 546-559 (1997).

16. Robinson, S.G. Applications for whole-core magneticsusceptibility measurements oi' deep-sea sediments: Leg115 results. Proc. ODP, Sei. Res. 115,737-771 (1990).

17. von Dobeneck, T. & Schmieder, F Using rock magneticproxy records for orbital tuning and extended time seriesanalyses into the super- and sub-Milankovitch bands. InUse of proxies in Paleoceanography: Examples from theSouth Atlantic (eels. Fischer, G. & Wefer, G.), 601-633(Springer Verlag, Berlin, 1999).

18. Berger, A & Loutre, M.F Insolation values for theclimate oi' the last 10 million years. Quat. Sci. Rev. 10,297-317 (1991).

19. Bickert, T. & Wefer, G. Late Quaternary deep watercirculation in the South Atlantic: reconstructions fromcarbonate dissolution and benthis stable isotopes. In TheSouth Atlantic: present emd past circulation (eds.Wefer, G., Berger, W.B., Siedler G. & Webb, DJ.),599-620 (Springer Verlag, Berlin, 1996).

20. Lisitzin, AP. Oceanic sedimentation: lithology andgeochemistry. AGU, Washington D.e., 400 pp (1996).

21. GingeIe, FX., Schmieder, F, von Dobeneck, T.,Petschick, R. & Rühlemann, e. Terrigenous t1ux in theRio Grande Rise area during the last 1500 ka: evidenceof deepwater advection 01' rapid response to continentalrainfall patterns? Paleoceanography, 14,84-95.

22. Breitzke, M. Elastische Wellenausbreitung in marinenSedimenten - neue Entwicklungen der UltraschallSedimentphysik unel Seelimentechographie. Berichte,Fachbereich Gemvissenschaften, Univ. Bremen, 104,298 pp (1997).

23. Adelseck, e.G. .Ir. Recent and late Pleistocene sedi­ments from the eastern equatorial Pacific Ocean: Sedi­mentation and dissolution. Ph.D. thesis, 192 pp, Univ. ofCalif., San Diego (1977).

24. Mikkelsen, N. On the origin oi' Etlunodiscus ooze. Mal'.Micropaleontol. 2, 35-46 (1977).

25. Stabeil, B. Variations of diatom flux in the easternequatorial Atlantic during the last 400,000 ycars("METEOR" cores 13519 andI3521). Marine (;eol. 72,305-323 (1986).

26. Gm'dner, 1.y. & Burckle, L.B. Upper Pleistocene Eth­modiscus rex oozes from the eastern eqllatorial Atlantic.Micropaleontol. 21,236-242 (1975).

27. Villareal, T.A Abllndance of the giant e1iatom Ethrno·discus in the Southern Atlantic Ocean and the ccntralPacific gyre. DiatOlIJ Res. 8, 171-177 (1993) .

28. Yoelcr , .I.A., Ackleson, S.G., Barber, R.T., Flament. P.& Baieh, W.M. A line in the sea. Nature 371, 689-692(1994).

29. Bassinot, Fe., Labeyrie, L.D., Vincent, E" Quidelleur,x., Shackleton, NJ. & Lancelot, Y. The astronomicaltheory of c1imate anel the age of the Brunhes-Matuyamamagnetic reversal. Earth Planet. Sci. Lett. 126, 91- J08(1994).

30. Rossignol-Strick, M., Paterne, M., Bassinot, FC.,Emeis, K.-e. & DeLange, G..I. An llnusual miel-Pleisto­cene monsoon period over Africa and Asia. Nature 392,269-272 (1998).

31. Bloemendal, 1., Liu, X.M. & Rolph, T.e. Correlation ofthe magnetic sllsceptibility stratigraphy of Chinese loessand the marine oxygen isotope record: chronologicaland palaeoclimatic implications. Earth Planet. Sci. LeU.131, 37J-380 (1995).

Cycles, trends and events ofPleistocene sedimentation in the oligotrophiesubtropical South Atlantie Ocean(to be subrnitted to Paleoceanography)

Frank Schmieder, Tilo von Dobeneck and Ulrich Bleil

Fachbereich Geowissenschaften, Universität Bremen, 28334 Bremen, (Jcrmany

Abstract. Most investigations of the mid-Pleistocene climate transition (MPT) were based on thc iee

volume proxy 8180. This study develops new aspects of this enigmatic period of Quaternary climate

evolution from twelve subtropical South Atlantic susceptibility records reflecting carbonate dissolution

cycles and hence changes in deep water chemistry. Comparative evolutionary spectra ofthese rock magnetic

proxy records and orbital forcing imply reciprocal exchange ofspectral energy between the 41 and 100 kyr

bands. A MPT interim state ofreduced carbonate deposition is defined from a temporal analysis ofwest­

east-asymmetry. Presumably resulting from enhanced influence of southern source waters, this interval

exactly fills the time lag between the first occurrence oflarger glacial ice shields (~920 ka) and the onset

of 100 kyr ice age cyclicity (~650 ka). It ends with an unusual, probably global scale paleoclimatie episode

at about 530 ka, which is documented in several uncommon lithologies and interpreted as terminal MPT

event.

Introduction

The South Atlantic Ocean plays a major role in the

framewark of global thermohaline circulation.

'North Atlantic heat piracy' (Berger and Wefer,

1996), the interglacial exchange of cold North

Atlantic Deep Water (NADW) for warm surface

waters from the south, is essential far today' s warm

climate in northern Europe. Switching on and off

the Atlantic Heat Conveyor in the global ice age

rhythm also modifies conditions in the deep ocean.

Glacial reduction ofNADW gives way for a thicker

layer ofLower Circumpolar Deep Water (LCDW)

in the Atlantic Ocean and lifts the NADW/LCDW

boundary. Beside contrasting in temperature,

salinity and nutrient concentration these two deep

water masses differ in corrosiveness with respect

to calcium carbonate and hence in the preservation

of calcareous shells. The fluctuating NADWI

LCDW lysocline leaves carbonate dissolution

cycle's in transition zone sediments as documen­

tation of paleoclimatic history.

The three major submarine ridges dividing the

subtropical South Atlantic into four pe]agic basins

are by far the best locations to find recordings of

this history as they intersect the transition zone

during glacia] as well as interglacial times. Penna­

nent oligotrophie conditions in this 'open ocean

desCl't' keep the sedimentation rates as ]ow as 0.5­

1 cm/kyr. On one hand, this makes it rather difficult

to build precise age models. On the other hand, gra­

vity coring ofthe top 10m ofthe sediment column

yields continuous, nearly undisturbed records ofthe

whole Quaternary. These sequences offer the oppor­

tunity to study not only cyclic changes but also long­

tellli trends of Pleistocene climate evolution,

Still enigmatic within this epoch is the mid­

Pleistocene transition (MPT) of the global climate

system (e.g., Pisias and Moore, 1981; Prell, 1982;

Ruddiman et a1. , 1989). In the course ofthe MPT,

the response to orbitally driven changes in insolation

received on Earth changed fundamentally. While

late Pliocene to early Pleistocene paleoclimate re-

50 Chapter 4

30 0 S

45°S '-'--__-'--_--'-__-'----_-----'__----'-__-'-----_-----L__---'---_-----'__---'---__-'-----_----'-_

50 0 W

[m]

1,000

o-1,000

-2,000

-3,000

-4,000

-5,000

-6,000

Figure 1. GeoB core locations defining the SUSAS transecL The depth profile follows the white dashed line in the map amiis plotted against longitude to depict the cores' affinities to the four major pelagic basins and their deep water bodies. It is notsuitable to identify the positions and depths of deep water passages.

cords exhibit l11ainly obliquity and precession rela­

ted variance (e.g., RaYl110 et al., 1989; Ruddiman

et al., 1989; Bloel11endal and de Menocal, 1989),

which can be explained as linear responses (e.g.,

Imbrie et al. , 1992), the primm'y rhythm oflate Pleis­

tocene climatic change with aperiod near 100 kyr

calls for more difficult explanations (for a summary

of research and proposed models see Imbrie et al.,

1993). The exact timing ofthe MPT and the question

whether it was a gradual ('mid-Pleistocene evolu­

tion', Ruddiman et al., 1989) 01' a sudden change

('mid-Pleistocene revolution', Berger et al., 1994)

were subject to many studies, which attempted to

model 01' characterize the transition with different

statistical techniques (e.g., Maasch, 1988; DeBlonde

and Peltier, 1991; Park and Maasch, 1993; Mudelsee

and Schulz, 1997; Mudelsee and Stattegger, 1997;

Clark and Pollard, 1998). With few exceptions (e.g.,

Raymo et al. , 1997) all ofthese investigations were

based on oxygen isotope records. In our view, cli­

mate proxy records not exclusively linked to global

ice volmne are necessary to advance understanding

ofthe MPT. Records ofcarbonate dissolution cycles

in the deep South Atlantic should provide an deci­

sive complementary perspective of this phenome-

non.

In many marine environments, variations in car­

bonate accumulation can be precisely traced by rock

magnetic methods. Here we present twelve high­

resolution magnetic susceptibility time series of

Cycles, trends and events of Pleistocene sedimentation 51

-----,-------~

I Cruise report I

Wefer et a1. (1988)-j

Wefer et a1. (1988)

Wefer et al. (1990)

Pätzold et al. (1993)

Pätzold et al. (1993)

8chulz et al. (1992)

Bleil et al. (1994)

Blcil et al. (1994)

Wefcr et al. (1996)

Wefer ct al. (1996)

'J I VVeI'er et al. (1996)

~~~_~lj199§1_...J

Core METEOR Water Position Core

expedition depth (m) Latitude Longitude length (m)

GeoB 1034-3 M6/6 3772 21 °44.1'8 05°25.3'E 10.65

GeoB 1035-4 M6/6 4453 21°35.2'8 05°01.7'E 10.61

GeoB 1211-3 M1211 4084 24°28.5'8 07°32.0'E 8.61

GeoB 1309-2 M15/2 3963 31 °40.0'8 28°40.0'W 9.48

GeoB 1311-1 M15/2 2901 31 °30.7'8 29°05.9'W 7.42

GeoB 1729-3 M20/2 4401 28°53.6'8 () 1000.I'E 7.45

IGeoB 2820-2 M29/2 3615 30°49.4'8 38c 26.4'W 7.67

GeoB 2821-1 M29/2 3941 30°27.1'8 38°48.9'W 8.19

GeoB 3801-6 M34/3 4546 29°30.7'8 08°18.3'W 9.37 IGeoB 3812-1 M34/3 4205 31 °36.9'8 19°45SW 5.32

IGeoB 3813-3 M34/3 4331 32°16.1'8 21°58.0'W 9.83

GeoB 3814-6 M34/3 4340 34 0 11.0'8 28°38.0'W 7.95----

Table 1. Water depths, positions anel core lengths of the GeoB gravity cores presented in this paper.

Pleistocene sediment cores from the subtropical

South At1antic.

Material and Methods

The cores were chosen from material recovered

during six expeditions with R.Y. METEOR in the

framework of a long-term paleoceanographic

research program (SFB 261 at the University of

Bremen). This core selection covers the time range

of the MPT and fon11s a transect across the entire

deep South Atlantic Ocean between 200 S and 35°S

(Figure 1). Table 1 summarizes positions, water

depths and core 1engths. With the exception ofGeoB

1311-1, which was recovered fi-om 2901m, aIl cores

are located within the broad transition zone between

NADW and LCDW during glacia1 as weIl as inter­

glacial times.

The westenm10st cores GeoB 2820-2 and 2821-1

are located on the east side of the Vema Channe1,

the predominant of two LCDW deep water passa­

ges. Cores GeoB 1309-2, 1311-1 and 3814-6 lay

on the eastern flank of the Rio Grande Rise in the

vicinity of the Hunter Channe1, GeoB 3812-1 and

3813-3 on the western slope of the Mid-Atlantic

Ridge. As the flow of corrosive LCDW into the

Angola basin is to some extent restricted by sur­

rounding ridges, especiaIly during interglacials (e.g.,

Bickert and Wefer, 1996), sedimentation there is

less affected by dissolution and sedimentation rates

are somewhat higher. While GeoB 1034-3 and

1729-3 from the northern and 1211-1 from the

southern flank of the Wa1vis Ridge reach back to

1200 ka, GeoB 1035-4 and 3801-6 do notl'each the

Brunhes/Matuyama boundary. They were selected

for this study as they contribute detailed records of

the last about 500 kyr and additionally document a

paleoceanographic event discussed later.

The pe1agic carbonate sediments mainly consist

ofnannofossi1 ooze with varying amounts offorami­

nifera. Five cores were disturbed by one 01' two,

GeoB 3801-6 by five thin turbidites. Susceptibility

signal correlations showed that simply removing

the turbidites from the sequence results in complete

time series, only in GeoB 3812-1 arecord section

was lost. Siliceous fossils are generally negligible

in the sediments, but two cores contain an enormous

diatom ooze layer. In GeoB 3813-3 this layer is

38 cm thick, in GeoB 3801-6 even 124 cm. As in

the case of the turbidites, removal of the diatom

layers results in complete, correlatable susceptibility

records. Obviously, these 1ayers have been deposi­

ted in relatively short time.

Paleomagnetic ana1yses were performed at

5-10 cm sampling intervals. Natural remanent mag-

52 Chapter 4

200 200

150R2 =0.83

150- . Ci(f) I . ü5 ill"? <0 70 D-0 . ". b 0=::. . •-.1.. ... =. :J

ill

.~ 100 ,:,..- . .-2 100 ur. ():fj '. j5 0

0. ß 80 :J

ill ~Q) 0ü if)(j) :J

~:J 50 (f) 50(f)

90

0 0100

100 90 80 70 60 50 0 200 400 600 800 1000 1200Carbonate content [%] Age [ka]

Figure 2. Comparision of magnetic susceptibility and carbonate content (T.Bickert, 1992) far care GeoB 1211-3. Far tbecross-plot on the left siele susceptibility elata were reduced to the 5 cm sampling intervall of the carbonte data. The inversecorrelation is very close in all period ranges, long-term trends anel Milankovitch cycles are weil reproduced. Due to bigherresolution, high-frequent signal features are expressed in more detail in magnetic susceptibIity.

netization was measured using a three-axis cryo­

genic magnetometer (Cryo genic Consultants

GM 400) and statically AF demagnetized in ten

steps up to 100 mT (2G demagnetizer). The direc­

tion of the characteristic remanent magnetization

(ChRM) was calculated by averaging over at least

three successive demagnetization steps.

Magnetic volume susceptibility K was measured

on the archive halves ofthe cores at a l-cm spacing

using a Baliington Instruments M.S.2.F spot sensor.

The suceptometer was operated in the sensitive

range and each measurement corrected with a sepa­

rate background reading.

In most marine sediments susceptibility prima­

rily quantifies magnetite content (Thompson and

Oldfield, 1986), which is generally part ofthe terri­

genous input, although it may occasionally be

supplemented by bacterial magnetite (Petersen et

al., 1986, Vali et al. , 1987). The susceptibility signal

of marine sediment series frequently reflects the

ratio ofbiogenic and lithogenic components (Robin­

son, 1990), which may vary due to changes in terri­

genous input, carbonate production and dissolution,

or a combination of these often climatically con­

trolled mechanisms. Susceptibility records have

been regionally established as excellent paleocli­

matic proxies (e.g., Mead et al., 1986; Bloemendal

et al. , 1988; Bloemendal and de Menocal, 1989;

Park et al. , 1993; Robinson et al., 1995; Chi and

Mienert, 1996).

Ifmagnetic susceptibility is interpreted as inverse

carbonate proxy, all potential exceptions like sand

and ash layers 01' sections affected by reductive

diagenesis must be excluded (e.g., Bloemendal et

al. , 1989; Frederichs et al., 1999a). Except for the

earlier mentioned turbidites, there is no indication

of such effects in any of the twelve cores. Their

magnetic susceptibility records reflect variations in

carbonate content as the comparison of magnetic

susceptibility and %CaC03 (Bickert, 1992) records

of GeoB 1211-3 show (Figure 2). The inverse cor­

relation of susceptibility and carbonate content is

very close in all period ranges. Long-term trends,

Milankovitch cycles and high-frequent signal fea­

tures are weIl reproduced, although comparison of

the latter is restricted due to the much lower resolu­

tion ofthe %CaC03 record (5 cm spacing). Only in

isotope stage 14 at about 550 ka considerable differ­

ences appeal'. The extrapolated end-member values

for 100% and 0% carbonate are -5.6.10-6 SI units

Cycles, trends and events of Pleistocene sedimentation 53

and 464.1.10-6 SI units, respectively. The suscepti­

bility of pure diamagnetic calcite, -15.10-6 SI units

(Thompson and Oldfield, 1986), is the theoretical

absolute minimum and logical origin ofthe suscep­

tibility axis.

As mentioned above, the three factors goveming

the carbonate content of marine sediments are pro­

ductivity fluctuations ofcalcareous organisms, dilu­

tion by telTigenous material, and calcium carbonate

dissolution. Volat et al. (1980) analyzed Pleistocene

sediments in the Pacific, Indian, and Atlantic Ocean

and concluded that dissolution is the most impoliant

factor influencing carbonate content in all three

oceans. John50n et al. (1977) found carbonate cycles

in late Pleistocene sediments recovered in the Vema

Charmel and on the lower flanks ofthe Rio Grande

Rise at water depths between 2900 and 4000 m.

They evaluated the extent to which these cycles may

be dissolution controlled by comparing carbonate

content data and a semi-quantitative measure of

dissolution. Finding a strong cOlTelation between

low carbonate content and a high dissolution index

they interpret the variations in carbonate content as

dissolution cycles.

While in the South Atlantic carbonate dilution

by varying telTigenous sedimentation is important

near coasts and in regions of enhanced fluvial and

eolian sedimentation, it diminishes with distance

from the continents (Schmidt et al., submitted). In

spite of the fact, that telTigenous input is much

greater in the westem than in the eastem South

Atlantic (Lisitzin, 1996), even our westemmost core

GeoB 2821-1 consists primarily (70±9 %) of

CaC03 (P. Müller, unpublished data). Varying

terrigenous input certainly effects the susceptibility

signals to some extent, but carbonate dissolution

by alternating influence of NADW and more

cOlTosive LCDW is believed to be the main cause

of signal variance in the records presented here.

Chronostratigraphy

Magnetostratigraphies of GeoB 1034-3 and

1211-3 were established by Thiessen (1993) and

for all other cores by von Dobeneck and Schmieder

(1999). Except for GeoB 1035-4 and 3801-6 all

cores reach the Brunhes/Matuyama boundary

(780 ka) and the Jaramillo Event (990-1070 ka). In

three cases the short Cobb Mountain Event

(1190 ka) was detected despite low sedimentation

rates. Oxygen isotope records exist for

GeoB 1034-3,1035-4,1211-3 (Bickert, 1992;

BickertandWefer, 1996)and l309-2(W. Hale,per­

sonal communication). Their stratigraphic interpre­

tability is limited by very low sedimentation rates,

particularly in the deeper, obliquity-dominated sec­

tions. By interpolating between the magnetostrati­

graphic tie points, spectral characteristics ofMilan­

kovitch cyclicity become apparent in the suscepti­

bility logs. These initial spectra display significant

maxima at around 100 and 40 kyr, but show little

01' no indications ofa precessional signal component

(von Dobeneck and Schmieder, 1999).

A pattem matching to standard 8180 records, e.g.

from equatorial Pacific ODP Site 677 (Shackleton

et al. , 1990) was feasible over most signal sections,

but remained ambiguous between oxygen isotope

stages 16 and 13 between about 650 and 500 ka.

Continuously high similarities, including long-term

fluctuations of about 500 kyr duration, exist bet­

ween all susceptibility signals and the benthic 813C

record at Pacific ODP site 806 (T. Bickeli, unpub­

lished data) based on an orbitally tuned 8 180 strati­

graphy (Berger et al., 1994). The complete and un­

disturbed susceptibility record ofcore GeoB 3814-6

(Figure 3b) cOlTelates particularly weIl and helped

to substantiate the other eleven matches.

The most widely used orbital tuning procedure

assumes a phase lock between climate and metro­

nome record (Martinson et al., 1987). In order to

assess the absolute lag of any proxy record with

54 Chapter 4

100 200 300 400 500 600 700 800 900 1000 1100 1200 1300 1400 1500Age [ka]

!~-1~,--~-+'------~+--+-----,~----'---+--'--'--+-----'-'--+-+-c---+---'-+--+-+---'- ' ',~_._._J

, ~~i_',_,_\ /_ !_i ! i! f !_~~__ \ I

ill:~1V1°1\MJ~~~\f\;Wr\

Polarity

813CODP 806(inverted)

Suscepti­

bility K

Obliquity(inverted)

K

filtered15-47 kYr H'---r'-r-;+I--'r'+-'r-+'r+-r-'-r'+'++-.-'+-T'-'i-+'r+-r--T+'i-Y-',-'+'-r-+-r-r+,-,---r-'r+-r+i-'-r+'-+--1

o

TargetCurve

K

filtered

35-47 kyr 'Y-7-H+-~~+-~+-~~~+-~+-;-+-'----'c-'~--'-+c~',--+--!--,-c,-+,,-',-'--'--',-+-+T-~

Figllre 3. Four step refinement ofthe age model for core GeoB 3814-6. (a) Susceptibility record dated by magnetostratigraphy(3 tie points). (b) Signal pattern corre1ation to the tuned 81'C record ofODP Site 806 (T. Bickert, unpubl. data, 44 tie points).In critical intervals other SUSAS records were inspected for supplementary information (c) Susceptibility filtered (35-47 kyr)in the obliquity range (dashed line) and tuned (solid line) to an astronomical obliquity signal shifted by -4.5 kyr (73 tie points).(d) Susceptibility filtered (15-47 kyr) and tuned to an astronomical (obliquity and precession) target curve (95 tie points).

respect to astronomical obliquity and precession,

an independent and precise absolute chronology is

required. 8180 phase lags have been determined on

basis ofradiometric ages (Hays et al., 1976: 9 kyr

for obliquity, 3 kyr for precession) and assumptions

on the coupling of insolation and ice mass (Imbrie

et al. , 1984: 7.9 (7.4-8.2) kyr for obliquity, 5.0 (4.8­

5.1) kyr and 4.2 (4.1-4.3) kyr for precession). To

estimate orbital phase lags for regional proxy para­

meters such as susceptibility it is therefore sufficient

to know the phase lags relative to a 8180 curve from

the same core.

Here we use a revised 8 180 time scale of core

GeoB 1211-3 (Bickeli, 1992) based on a correlation

to the SPECMAP stack (Imbrie et al., 1984). Cross

spectral analysis of 8 180 and K yields a coherence

of 0.99 and 0.96 for the 100 kyr and 41 kyr cycles

and reasonable phase angles of -41 ° and -30°. This

implies that susceptibility leads 8180 by 3.4 kyr in

the obliquity band and therefore lags the obliquity

forcing function by 7.9 kyr-3.4 kyr=4.5 kyr and is

in good agreement with Imbrie et al. 's (1993) values

calculated for early proxy responses in the southern

hemisphere. The temporal resolution of the 8 180

record was too low to determine the precessional

phase lags within acceptable elTor margins.

The subsequent orbital tuning process was per­

fonned in two stages. At first the obliquity-related

signal component was extracted applying a 1si order

butterworth bandpass (35-47 kyr) filter in forward

and reverse direction. Each maximum and minimum

in the filtered signal was assigned the age (+4.5 kyr)

Cyc1es, trends and events ofPleistocene sedimentation 55

Age [ka]

0.53

: : GeoB: : 2821-1

: : OA6

: : 0.74

:037 :

OA7: ,

:034: :

:051:

500 1000 1500

0,56

'0:8 : :GeoBiv:H .: 2820-2

0.49

: :GeoB: : 1311-1 -­039

500 1000 1500

~.__••_ "',!""f--:: : •GeoB 1-

:1:0 :3813-3

0.70"':

: :GeoB: :3801-6

: : GeoB: : 1729-3

:OAO :

: :0.70

:0.47: :

: :089

. ,:GeoB'L :: 3812-1

TIH:~:

:0.37: ,: :050

500 1000 1500 0

500 1000 1500 0

104

0.69

5

5

5

5

5

:057: . I. r:068"-. t10 ---f-,-,-,-,--,i-,--,-.---c-,-,-..---.t- ---f-,-,-,-,--,i--r-rT':!'-.-n+

o-+--'-.-L.L--'--'+-.L-'-L.L...J-'---'--'-q: . GeoB. ·1034-3

~

~10 --h--.-r..,.--,'-rr-r-r-+-r-+cr-rl-- -t--,--,-,---,----r-r-r-.-r--r-r-..-rl­

o-l:--'---'--'--'----S-'-7'----'---++-"--'--'-"---j- -k--'---'--'---,--*-,-,-,---,---+~---'--'--L-'---I-

10 --h--r-r--,-,i';-o.,--,-,o--r-r-,---,--'-+-- --h-,-,---,-,i..,..,--,---,-,---,..--rl-­

o-t--'-'-J....L...J;;-lW---'-lL.L.L.L.L-'-f-- --k-'--'-L...L...J;--'-'-'----'--"--'---'---'--'--L.j--

10 --h--r-r--rl!e,-L,---.++-r-,---,--'-+-- --h--r-r--rl!e,-L,---.++-r-..--rl-­

o---k-'-'-L--'-fi'-Y---'-';-+-'--'---'---'-f-- ---k-'-'-L--'-fi'-Y---'-';-+-'---'--'--L.j--

10 --h-,-,---,-,i..,..,--r-r-,---,-,---,,-f- --h--r-r..,.--,'-,-,--r-r-,---,-,---,rl-­

o

E~ 10 --h-,-,-"""'';-'---r-r;-T--r--T-.--rl-- --h-'-T"""''''''-'---':--i-r''0+",5+-7rl--

ö.. 0 -k--'-.-L.L-'--If.!,.L-'-4--!-J--'--'-"---j- -k:'--'-L--'--4-.l,-J--'-4--!-J--'--'-"---j­(])

o

Figure 4. Age depth relation ofall SUSAS cores. Sedimenta­tion rates are generally fairly constant. Dashed lines mark theMPT interim state, bounded by two transitions. During thisinterval sedimentation decreased. The grey vertical bar denotesthe time ofa terminal MPT event indicated by unusuallitholo­gies in several SUSAS cores (for abreviations see Figure 5).

of its postulated eguivalent in the astronomical

record (Berger and Loutre, 1991), starting from the

most uncritical sections near magnetostratigraphic

01' other reliable tie points. As illustrated in Fig­

ure 3c, changes in the correlation age model are

small, typically less than a half-cyc1e of obliguity.

The second tuning step procceds from the observa­

tion that faintly visible precessional peaks in the

spectra gradually sharpened at both refinements of

the age model (von Dobeneck and Schmieder,

1999). To benefit from this precessional signal com­

ponent, a target curve of nOl1nalized obliguity and

precession index (mixing ratio of 2:3 chosen by

visual evaluation) and 4.5 kyr lag was calculated,

comprising notjust more, but also more prominent

signal features than either single orbital parameter.

The obliquity-tuned primary susceptibility record

was filtered using a wide bandpass (15-47 kyr) to

inc1ude obJiguity and precession cycles. As many

extrema as possible were l11atched (Figure 3d),

admittedly at the resolution limit of these records.

The resulting final age models are again j ust slight

moelifications ofthe previous models anel possibly

not even more precise in absolute ages as the pre­

cession time lag is undetermineel. Nevertheless, this

higher-fiequency tuning leads to a bettel' mutual sig­

nal cOlTelation ofthe twelve records.

Results and Discussion

Figures 4 and 5 sUl11l11arize the combined results of

magnetic age modeJing for the complete transect.

Resulting sedimentation rates are fairly constant

over time. A few question marks remain in core

sections, where individual signal features seem to

be missing 01' incomplete (e.g., GeoB 1311-1 at

420 ka, Ge oB 2820-2 at 800 ka). The high

confOlmity of all records led us to cOlnpute the

SUSAS (Subtropical.§outh Atlantic .§usceptibility)

stack, which is free of local effects anel therefore

representative for the whole Quatemary oligotrophic

56 Chapter 4

100 kyr State Interim State :!t~;~ 41 kyr State 1'---------------+--_f---- -1-_-l-________ _ J

o 100 200 300 400 500 600' 700 800 900 1000 1100 1200 1300 1400 1500 1600I, ! ! ! I ! ! ! ! I ! ! I , I ! ! ! , I ! ! , ! I I ! I ! I ! I ! ( I , I , ! I I I I , j ! ! ! ! ! ! ! ! ! I ! ! ! I I , ! ! i I ! ! ! , I , ! ! , I ! ! ! , I , ! ! ! I ! ! ! , I ! ; ! , ! , ! ! ! I ! , ! , I ! ! ! \ I ! I I I 11 ! I I I ' I ! I I I ! ! t I ! LU1J...uLLu.1..LLJ..l..1i.L.Lul~+-

::U~~1r1;;vvv~~~"OOkY~' ~~~ci:1~~i~""11~~"~~,!;:~ ~~B3i~;~31111

. .' , -----i·~i

75~ L Fine Laminated Layers Ii\j I;~ . . . .. 0 Laminated Diatom oo~ GeoB 1035-4 I:S: i

o· 1.18 .' •••_._'_.__ WO_445;3_JTL~ I

i~~ G"B1729"8i~!'0.47 :0.89 WO 4401 m

UJIGeoB 3801-6 ~

..........,- WO 4546 m ~

"""TH'~ GeoB 3812·' J-------~c___:__--c___:__----~.c-----------:---:0"--'-.5=---:0......,.,..______...________...__----WO 4205 m <;(

:E

"tAJGeOB 3813.3 :!WO 4331 m

500.

250 GeoB 1309-2 CI>

o---e:.:.:....:.-----------+-~-----i-~~---_______,Ä------:.--=-='---------=_______,_____,._--_______,-------'=----"--"'-"--"-___Ir___i ~200 ~

t::100 ~

(!J--c.:-=--=--------------:---c----;-:--:.:..:.-----------I\--------i-..::...:.-=------.--------,r-------------'--'-""----'=-"'---'--'-'--'-----1.!2

Cl::

21o·f iV-V·,' .,.f-'.. _."--d\",j;- -I,,·,· I··"'-"· .+VI,'.. j

-1

-2 ~ ""'.: : •

to:~~:,~"""""" "" "" """ I' ,C,. I" "",h1~" 'I '" '" ""1",, ,11,!~ C!:~~\,~i:,~~~~~~;~~,~~,~~"~,\~~,~ ,o 100 200 300 400 500 600 700 800 900 1000 1100 1200 1300 1400 1500 1600

Age [ka]

Cycles, trends and events of Pleistocene sedimentation 57

deep South At1antic Ocean. Because of different

amplitude ranges, the stack was determined as an

arithmetic mean ofnormalized (subtraction ofcore

mean and division by standard deviation), interpo­

lated (Llt = 2 kyr) records.

Some important features of the SUSAS series

discussed in the following are visible by closer

visual inspection ofFigure 4:

• an east-west trend of increasing average mag­

netic susceptibilities,

• changing signal cyclicities during the Pleistocene

with particularly uniform pattems during the

early (41 kyr state) and late Pleistocene (100 kyr

state),

• a shift towards enhanced magnetic susceptibili­

ties and reduced sedimentation rates during the

intermediate time interval,

• the premature occurrence ofa near-1 00 kyr cycle

at about 1150 ka,

• a paleoceanographic event at about 530 ka

documented in unusual litho10gies in several

SUSAS cares.

Adapted statistical analysis must be applied to gain

insight into these phenomena and their interrela­

tions.

Figure 5. (opposite) Individually tuned SUSAS records(WD = Water Depth) and resulting susceptibility (arithmeticmean with standard deviation band) and sedimentation ratestack. A diamagnetic susceptibility of -15 x10-6 SI units rep­resenting pure carbonate sediment was chosen as axis mini­mum. To compensate for different signal levels in the stack­ing, each record was normalized by subtracting its mean anddividing by its standard deviation. Labels at the SUSAS stackindicate even oxygen isotope stages. Horizontal dotted linesmark baseline averages for pre-, syn- and post-MPT statesexcluding the bounding transitions. During the interim state,all cores display increased susceptibilities and reduced sedi­mentation rates. The timing of the three intervals approxi­mately corresponds to the Laplace (1800-1200 ka), eroll(1200-600 ka) and Milankovitch chron (600-0 ka) defined byBerger and Wefer (1992). Numbers in each section denotemean sedimentation rates; grey vertical bars apremature100 kyr cycle centred near 1150 ka and a telminal MPT eventfeaturing unusuallithologies at about 530 ka.

Cycles

Evolutionary spectral analysis is an excellent

method to visualize the variability ofcyclic climate

characteristics in arecord through time, either

presented as a sectionalized analysis (e.g., Pestiaux

and Berger, 1984; Bloemendal and deMenocal,

1989; Mwenifumbo and Blangy, 1991; Tiedemann

et al., 1994; Berger and Jansen, 1994) ar in the form

of'spectrograms' (e.g., Joyce et al., 1990; Yiou et

al. , 1991; Birchfield and Ghil, 1993; Glützner et

al., 1997; Harris et al., 1997; Paillard, 1998). These

three-dimensiona1 diagrams result from spectra

generated within a moving window and thus com­

bine frequency and time domain. The SPECGRAM

algarithm embedded in the MATLAB Signal Pro­

cessing Toolbox was used to calcu1ate the spectro­

grams in Figure 6. For all time series analyzed a

500 kyr frame was advanced at 10 kyr steps and, in

order to minimize cut-off effects, tapered with a

Hmming window, thereby focusing to the central

section at each step. Long-term trends (periods

>350 kyr) were previously removed by highpass fil­

tering to minimize disturbance by this signal compo­

nent. The resulting data matrix was normalized to

an average spectra1 density of 1 by dividing all

values by the total matrix mean. This procedure

merely scales all spectrograms uniformly and

enables their comparison, but does not alter relative

variations in the time or frequency domain. As a

consequence ofthe window length used no spectra

cou1d be calculated for the first and the last 250 kyr

ofthe record. In order to avoid 10ss ofinformation

on the early Pleistocene the SUSAS stack was exten­

ded beyond the age of 1530 ka by the normalized

signal of GeoB 1311-1 which reaches back to

1730 ka (Figure 4).

In the resulting spectrogram (Figure 6c) the

precession re1ated signal is poorly documented due

to bioturbation damping with the exception of

sections ofrelative1y higher sedimentation rates and

58

0.07

0.06

0.05

~ 0.04

>­uc~~ 0.03

lL

0.02

0.01

0.00

Chapter 4

19.0

22.423.7

29

41

54

95124

300 400 500 600 700 800 900 1000 1100 1200 1300 1400 300 400 500 600 700 800 900 1000 1100 1200 1300 1400

0.07

0.06

0.05

};:0 0.04

>­uc~~ 0.03

lL

0.02

0.01

0.00300 400 500 600 700 800 900 1000 1100 1200 1300 1400 300 400 500 600 700 800 900 1000 1100 1200 1300 1400

0.07

0.06

0.05

]<-'=. 0.04

>­uc

~ 0.03~lL

0.02

0.01

0.00300 400 500 600 700 800 900 1000 1100 1200 1300 1400

Age[l<aJ300 400 500 600 700 800 900 1000 1100 1200 1300 1400

Age[l<a]

Figure 6. Evolutionary spectral analysis of driving and responding climate variables. The spectrograms were calculated byadvancing a tapered (Hanning window) 500 kyr frame at 10 kyr steps. All records were previously highpass filtered « 350 kyr)to minimize disturbance by low-frequency signal components.(a) Mid-month insolation at 65°N for July (Berger and Loutre,1991), (b) ETP curve, calculated from eccentricity, obliquity and precession by calibrating the cumulative spectral intensity ineach band to equal that of(c), the SUSAS spectrogram. (d) For a comparison offorcing and response the numerical differenceof (c) and (b) was calculated. In general, spectrograms of core GeoB 2821-1(e) and GeoB 3814-6 (f) displaya good overallcorrelation to each other and the SUSAS stack (c), slight deviations occur during the MPT interim state.

Cycles, trends and events ofPleistocene sedimentation 59

enhanced forcing (e.g., prior to 900 ka). The in­

fluence ofobliquity on the SUSAS stack diminishes

continuously with a major decrease near 1200 ka

and a second, somewhat less enhanced reduction at

about 700 ka. Near 100 kyr periods clearly dominate

the signal in late Pleistocene since 650 ka, but also

give important contributions to older sections, par­

ticularly expressed between 1250 and 1050 ka. In

this time range strong spectral components between

the obliquity and the eccentricity band emerge. At

about 1150 ka they fuse with near 100 kyr cyclicity

into a broad maximum. This merging is a result of

the compromise between time and fiequency resolu­

tion. A conventional sectionalized spectral analysis

with a Ionger time window (1530-920 ka) separates

a component with aperiod ofabout 70 kyr (Figure 7,

top). Anomalous spectral peaks near 70 kyr have

been repOlied from Pleistocene paleoclimate records

ofthe North Atlantic (e.g., Ruddiman et a1., 1989),

the Indian (e.g., Robinson, 1990), and the Pacific

Ocean (e.g., Bassinot et a1., 1994a), but few attempts

have been made to explain their origin. Considering

the simultaneous presence of strong variance in the

100 and the 41 kyr band in the SUSAS record

(Figure 5, top and 7c), a nonlinear interference of

these two frequencies (1/41-1/100=1/69), pre­

viously discussed by Robinson (1990) and Muller

and Mac Donald (1997), is a plausible explanation.

A simple nonlinear model of sea-level change

produces a spectral peak near 72 kyr beside moving

energy from the orbital forcing bands into the

approximately 100 kyr band (Berger et a1., 1996).

Similar to our results Mudelsee and Stattegger

(1997) identified high 100 and 41 kyr amplitudes

in the benthic oxygen isotope records ofODP sites

607 and 659 in the time period around 1200 ka.

They hypothesize that this interval of generally

stronger climate fluctuations was 'a first but

unsuccessful "attempt" of the climate system to

attain a nonlinear "late Pleistocene ice ages" state'.

Our spectrogram analysis strongly supports this

assumption.

The evolution ofthe 100 and 41 kyr cycIe docu­

mented in the SUSAS spectrogram matches in great

detail the results of statistical analysis of benthic

oxygen isotopes of DSDP site 607 am] ODP sites

659 and 677 (Park and Maasch, 1993; Mudelsee

and Stattegger, 1997): strongest Pleistocene 41 kyr

amplitudes between 1400-1200 ka and nearly con­

stant values since about 1100 ka as weIl as increas­

ing 100 kyr cycles at about 650 ka and enhanced

values between 1200 and 1000 ka. Despite some

deviations the impressive correlation ofour analysis

and the calculations made for benthic 8 180 time

series from the midlatitude North Atlantic

(DSDP 607, ODP 659), and the equatorial Pacific

(ODP 677) suggests aglobaI character ofthe change

fr0111 41 to 100 kyr cyclicity. Evidently the cyclic

variations ofthe SUSAS stack are closely linked to

global ice volume via changes in deep water chemis­

try.

How is the observed evolution of the cyclicity

related to the driving orbital fluctuations? In Fig­

ure 6a-d spectrograms of forcing and responding

variables are compared. The mid-summer insolation

signal at 65oN (Figure 6a, Berger and Loutre, 1991)

and a norn1alized ETP curve (Figure 6b) are shown

as reference. ETP curves are artificial, but often

employed target records composed by calibrating

and adding the time series of Eccentricity, Tilt

(obliquity) and Precession (Imbrie et a1. , 1984).

Here, the cumulative spectral intensity in each band

ofthe ETP spectrogram was calibrated to equal that

of the SUSAS spectrogram. Being based on the

same orbital variations, both reference signals exhi­

bit identical amplitude modulation patterns, but very

different spectral power in each band. In the eccen­

tricity band this difference amounts to several orders

of magnitude - an expression of the 100 kyr cycle

problem (Imbrie et a1. , 1993).

60 Chapter 4

Figure 7. Spectral analysis ofthe SUSAS stack far the threePleistocene states defined in Figure 5.

taneous decrease in obliquity although the gradient

is steeper as in Figures 6a, b. The reduction near

700 to 600 ka and the subsequently low amplitudes

during the entire late Quaternary however do 110t

mirror the evolution ofthe forcing signal. Quite the

reverse, obliquity amplitudes increase again during

the late Pleistocene (Figure 6a, b) and thus praduce

negative residues (Figure 6d). Obviously the ability

of obliquity to drive the observed carbonate disso­

lution cycles decreased during Pleistocene times and

remained low since the onset of 100 kyr cyclicity

at about 650 ka. This finding is in accordance with

the suggestion ofRuddiman et al. (1989) and Joyce

et al. (1990) that the climatic system was less sensi­

tive to obliquity forcing during the late Pleistocene

and as well with the results of Imbrie (1992) who

found, that response in the obliquity band was fairly

constant over the past half-million years.

The coincidence of target and proxy records in

the 100 kyr band is even lower, as indicated by

larger residues and steeper gradients. While the ETP

modellooses 100 kyr power in the late Pleistocene

(Figure 6b), the SUSAS stack, like most climate

records, documents an intensification resulting in a

positive residue. The preceding residue maximum

originates from the early Pleistocene 100 kyr cycle

near 1150 ka. Two low-frequency positive maxima

centered at 950 and 600 ka in the SUSAS and the

residual spectrogram (Figure 6c, d) result from re­

mainders of the earlier mentioned base line transi­

tions ofthe long-term susceptibility trend (Figure 5).

The most striking result of the residual spectro­

gram is the reciprocity of spectral intensities in the

41 and 100 kyr bands (dashed outlines in Figure 6d).

In relation to the ETP model, the SUSAS 100 kyr

cyclicity develops over-proportionally in sections

with reduced response to obliquity (blue outline)

and retreats, where the response to obliquity is

strong (yellow outline). These findings imply an

exchange of spectral energy between the obliquity

>:, >:, >:,>:, -" -" -"-" r- '<T 0

:; C') N 0)

N N

I I

I I

>, ~ ~

-" >, >,'<T -" -"N LO '<T~ 0) LO

I I I70 kyr

I I ~ I

0.00 0.01 0.02 0.03 0.04 0.05 0.06 0.07

Frequency [1/kyr]

20

o

10

-10

-10

-30

-20

co2-;:;.

20';nc -10Q)

u

~ 10Ü -20Q)CL

Cf) 0-30

The numerical difference between the spectro­

grams of the SUSAS stack and the ETP curve is

displayed in Figure 6d. With restrietions due to the

overalliow variability in the 19 and 23 kyr bands,

the assumption ofa constant proxy response appears

justified for precession, as both records 1argely

compensate throughout the whole record. This

hypothesis can also be drawn from simple visual

comparison ofFigures 6a and c. The reduced 23 kyr

maximum centered at 600 ka and enhanced values

in the 19 kyr band near 400 ka may result from the

influence ofdifferent sedimentation rates (Figure 4)

on the definition of these high-frequency cycles.

41 kyr cyclicity is less affected by bioturbation

damping (von Dobeneck and Schmieder, 1999).

Yet, residues in the obliquity band show a more

variable, non-stationary response to forcing

(Figure 6d). The spectra1 intensity decline at 1200

to 11 00 ka (Figure 6c) may result from the simul-

Cycles, trends and events of Pleistocene sedimentation GI

and eccentricity band. A possible explanation is

provided by climate models invoking achanging

insolation threshold (e.g., Saltzman and Verbitsky,

1993; Raymo, 1997; Paillard, 1998). Intensified

northern hemisphere summer insolation capable of

triggering major deglaciations hence should result

fl:om an interference ofobliquity maxima with pre­

cession index minima. At times, when the insolation

threshold is relatively low, obliquity maxima alone

will trigger the withdrawal ofcontinental ice shields

and 41 kyr cyclicity prevails. When the insolation

threshold is higher, the required peak insolation is

only reached by optimum interaction of obliquity

and precession. As the precession amplitude is

modulated by eccentricity, sufficiently large pre­

cessional peaks occur only during 100 kyr eccentIi­

city maxima suppressing obliquity and causing the

observed 100 kyr climate cycles. While a linearly

increasing threshold (e.g., Raymo, 1997; Paillard,

1998; Raymo, 1998) results in a single transition

from 41 to 100 kyr predominance, a slightly fluc­

tuating threshold should also be capable to explain

the 100 kyr excursion near 1150 ka.

The good overall correlation of all records dis­

cussed in the time domain is also visible in the

synchronous time-frequency view of individual

spectrograms. Slight deviations in the spectrograms

of cores GeoB 2821-1 and 3814-6 possibly relate

to local effects (Figure 6e, f) and are concentrated

in the MPT interim state, where both records exhibit

higher variance in the obliquity band than the stack.

In GeoB 3814-6 this response includes a strong

54 kyr component, which is also visible in

GeoB 1211-3 (not shown) and in the spectrum of

the corresponding SUSAS stack section (Figure 7,

center). In both cores this subordinate obliquity

cycle amplitude reaches values comparable to the

41 kyr cycle although the latter, summarizing four

major terms, contributes six times as much to

astronomical obliquity variation (e.g., Berger and

Loutre, 1992). The mid-Pleistocene increase in

response to obliquity is even more puzzling as it

precisely coincides with a time of reduced ampli­

tudes in the forcing function (Figure Ga, b). The

three diverging cores were recovered from southern

ridge slopes 01' deep water passages (Figure 1) and

are thus more intensely influenced by southern

source waters. The influence ofobliquity-dominated

high latitude climate forcing was possibly intensi­

fied during the MPT interim state.

The evolutionary spectral analysis ofthe SUSAS

stack documents a continuous Pleistocene decline

of41 kyr cyclicity, but a comparatively abrupt inten­

sification of 100 kyr cyclicity after about 650 ka.

Oxygen isotope records indicate larger ice shields

since the beginning of the MPT interim state at

920 ka (e.g., Berger and Jansen, 1994). Why did

100 kyr cyclicity lag the initial ice volume increase

by approximately 280 kyr (Mudelsee and Schulz,

1997)? The observed baseline shift in the SUSAS

records (Figure 5) precisely fills this time lag and

ShOll1d shed light on this question.

Trends

A prominent feature of the SUSAS stack and all

individual records summarized in Figure 5 is a base­

line shift towards 40% higher average susceptibili­

ties during the MPT interim state (920 to 640 ka).

Confined by transitional intervals lasting about 80

and 140 kyr at its onset and termination, this period

exhibits the most distinct dissimilarities between

individual core logs. The preceding and following

climate states show similar mean susceptibilities

modulated by clearly developed and coherent 41

and 100 kyr cycles, respectively. An exception is

the 'premature' near-100 kyr cycle at around

1150 ka discussed above. The MPT susceptibility

shift is notably milTOl'ed in lowered average sedi­

mentation rates (Figure 5, bottom).

62 Chapter 4

~~yr S-;-t~ [500-' 0 kaJlI

Interim State (920 • 640!ffiJJ [ 41 kyr State (> 1000 ka)-Ii

_~~_~_~ __J

- 40"W 30"W 20"W 10"W 0' 10"E 40"W 30"W 20"W 10'W 0" 10"E 40"W 30"W 20"W 10"W 0" WECf) 800 800 800<D 2821- 1 3814-6 3813_3 3814-66 500- 500 • I. 500 2821-1 • 500-,-

3813-3 2~0-21309-2 • • 1309-2

.~ 300 3'12-1 300 300 • • • 3003812-'12820-2 3813-3

:n 200 • 200 1729-3 200 200"ß. •(lJ

100 100 • 100u(fJ • 1211-3::J 1211-3Cf) 50 50 50u o -~ 30 wO', J30 • 30c 20

• 20_ Enhanced Susceptibility 1034-3 - 20OJ 10 1034-3 10 10 10(\l Constant W·E Gradient:2:

0 -, 0 0

'L' 40"W 30"W 20"W 10"W 0" 10"E 40"W 30"W 20"W 10"W 0" 10"E 40"W 30"W 20"W 10"W 0' 10"E>. I I I I ._L- I I I.Y 3_0 3814-8 ~30 3814-8 3.0

N

• 3813-32821-1 • •E 20 2821-1 • • • • 3813-3

~ • 2.01309-2 1309-2

C •C 3801-8 2820-2 •0 10 • 1_0~

3812-1 1211-3 3812-1

1729-3 • 1729-3::J •E •0.5 0 0_5::J 1035-4U 0U 1311-1« 0_3 0_3-

21034-3 Constant Magnetite 1034-302L • 0.2 •~ Accumulation

c Constant W·E GradientOJ(\l 01 . 01 0.1

:2:-~l -,

40"W 30"W 20'W 10"W 0" 10"E 40'W 30"W 20"W 10"W 0" 10"E 40"W 30"W 20"W 10"W 0" 10"E1_2 1.2 I I I 1_2 1.21_1 3801-8 1035-4 1.1 Reduced Sedimentation 1.1 1034-3 1_1

'L' •>. 1_0 1211-3 10Collapsed W·E Gradient

10 • 1_0~ 0_9 • 0.9 0_9E 0_8

1034-308 08Q 2820-2

(lJ 0-7 0.7 0.7 •ro 1034-3• •0::: 0.6 2821-1 3814-8 • 0_6 3814-8 3813-3 0.6 1309-2c • • • 2821-1 •0 3812-1

2821-1 • •~

0_5 • - 0_5 -0_5 •2820-2 • 1309-2 • 0 3812-1

C 1729-3 1311-1(lJ OA 0 OA 3812-1 • OA •E '1311-1 • • 1211-3 3814-8

iJ 2820-2 0(lJ 1311-1

Cf) 0_3 0.3 0.3 0.340 0 W 30 0 W 20 0 W 10 0 W 0° 10 0 E 40 0 W 30 0 W 20 0 W 10 0 W 0° 100 E 40 0 W 30 0 W 20 0 W 100 W 0° 10 0 E

Longitude

Figure 8. Combined spatial and temporal trend analyses ofmagnetic susceptibility, magnetite accumulation (caIculated bysplitting susceptibility into a diamagnetic background and ferrimagnetic, grain-size independent magnetite signal) and sedi­mentation rate_ Major asymmetries and shifts are sunu11arized by white regression Iines, dark-gray shading delimits mean datarange, light-gray shading western and eastern South Atlantic care sets. Three cores were excIuded from the analysis: coresGeoB 1035-4 and 3801-6, as they do not reach beyond the 530 ka event, and core GeoB 1311-1 recovered from 290 1 m waterdepth cIearly above the LCDW/NADW transition zone and therefore showing a somewhat different evolution.

The cause for the overall rise in magnetic suscep­

tibility during the MPT interim state is revealed by

a temporal analysis of west-east trends. Plotted

against longitude, core mean susceptibilities far the

postulated three climate states (Figure 8 top) reflect

the well-known west-east asymmetry ofterrigenous

particle flux and accumulation in the South Atlantic

(e.g., Balsam and McCoy, 1987; Lisitzin, 1996).

While the gradients are almost identical throughout

Pleistocene, a shift to higher susceptibilities is evi­

dent during the MPT interim state. This relative in­

crease ofmagnetic mineral concentration cannot be

explained in terms oftemporal changes in magnetite

accumulation (Figure 8 center), as its west-east de­

cline from 2.5 to 0.3 g/m2kyr remains fairly constant

over time. A time-varying dilution ofthe terrigenous

Cycles, trends and events of Pleistocene sedimentation 63

fraction by non-magnetic carbonate must therefore

be responsible for the observed shift.

Mean sedimentation rates (Figure 8 bottom)

deduced from our age models are mainly controlled

by carbonate accumulation. Both for the 41 and

100 kyr climate states, sedimentation rates vary

from about 0.5 cm/kyr in the western to around

0.9 cm/kyr in the eastern South Atlantic. During

MPT interim state sedimentation rates on either side

ofthe mid-Atlantic Ridge are restricted to between

0.4 and 0.6 cm/kyr. The decline primarily aifects

the eastern part and brings the west-east asymmetry

to collapse.

In the working area cyclic variations ofthe sedi­

ment CaC03 content at Milankovitch frequencies

are mainly due to orbitally driven lysocline shifts

resulting from an interplay of NADW and more

corrosive LCDW (Bickert and Wefer, 1996). In the

Angola Basin (cores GeoB 1034-3, 1035-4,1729-3

and 3801-6) the glacial-interglacial contrast is

particularly manifest as the surrounding bathymetric

highs restrict the access of LCDW during inter­

glacials. The same reasoning should apply for long­

tern1 changes in carbonate accumulation. We there­

fore assume that the influence of southern-source

deep water was greatly enhanced during the MPT

interim state. This conclusion is supported by a

simultaneous decrease of glacial and interglacial

kaolinite/chlorite ratios in core GeoB 2821-1

(Gingele et a1., 1999), interpreted to be induced by

chlorite-enriched deep southern waters. Conse­

quently, NADW should have been reduced during

that interval. Indeed, 013C records from Atlantic and

Pacific ODP sites indicate significantly weaker

NADW between 900 and 400 ka (Raymo et a1. ,

1997). In view of this global evidence we suggest,

that the MPT should not be regarded as a gradual

transition frOl11 a '41 kyr world' to a '100 kyr

world', but rather as a third, contrasting climate

state.

Terminal MPT event

In several of the SUSAS COl'es unusual sediment

facies occur coincidentally at around 530 ka, dose

to the end ofthe terminal MPT transition. Granu!o­

metric analyses of cores GeoR 2820-2 and 2821-1

show a sharp grain-size shift with an almost total

loss ofthe> 63 flm fraction (Breitzke, 1(97), possi­

bly related to the mid-Brunhes dissolution cyc1e

(e.g, Adelseck, 1977). In core GeoB 2820-2 this

episode corresponds to a short hiatus. The continu­

ous late Pleistocene record of core GeoB 1035-4 is

interrupted by a thick turbidite dating to the same

age. Core GeoB 3812-1 displays a sharp color

change at a delicately laminated horizon (Figure 9).

Most impressive are thick intercalated laminated

diatom ooze layers in cores GeoB 3801-6 (124 cm)

and 3813-3 (38 cm). As discussed above, the

laminations must have been deposited in a very Sh011

time and should reflect an extremely short-term

climate variability. Very high sedimentation rates

in the overlying carbonate sediments may have

contributed to preserve the siliceous sections in both

cores. The almost monospecific layers consist of

the giant diatom Ethmodiscus rex (Rattray) Wise­

man and Hendey (C.B. Lange, pers. comm.), the

largest solitary (not chain-fol1ning) diatom known,

reaching diameters of2-3 mm (Wiseman and Hen­

dey, 1953; Round et a1., 1990). They are entirely

uncommon at these latitudes as Ethmodiscus rex is

found in plankton primarily in equatorial regions

in a range of temperature frOl11 19° to 29.5°C

(Lisitzin, 1996). Recently an unusual diatom ooze

layer was detected in core GeoB 5112-4 recovered

on R.V. METEOR cruise M41/3 at 23°49,5'S

16°15,5' W from a water depth of3842 m (Pätzold

et a1. , 1999). C01Telating the magnetic susceptibility

record of this core to the SUSAS stack proves that

the diatoms were deposited at the same time as in

the two SUSAS cores (Frederichs et a1. , 1999b).

Thick ooze deposits of this diatom were mainly

64 Chapter 4

Age [kar528 530 532 534 536 538 540 542 544 546 548 550 552 554 556

306 307 308 309 310 311 312 313 314 315 316 317 318 319Core Position [ern]

Figure 9. Terminal MPT event documented in a sharp colourtransition at unbioturbated finely laminated layers in coreGeoB 3812-1.

reported from equatorial regions ofthe world oeeans

(e.g., Gardner and Burekle, 1975; Mikkelsen, 1977;

StabeIl, 1986; Lisitzin, 1996). Even there, their

oeeurrenee is puzzling, beeause EthnlOdiscus rex

very rarely oceurs in plankton sampIes (not more

than 0.5 eells/m3; Lisitzin (1996)). However, enig­

matie deep populations have been observed in the

Paeific Oeean (Villareal, 1993). Several hypothesis

have been proposed to clarify the paradoxieal

'Ethlnodiscus rex problem' (Gardner and Burekle,

1975), including differential dissolution, foeusing

by bottom eUlTents and intense productivity blooms.

Results from the Joint Global Oeean Flux Study

(JGOFS), emphasize the potentially important role

of oeeanie frontal zones for the rapid aeeumulation

of diatom biomass (Yoder et a1., 1994). They indi­

eate dramatic biological responses to circulation and

mixing processes assoeiated with open-ocean fron­

tal systems separating cold from wann waters and

gave rise to the idea that thick diatom layers may

be deposited in such environments.

Pieking up this suggestion and taking into account

the OCCUlTence of the tel111inal event at the end of

the MPT interim state of strongly reduced NADW

flux we believe it to result from elemental changes

in the global ocean circulation system. Possibly an

oceanic frontal zone shifted into the subtropical

South Atlantic for a short time and thus initiated

the massive diatom growth. The diversified terminal

MPT events at around 530 ka documented in South

Atlantie Oeean sediments might weIl be related to

other paleoc1imatic 'puzzles'. Recently, Rossignol­

Strick et a1. (1998) found an anomalous sapropel

layer in the MeditelTanean Sea dated 528-525 ka

and interpreted it to result from a 'massive odd 1110n­

soon'. Previously Bassinot et a1. (1 994b) reportecl

extremely high and low 0180 values during isotope

stages 13.2 and 13.3, respectively, in a giant piston

core from the equatoriallndian Ocean. Analogously

to the sapropel layer the unusually depleted value

at stage 13.3 ean be interpreted as a regional effeet

resulting from low surface-water salinity due to

heavy monsoonal fluvial discharge (Rossignol­

Strick et a1. , 1998). But insolation eonditions would

not predict heavy monsoon rainfall over Afriea and

Asia during that time. However, also a terrestrial

climate proxy indieates an extraordinary warm

climate. Chinese loess sequences and interbedded

paleosols are known to reeord paleoc1imate, as loess

was deposited in a cold, dry climate, while soils

formed during wet and mild episodes. Paleosol S5,

which can be related to oxygen isotope stages 15­

13, is by far the thickest and most weathered 0 f the

whole sequence (An and Wei, 1980). At its top, near

stage 13, the highest magnetic susceptibilities of

the past 2500 kyr (Kukla et a1., 1988; Heller et a1.,

1991; B10emenda1 et a1. , 1995) imp1y an extremely

warm and humid climate in Asia, although oxygen

isotope records identify this stage as one ofthe least

expressed interglacials during the late Pleistocene.

Similarly as the MPT interim state, the terminal

event can not be explained by astronomieal forcing

nor is it documented in global iee-volume records.

Conclusions

The rock magnetic view at sediment sequences from

the oligotrophie subtropical South Atlantic uneovers

several new and decisive aspects of the mid-

Cyc1es, trends and events of P1eistocene sedimentation 65

P1eistocene climate transition (MPT). The ana1yses

are based on high-resolution magnetic age models

which were built despite very 10w sedimentation

rates for aseries oftwe1ve sediment cores by orbital

tuning of their magnetic susceptibility records.

Located on the submarine ridges in water depths

affected alternate1y by NADW and LCDW these

sediment sequences have recorded ocean history

over the past 1500 kyr in a region very important in

the framework ofthe global thermohaline circu1a­

tion. The good overall cOlTe1ation of all records let

us to compute the SUSAS stack.

Cyclic patterns, which partly were used for the

magnetic cyc10stratigraphy, essentially reflect or­

bitally forced changes in deep water chemistry. In

the course of the MPT they document the well­

known change from variations with aperiod of

41 kyr to the 1ate P1eistocene 100 kyr ice age

rhythm. Compared to the driving functions in an

evo1utionm'y spectra1 analysis an a1ternating ex­

change of spectra1 energy between the 41 and

100 kyr bands is implied. In accordance with the

suggestions ofRuddiman et al. (1989) and Joyce et

al. (1990) residues between the SUSAS stack and

an adapted ETP record characterize the c1imate

system as less sensitive to obliquity changes during

the 1ate P1eistocene. According to our analysis the

shift towards strong1y reduced response to obliquity

forcing takes place at about 650 ka and is synchro­

nous with the onset of 100 kyr cyc1icity in our data

and as weIl in global ice volume (e.g., Ruddiman et

al., 1989; Mude1see and Schulz, 1997). This result

substantiates thresho1d models, which mimic the

occurrence ofnear-1 00 kyr cyc1es in the late Quater­

nary by introducing a decreasing atmospheric pC02

level (e.g., Raymo, 1997; Paillard, 1998). The

residual analysis also documents reduced response

in the 41 kyr band coinciding with the premature

occurrence ofa near-1 00 kyr cyc1e at about 1150 ka.

Being previously reported from severa18 180 records

(e.g., Mude1see and Stattegger, 1997), this incident

involves global ice volume.

In addition to features which mimic oxygen iso­

topes, the rock magnetic proxy unveils a MPT in­

terim state of reduced carbonate deposition obvi­

ous1y not directly linked to changes in global icc

vo1ume. Bounded by transitions lasting fi'om 1000

to 920 ka and from 640 to 500 ka this interval pre­

cise1y fills the time lag between the first occurrence

of 1arger glacia1 ice shie1ds and the onset of near­

100 kyr cyc1icity (e.g., Mudelsee and Schulz, 1997).

Our trend ana1yses implies that the outstanding fea­

ture ofthis episode is enhanced influence of south­

ern source deep waters. Support for this hypothesis

comes from ana1yses of 813C records imp1ying a

consistent reduced influence ofNADW during this

time interval (Raymo et al., 1997).

At the end of the MPT interim state a terminal

event is documented in severa1 SUSAS cores. It is

believed to be due to strong changes in the ocean

circulation system linked to the end of the MPT

interim state and the beginning of the late Pleisto­

cene 100 kyr state. A global character of this event

is suggested as it occurs coincidentally with other

unusual c1imate excursions which point towards

extremely warm and humid c1imates in Asia and

Africa (e.g., Kukla et al. , 1988; Bassinot et al. ,

1994b; Rossignol-Strick et al., 1998).

Acknowledgements

We thank T. Bickert for kindly providing CaC03data

ofGeoB 1211-1 and the unpublished 813C record of

ODP Site 806. This study was funded by the Deutsche

Forschungsgemeinschaft (Sonderforschungsbereich

261 at Bremen University, contributionNo. XXX). F.S.

was supported by the Deutsche Forschungsgemein­

schaft in the framework ofGraduiertenkolleg 221.

66 Chapter 4

Referellces

Adelseck, e.G. Jr., Reeent ,md late Pleistoeene sediments from

the castern equatorial Paeifie Oeean: sedimentation and

dissolution, Ph.D. thesis, Univ. of Calif., San Diego,

192 pp, 1977.

An, Z.S. and L.Y. Wei, The fifth layer paleosol in the Lishi

Loess and their paleoelimatie signifieanee, Acta Pcdolo­

gica Siuim, 17. 1-10, 1980.

Balsam, W.L. ami F. W. I\1eCoy, Jl'., Atlantie sediments:

glaeial/interglaeial eomparisons, Palcoceauogmph.1', 2.

531-542,1987.

Bassinot, F.e., L. Beaufort L, E. Vineent, L.D. Labeyrie. F.

Rostek, P.J. Miiller, X. Quidelleur and Y. Laneelot, Coarse

fraetion fluetuations in pelagie carbonate sediments [rom

the tropieal Indian Oeean: a I 500-kyr reeord ofearbonate

dissolution, Palcoccauogm/Jhy, 9, 579-600, 1994a.

Bassinot, F.e., L.D. Labeyrie, E. Vineent, X. Quiclelleur, N.J.

Shaekleton ami Y. Laneelot. The astronomieal theory of

elimate ancl the age of the Brunhes-Matuyama magnetie

reversa!, Earth Plane!. Sci. Lct!., 126, 91-108, 1994b.

Berger, A. ami M.F. Loutre, Insolation va lues for the elimate

ofthe last 10 million years, QIIa! Sei Rc\', 10,297-317,

1991.

Berger, A. ancl M.F. Loutre, Astronomieal solutions for

paleoclimate stuclies over the last 3 million years, Earth

Plauct. Sei. Lett., 111,369-382,1992.

Berger, W.H. and G. Weftr, Klimageschichte aus Tiefsee­

sedimenten, NatllrJvissensc/wjtcn, 79, 541-550, 1992.

Berger, W.H. and E. Jansen, Mid-Pleistoeene elimate shift

the Nansen eonneetion, in The Polar Oceans aud their

Rolc in Shapiug thc Glohal Envil"Onlllcut, Geophys.

Monogr., 84, 295-311, AGU, Washington, 1994.

Berger, W.H., M.K. Yasuda, T. Biekert, G. Weter and T.

Takayama, Quaternary timc seale for the Ontong Java

plateau: Milankoviteh tcmplate for Ocean Drilling Pro­

gram Site 806, Gcology, 22, 463-467, 1994.

Berger, W.H. ancl G. Weier, Expeditions into the past: pa­

leoceanographic studies in the South Atlantic, in Thc South

Atlantic: Present a/1(l Past Circulation edited by G. Wefer,

W.H. Berger, G. Siedler ancl D.J. Webb, Springer Verlag,

Berlin, pp 363-410, 1996.

Berger, W.H., T. Bickert, M.K. Yasuda ancl G. Wefer,

Reeonstruetion of atmospherie CO 2 from iee-eore data and

the deep-sea reeord ofOntong Java plateau: the Milanko­

viteh ehron, Gco!. Ruudschau, 85,466-495, 1996.

Biekert, '1'., Rekonstruktion der spätquartärel1 Bodenwasser­

zirkulation im östlichen Südatlal1tik über stabile Isotope

bentl1iseher Foraminiferen, Ph.D. thesis, Bel'. FB Geow.

Univ. Bremcn, 27, 205 pp, 1992.

Bickert, T. and G. Wefer, Late Quaternat·y deep water

cireulatiol1 in the South Atlantic: reeol1struction from

carbonate dissolution and bel1thic stable isotopes, in The

South Atlantie: Presentaud Past Cireu/ariou edited by G.

Wefer, W.H. Berger, G. Siedler and D.J. Webb, Springer

Verlag, Berlin, pp 599-620, 1996.

Birchfield, G.E. and M. Ghil. Climate evolution in the Plioeene

and Pleistocene from marine-sediment reeords ami sinlllla­

tions: interna I variability versus orbItal foreing, J G('()­

pln's. Res, 08,10385-10399,1993.

Bleil, U. and eruise partieipants, Report amI prcliminary results

of METEOR eruise 29/2, MonteVIdeo Rio de J,lnciro,

15.7.1994· 8.8. I 994. Ba. FIJ Gc01\' Um1' Bremcn.59.

153 pp, 1994.

Bloemencla!, .1., 13. Lamb ancl .I. King. Paleoenvironmcnt:i1

implications 01' rock-magnetic properties 01' Latc QLwter­

nary sediment COlTS from the eastern equatorial Atlantic.

Pal('()eclluography, 3, 61-87,1988.

Bloemendal, J. and P. deMenoeal, Evidence for a change in

the periodicity oftropieal climate cycles at 2.4 Myr from

whole-eore magnetie suseeptibility measuremel1ts, Nature

342, 897-900, 1989.

Bloemenclal, .1., .l.W. King, L. Tauxe ami J.-P. Valet, Rock

magnetic stratigraphy 01' leg 108 sites 658, 659, MI, amI

665, eastern tropieal Atlantic, Pme" ODP, Sei. Reslllrs,

108,415-428, 1989.

Bloemendal, J., X.M. Lill and T.e. Rolph, CorrelatioJ1 of the

magnetic sllseeptibility stratigraphy 01' Chinese loess and

the marinc oxygen isotope record: chronological ami

palaeoclimatic implications, Earth Planct. Sci. Lct!, 131,

371-380,1995.

Breitzke, M. Elastische Wellenausbreitung in marinen Sedi­

menten - neue Entwicklungen der Ultraschall Sediment­

physik lind Sedimentechographie. Bel'. FB Geo\\'. Uni1'.

Bremen, 104,298 pp, 1997.

Chi, .I. ancl J. Mienert, Linking physieal property records 01'

Quaternary sediments to Heinrich evenls, /l1ar. Geolog1'.

131,57-73,1996.

Clark, P.U. and D. Pollard, Origin of the middle Pleistocene

transition by ice sheet erosion ofregolith, Paleoceanogra­

phy, 13, 1-9, 1998.

DeBlonde, G. ancl W.R. Peltier, A one-dimensional model 01'

continental ice volume fluctuations through the Pleisto­

cene: implications for the origin of the Mid-Pleistoeene

climate transition, J Clim., 4, 3 I 8-344, 199 I.

Frederiehs, 'rh., U. Bleil, K. Däumler, T. von Dobeneek ancl

A. Schmidt, The magnetic vievv on the marine paleoen­

vironment: parameters, techniques and potentials of rock

magnetic studies as a key to plaeoclimatie and paleoceno­

graphie ehanges, in Use 01' Proxies in Paleoecunogra­

phy: Examplesji'om the SOllth Atlantie edited by Fischer,

G. and G. Wefer, Springer-Verlag Berlin Heidelberg, in

press, 1999a.

Frederichs, Th., K. Fabian and J. Funk, Physical properties

studies, in Report und prelimin([/)' re.l'lllt.l' 01' METEOR

cmise M4/ /3, Vitöria - Salvador de Buhiu, /8.4. /998 -

Cyclcs, trends and events of Pleistocene sedimentation 67

! 5.5.! 998, Ber. F/3 Gco\\'. Ulli,'. Bremell, in press, 1999b.

Gardner, ,LV. and LH. Burekle, tIpper Pleistoeene Etll­

lI/odiscus rex oozes ti'om the castern equatorial Atlantie,

Micropo!eolltol. 21, 236-242, 1975.

GingeIe, F.X., F. Schmieder, T von Dobeneek, R. Petsehiek

and C. Rühlemann, Terrigenous t1ux in the Rio Grande

Rise area during the last 1500 ka: evidenee of deepwater

adveetion or rapid response to eontinental rainfall

patterns?, Paleoceallogmpln', in press, 1999.

Grlitzner, l, F.C Bassinot and J. Mienert, High-resolution

eompressional-wave veloeity measurements in Pleistoeene

sediments oethe Ceara Rise (western equatorial Atlantie):

implieations for orbital driven sedimentary eycles, Pmc.

GDP, Sci Rcsults, 154, 135-149, 1997.

Hays, lD., l Il11brie and NJ. Shaekleton, Variations in Earth's

orbit: pace-makel' ofthe iee ages, Sciellce, 194, 1121-1132,

1976.

HaITis, S.E., AC. Mix and T King, Biogenie and terrigenous

sedimentation at Ceara Rise, western tropieal Atlantie,

supports P1ioeene-Pleistoeene deep-water linkage between

hemispheres. Pl'Oc. GDP. Sci. Results, 154,331-345, ]997.

Heller, F., X Liu, T. Liu and T. Xu, Magnetie suseeptibility

01' loess in China, Earlll Plalle!. Sei Let!, 1OJ, 301-3] 0,

1991.

Imbrie, l, Mcasuring the gain ofthe climate system 's response

to Milankoviteh foreing in the preeession and obliquity

bands, in LOllg-tel'lll clill/atie variations -- data amlmode/­

/ing, Springer, Berlin, pp 403-410, 1992.

Imbrie, l, lD. Hays, D.G. Martinson, A. McIntyre, A.C

Mix, The orbital theory of Pleistoeene elimate: support

üom a revised chronology of the marine 0180 reeord, in

Milankovitcll (md Clill/ate Part!, edited by A. Berger, l

Imbrie, l Hays, G. Kukla, B. Saltzman, Reidel, Dordreeht,

pp 269-305, 1984.

Imbrie, l, A. Berger, E.A. Boyle, S.c. Clemens, A Duffy,

W.R. Howard, G. Kukla, l Kutzbaeh, D.G. Martinson,

A. McIntyre, A.C Mix, B. Molfino, II Morley, LC

Peterson, N.G. Pisias, W.L. Prell, M.E. Raymo, N.J.

Shaekleton and lR. Toggweiler, On the strueture and

origin of major glaeiation eycles 1. Linear responses to

Milankoviteh foreing, Paleoceanograplly, 7, 701-738,

1992.

Imbrie, l, ABerger, E.A. Boyle, S.C Clemens, A Duffy,

W.R. Howard, G, Kukla, l Kutzbaeh, D.G. Martinson,

A. Mclntyre, AC Mix, B. Molfino, J.J. Morley, L.C

Peterson, N.G. Pisias, W.L. Prell, M.E. Raymo, N.J.

Shaekleton and lR. Toggweiler, On the strueture and

origin of major glaeiation eycles 2. The 100,000-year

eycle, Paleoceanogmplzy, 8, 699-735, 1993.

Johnson, D.A., M. Ledbetter and L. Burekle, Vema Channel

paleo-oeeanography: Pleistoeene dissolution eycles and

episodie bottom water t1ow, Mal'. Geology, 23, 1-33, 1977.

Joyee, lE., L.R.C Tja1sma, lM. Prutzman, High resolution

planktie stable isotope reeord and speetral analysis for the

last 5.35 m.y.: Oeean Drilling Progral11 Site 625,Ilortheast

gulf ofMexieo, Pa!eoccallogl'llplly, 5, 507-529, 19()0.

Kllkla, G., F. Heller, X.M. Liu, T.C Xu, T.S. Liu alld Z.S.

An, Pleistoeene elimates in China da ted by magnetie

suseeptibility, Geolog)', 16, 81] -814, 1988.

Lisitzin, A.P., Oeeanie sedimentation: Jithology and gco­

ehemistry, AGU, Washington D.C, 400 pp, 19%.

Maaseh, K.A., Statistieal deteetion 01' the micl-P!elstoeene

transition, C/im. D1'II., 2, 133-143, 1988.

Martinson, D.G., N.G. Pisias, J.D. Hays, J. Imbric, r.c.Moore, N..!. Shaekleton, Age dating and the orbital theory

of the iee ages: development 01' a high-resolution 0 (0

300,000-year ehronostratigraphy, QUa!. Rcs .. 27, 1-29,

1987.

Mead, G.A. and L Tallxe, Oligoeene paleoeeanography of

the South Atlantie: paleoelimatie implieations ofsediment

aeeumulation rates and magnetie suseeptibility measurc­

ments, PalCOCC(//lOgmpllv, I, 273-284, 1986.

Mikkelsen, N., On the origin of Ethrnodiscus OOl:e, Mal'

Ahcl'Of!alcolllol., 2, 35-46, 1977.

Mudelsee, M. and M. Sehulz, The Mid-P1eistoeene climate

transition: onset of 100 ka eyele lags iee volume budd-up

by 280 ka, Eartlz Planet. Sci. Lett, 151. 117··123, 1997.

MudeIsce, M. and K. Stattegger, Exploring the strueture 01'

the mid-Pleistoeene revolution with advaneedmethods of

time-series analysis, Geol. Rundschau, 86,499-511,1997.

Muller, R.A. and G ..!. MaeDonald, Simultaneous presenee of

orbital inelination and eeeentrieity in proxy elimate reeords

from Oeean Drilling Program Site 806, Geolog)', 25. 3-6,

1997.

Mwenifumbo, C.l, .!.P. Blangy, Shor(-(erm speetral analysis

of downhole logging measurements from Site 704, Proc.

GDP, Sei. Results, 114, 577-585, 1991.

Paillard, D., The timing of Pleistoecne glaeiations from a

simple multiple-state elimate model, Nature, 391, 1998.

Park, l, S.L. D'Hondt, l W. King and C. Gibson, Late Creta­

eeous preeessional eycles in double time: a warm-Earth

Milankoviteh response, Science, 261, 1431-1434, 1993.

Park, .!. anel K,A. Maaseh, Plio-Pleistoeene time evolution of

the 1OO-kyr eycle in marine paleoclimate reeords, J. Geo­

plzys. Res., 98, 447-461,1993.

Pätzold, l und Fahrtteilnehmer, Bericht und erste Ergebnisse

über die METEOR-Fahrt MI5/2, Rio de Janeiro - Vitoria,

18.1. -7.2.1991, Bel'. PB Geow. Univ. Bremen, 17,46 pp,

1993.

Pätzold, land eruise partieipants, Report and preliminary

resll1ts of METEOR eruise M41/3, Vit6ria - Sa1vador de

Bahia, 18.4.1998 - 15.5.1998, Bel'. PB Geow. Univ.

Bremen, in press, 1999.

Pestiaux, P. and A. Berger, An optimal approach to the speetral

eharaeteristies of deep-sea elimatie reeords, in Milanko­

vitell and Climate Part! edited by A. Berger, l Imbrie, l

68 Chapter 4

Hays, G. Kukla ancl B. Saltzman, Reidel, Dordreeht, pp

417-445,1984.

Petersen, N., T von Dobeneek and H. Vali, Fossil baeterial

magnetite in deep-sea sediments from the South Atlantie

Oeean, Nature, 320, 611-615, 1986.

Pisias, N.G. and T.e. Moore, The evolution of Pleistocene

climate: a time series approach, Earth. Planet. Sei. Lett.,

52,450-458, 1981.

Prell, W.L., Oxygen and earbon isotope stratigraphy for the

Quaternary of Hole 502B: evidence tor two modes of

isotopie variability, Initial Rep., DSDP, 68,455-464, 1982.

Raymo, M.E., The timing of major climate transitions, Pa­

leoeeanography, 12,577-585,1997.

Raymo, M.E., Glaeial puzzles, Science, 281, 1467-1468, 1998.

Raymo, M.E., W.F. Ruddiman, J. Backman, B.M. Clement

and D.G. Martinson, Late Pliocene variation in northern

hemisphere ice sheets and North Atlantic deep water

cireulation, Paleoeeanography, 4, 413-446, 1989.

Raymo, M.E., D.W. Oppo and W. Curry, The mid-Pleistocene

elimate transition: a deep-sea carbon isotopie perspeetive,

Paleoceanography, 12, 546-559, 1997.

Robinson, S.G., Applications for whole-core magnetie suscep­

tibility measurements of deep-sea sediments, Proe. GDP,

Sei. Results, 115,737-771,1990.

Robinson, S.G., M.A. Maslin andl.N. McCave, Magnetic sus­

eeptibility variations in Upper Pleistocene deep-sea sedi­

ments ofthe NE Atlantic: implications for ice rafting ancl

paleoeireulation at the last glaeial maximum, Paleoeean­

ography, 10, 221-250, 1995.

Rossignol-Strick, M., M. Paterne, F.e. Bassinot, K.-e. Emeis

and G.I. DeLange, An unusual mid-Pleistocene mon8Oon

period over Afriea and Asia, Nature, 392, 269-272,1998.

Round, F.E., R.M. Crawtord and D.G. Mann, The diatoms:

biology and morphology ofthe genera, Cambridge Univ.

Press, Cambridge, 747 pp, 1990.

Ruddiman, W.F., M.E. Raymo, D.G. Martinson, B.M.

Clement and I. Backman, Pleistocene evolution: northern

hemisphere ice sheets and North Atlantic ocean, Pa­

leoeeanography, 4, 353-412, 1989.

Saltzman, B. and M.Y. Verbitsky, Multiple instabilities and

modes of glacial rhythmicity in the Plio-Pleistocene: a

general theory of late Cenozoic climatic change, Clim.

Dyn., 9, 1-15, 1993.

Schmidt, A., T. von Dobeneck and U. Bleil, Magnetic

characterization of Holocene sedimentation in the South

Atlantic, submitted.

Schulz, H.D. und Fahrtteilnehmer, Bericht und erste

Ergebnisse über die METEOR-Fahrt M20/2, Abidjan ­

Dakar, 27 .12.1991 - 3.2.1992, Ber. FB Geow Univ.

Bremen, 25, 173 pp, 1992.

Shaekleton, N.J., A. Berger, W.R. Peltier, An alternative

astronomical calibration ofthe lower Pleistocene timescale

based on ODP Site 677, Trans. Royal Soc. Edinburgh:

Earth Sei, 81, 251-261,1990.

StabeIl, B., Variations ofdiatom flux in the eastern equatorial

Atlantic during the last 400,000 years (METEOR cores

13519 and 13521), Mar. Geol., 72, 305-323,1986.

Thiessen, W., Magnetische Eigenschaften des östlichen Süd­

atlantiks und ihre paläozeanographisehe Relevanz, Ph.D.

thesis, Ber. FBGeow. Univ. Bremen, 41, 170 pp, ]993.

Thompson, R. and F. Oldfield, Environl11ental l11agnetisl11,

Allen and Unwin, London, 227 pp, 1986.

'riedel11ann, R., M. Sarnthein and N.l Shackleton, Astronomie

til11escale for the Pliocene Atlantie 8 130 and dust tlux

records ofOcean Drilling Program site 659, Paleoccano­

graph)!, 9, 619-638,1994.

Vali, H., O. Förster, G. Al11arantidis and N. Petersen, Magncto­

tactic bacteria and their magnetofossils in sediments, Fllrth

Planet. Sei. Lett, 86, 389-400,1987.

Villareal, TA., Abundance ofthe giant diatol11 Ethmodisells

in the Southern Atlantic Oeean and the eentral Pacific gyrc,

Diatom Res, 8, 171-177, 1993.

Volat, lL., L. Pastourct and e. Grazzini, Dissolution and car­

bonate fluctuations in Pleistocene deep sea cores: a re­

view, Mar. Geology, 34, 1-28, 1980.

von Dobeneck, T and F. Schmiede!', Using rock magnetic

proxy records for orbital tuning and extended time series

analyses into the super- and sub-Milankovitch bands, in

Uye ofProxics in Paleoecanography: Examplesfi'o/ll the

SOllth Atlantie edited by Fischer, G. and G. Wefer, Sprin­

ger-Verlag Berlin Heidelberg, in press, 1999.

Wefer, G. und Fahrtteilnehmer, Bericht über die METEOR

Fahrt M 6/6, Libreville - Las Palmas, 16.2.1988 ­

24.3.1988, Ber. FB Geow. Univ. Bremen, 3, 97 pp, 1988.

Wefer, G. und Fahrtteilnehmer, Bericht über die METEOR

Fahrt M12/l Kapstadt - Funchal, 13.3.1990 - 14.4.1990,

Ber. FB Geow. Univ. Brelnen, 11,66 pp, 1990.

Wefer, G. and cmise participants, Report and prelil11inary

results of METEOR cmise M34/3, Walvisbay - Recife,

21.2 - 17.3.1996, Ber. FB Geow. Univ. Bremen, 59,

153 pp, 1996.

Wiseman, J.D.H. and N.!. Hendey, The significance and

diatom content of a deep-sea floor sampie from the

neighbourhood of the greatest oceanic depth, Deep-sea

Res, 1,47-59,1953.

Yiou, P., C. Genthon, M. Ghil, J. Jouzel, H. Le Treut, I.M.

Barnola, C. Lorius and Y.N. Korotkeviteh, High-frequency

paleovariability in climate and CO2 levels from Vostoek

ice eore reeords, 1. Geophys. Res., 96, 20365-20378,1991.

Yoder, lA., S.G. Ackleson, R.T. Barber, P. Flament and

W.M. Balch, A line in the sea, Nature, 371,689-692, 1994.

69

Terrigenous flux in the Rio Grande Rise area during the past 1500 ka:Evidence 01' deepwater advection 01' rapid response to continentalrainfall patterns? (Paleoceanograplzy, 14, 84-95, 1999)

Franz X. CJingele, I Frank Schmieder,2 Tilo von Dobeneck,2 Rainer Petschick,3

amI Carsten Rühlemann2

Abstract. Sud~1ce sediment sampies and three gravity COlTS from the eastern terrace of the Vema Channc1, fhe westernnank of the Rio Grande Rise, and the Brazilian continental slope were investigated tor physical properties, grain size, anclclay mineral composition. Discharge of the Rio Doce is responsible tor kaolinite enrichments on the slopc south of 2()"and at intermediate depths of the Rio Grande Rise. The long-distance advection of kaolinite with North Atlantic DeepWater from lower latitudes is of minor importance as evidenced by low kaolinite/chlorite ratios on the Mid-AtlanticRidge. Cyclic variations of kaolinite/chlorite ratios in all our cores, \Vith maxima in interglacials, are attributcd tn 100v­and high-latitude forcing of paleoclimate on the Brazilian mainland and the related discharge of the Rio Doce. A long­term trend to\Vard more arid and "glacial" conditions from 1500 ka to present is superimposed on the glacial-intcrglacialcyclicity.

1. Introduction

The Rio Crande Risc!Vema Channel/Hunter Channel regionis a key area in the western South Atlantie to study spatial asweil as temporal variations in the history of deepwater masses.Southern source bottom water passes from the AI'gentine to theGrazil Gasin via the Vema and Hunter Channels. Above, south­ward flowing North Atlantie Deep Water (NADW) is recordedfrom 4000 to 2000 m water depth, overlain by Upper Cireum­polar Water (UCPW) and Antaretie Intermediate Water (AAIW)[PelerSOI/ al/(! SlrallIIlW, 1(90). The Rio Crande Rise (RCR),wh ich reaches water depths of 800 m, offcrs the opportunity tosampie sediments situated in different water masses and look fortracers of changes in thermohaline circulation.

Numerous studies have focussed in partiClilar on variations inthe propagation of Antarctic Gottom Water (AAGW) andNADW in the South Atlantic using various tracers from benthicforaminifera assemblages [lI/fackel/seil ef al., 1993; SchIIlied! alld

MackeIlseIl, 1997], Cd/Ca ratios in benthic foraminifera [Boy!e,

1988, 1994; Oppo al/d Rosellf!w!, 1994], carbon-13 eomposition[Omy ef a!., 1988; Dup!essy ef al., 1988; IlIackellsell cf al.,

1994]. diatoms [Jolles alld .Johl/soll, 1984], grain size [Masse cf

al., 1994]. and elay minerals [Biseaye, 1965, (,halll!ey, 1975,JOlles, 1984; Dieklllalll/ cf al., 1996).

On the basis of their strong latitudinal and reciprocal distri­bution patterns in surface sediments of the South Atlantie, kao­linite and chlorite were inferred to be useful tracers ofthe majordeepwater masses in the vicinity of the RCR. An advection of

I Baltic SeH Research Institute, Rostock-Warncmucnde, Germany.

2 University 01' Bremcn, Bremen, Germany.

.\ University 01' Frankfurt, Frankfurt, Germany.

Copyright 1999 by the Allleriean Ceophysieal Union.

Paper number I998PA9000 12.0883-8305/99/1 998PA900012$12.00

chlorite with southern source dcepwatcr (AAI3W) was found byIiiscaye [1965] and confirmed by later studies [Jolles. I()84;Pelschick cl al., I99CJ]. The OCCUITence 01' 'I kaolinite maximumat intermediate depths (above 4000 m) was attrlbuted to advec­tion 01' this mineral by NAD\V [ClwlII!e)" 1975). The propaga­tion of kaolinite with NADW is 'I well-documented feature inmany cores 1'rom the eastern South Atlantie [f)ie!w/{/1I11 cl 01.,1996].

;\ Iternatively, JOlles [1984] suggested an isopycnal transportmodel for the kaolinite enriehment on the RCR. On the basis 01'the assumption that the suspension load 01' the Nf\DW in thewestern South Atlantic is too small to sustain a subslantialenrichment (clearwater minimum), kaolinite input by the RioDoee, southward transport with the I3razil current, deposition onthe Sao Paulo (Santos) Plateau, and resuspension and isopycnalflow to the RCR was proposcd. These different interpretationshave implieations on the evaluation 01' temporal changes 01'kaolinite content in sediment cores, which record either oeeanicprocesses such as NADW fluctuatlons, sea level ehanges, orvariations in fluvial discharge.

We examined 51 surface sampies from two transeets toevaluate potential sources and reeent propagation 01' clayminerals (Figure I). One profile sampies the continental slopeoff the Rio Doce. The second transect runs from the slope acrossthe Santos Plateau, the western and castern flank 01' the RCR, tothe Mid-Atlantic Ridge (MAR).

The temporal variation in the supply 01' clay minerals c1uringthe past 200 kyr was investigated in sediment eores from theslope (CeoB 2110) and the castern terrace of the Vellla Channel(CeoB 2822). A complete reeord of clay mineral supplycovering the past 1500 kyr was found in core GeoB 2821. Thiscore is situated in a key position on the western flank of theRCR just above the present AABW/NADW boundary. Cross­speetral analysis on the kaolinite/chlorite ratio of this eore wascarried out to deterllline phase relationships to orbital cycles andeompare phase angles 01' related paleoeeanographic and paleo­climatie proxies.

70 Chapter 5

-45' _40' _35' _30" _25" _20' -15'

kaolinite/chlorite 0 < 1 c. 2-5 • > 10

ratio -ll. 1-2 ~ 5-10' .... _-

E

Argentine Basin

_20"

-25~

-30'

-35'

Souree GEBCO.

w

_45" _40"

Brazil Basin

-25" -20' _15'

-20'

-30"

MidoceanRidge

10 kaolinite/chlorite9 ratio8

_J _65

4

32

1

o

/

__ 8 _Santos Rio GrandePlateau Vema Rise

Channel HunterChannel

0

E2SL

"- 2<lJU

Qj 3ro;';

4

50"W 45"W 40"W 35"W 30"W 20"W 15-W

Figurc 1. Area 01' investigation. core locations, and kaol initeichloritc ratios 01' 5 I surface sampies in the investigation area. Depth contours afterGeneral Bathymetric Chart 01' the Oceans (GEBCO). Kaolinite/chlorite ratios are also projected on a hypothetical transect Ii'om the Santos Plateauacross the Rio Grande Risc (RGR) to the Hunter Channel and Micl-Atlantic Riclgc (MAR). Extrcmd<aolinite/chlorite values (>10) from sites clircctlyotT thc Rio Doce are cxcluclecl from the projection.

2. Material and Methods

Sediment surface sampies were recovered with giant boxcorer and multicorer during Meteor cruises M 15/2 [Pätzold etal., 1993J, M23/2 [Bleil et al., 1993J, M29/2 [Blei! et al., 1994J,and M34/3 [Iflerer et al., 1996]. Sample treatment and claymineral analyses are given in detail by Petschick et al. [1996].Sediment cores (Table 1) were retrieved with a gravity corer oncruises M23/2 (GeoB 2110-3/4) and M29/2 (GeoB 2821-1 andGeoB 2822-2). Core GeoB 21 10 from 3008 m water depth onthe continental slope above the Santos Plateau (Figure I) con-

sists 01' grey hemipelagic clays with varying amounts 01' car­bonate. Core GeoB 2822 from the eastern terrace 01' the VemaChannel (4267 m water depth) comprises terrigenous muds withsmall amounts 01' calcareous nannofossils and foraminifera.Increasing amounts 01' carbonate characterize core GeoB 2821from the western flank 01' the RGR (3927 m water depth).

The cores were sampled at 5 cm intervals. Samples weretreated with 10% acetic acid to remove carbonate, sieved over a63 ~lm mesh, and split into >2 ~lm and <2 pm fractions by set­tling technique. The weight percentage 01' the sand fraction (>63pm) was negligiblc. Weight percentages 01' the silt (2-63 ~lm)

and clay fraction «2 ~lm) were used to compute a silt/clay ratio.

Terrigcnous flux in tbc Rio Grande Risc arca during tbc past 1500 ka 71

Table I. Location of Scdimcnt Cores

Waler Lalilllde, CrlliseCore Deplh. III Longillldc Rcport Agc Model

Geo13 3008 28°389'8. Bleil cl al. Bleil Cl al. [1993J and2110-3/4 45°312'W [1993J C. Rlihlclllann,

(lIllpliblishcd dala, 1(93)GeoB 3927 30°271 '8, Bleil cl a!. F. SchIllieder,2821-1 38°481 'W [1994J (lIllplihlished dala, 19(4)GcoB 42()7 300 14.3 's, Bleil cl a!. Bleil cl a!. r199412822-2 3,)°08.5 'W [1994J

The clay fraetion «2 pm) was analyzed by X-ray diffraetion(CoKu radiation) on oriented mounts for the four clay mineralgroups kaolinite, smectite, illite, and chlorite following standardproeedures given in detail by Pc/schick e/ af. [1996]. Theseprocedures assume that the four main elay mineral groups addup to 100% in the fraction <2 pm and involve the use of theweighting faetors of Biscayc [1965]. As a eonsequence ourabsolute percentages for the individual elay minerals cannot becompared to those of .lanes [1984], who used weighting faetorsof I-Iea/h (lild Pisias [1979]. However, general patterns of elaymineral distribution are similar.

Magnetic suseeptibility was measured on the arehive halvesof the eores at a I cm spacing using a Bartington InstrumentsM,S,2.C loop sensor. Lomb-Scargle Fourier transform [Loll/h,I97G; Scargfc, 1982, 1989] embedded in the Spectrum program[Schuh ami S/al/eggcr, 1997] was used for all spectral analysis,This speetral estimation method ean be direetly applied tounevenly spaced geologieal times series. Cross-spectral analysisfor GeoB 2821 was performed using Welch' s [] 967] overlappedsegment average (WOSA) procedure to reduce spectral peaksoriginating fl'om random fluctuations.

2.1. StratigraphyofGeoB2110

A preliminary stratigraphic framework was established onshipboard counts of the planktonie foraminifera Gfoboro/alia

Il/enardii [Muli/za, 1993]. On the basis ofthe cyc]ic appearaneeof this speeies, Ericson (lild Woflin [1968] defined abiostratigraphic zonation scheme using a letter notation from Z(Holoeene) to Q in order of increasing ages. Zones Z to U ean bedirectly correlated to oxygen isotope stages with following agesat zone boundaries: ZN, ]2 ka; Y/X, 80 ka; X/W, ]30 ka; WIV,185 ka; and V/U, 370 ka, Though minor shifts in stageboundaries occur because of the poor resolution of the G,mel1ardii eounts, the preliminary stratigraphic fi-amework waseonfirmed for site GeoB 21] 0 by earbonate stratigraphy afterDall/u/h [1977]. Oxygen isotope measurements on planetie andbenthie foraminifera were earried out on a Finnegan massanalyzing technique (MAT) 25] mass speetrometer (Figure 2).The measurements of the planktonic foraminifera Gfobigeri­

noides sacculifcr were partly eomplicated by low sedimentationrates and enhaneed carbonate dissolution but enabledidentification of oxygen isotope stages 6,0, 6.2, 7,0, and 7, I.Supplementary measurements of Uvigeril1a spp. near isotopestage 4 clarified the stratigraphie position of stages 4.0 and 5.0.In addition to isotope stratigraphy, some eharaeteristie patternsof the high-resolution magnetic susceptibility record wereeorrelated to the SPECMAP stack (Jmbrie el al., 1984] andsupplied four more tie points (near oxygen isotope stages 2.2,

3,0, 5,5, and 6.4), Because of different phase lags of plünktonlcand benthic foraminifera and magnetic susceptibility, no phaseinformation can be dedueed from this combined age model.

2.2. Stratigraphy 01' GeoB 2821

The Cl. Il/enardii counts for core GeoB 2821 elid not producca reasonable pattern. This is due 10 the wide sampie spaeing ofthe shipboard sampies and the low sedimentation rates of thisCOlT. Since no 0180 stratigraphy is available an agc modelderived by orbital tuning of the susceptibility rccord \\as ustdhere, This cyclostratigraphy was established in the framework ofa stratigraphical synthesis of 12 sediment cores from the sub­tropical South Atlantic Ocean (subtropical South Atlantic sus­ceptibility (SUSAS) stack) [VOll Dobcneck and Sc/I/I/icder,

1998], based on cyclic and highly coherent magnetic susccpti­bility logs. The eorrelation of suseeptibility and carbonate con­tent is inverse as in many marine environments [e.g., Robinson,

1990]. This indicates that the rock magnetic signal is a result ofvarying dilution of the (terrigenous) magnetic sediment fractionby glacial-interglacial carbonate dissolution cyclcs. On the basisof a detailed stepwise alternating field demagnetizatlon of theNatural Remanent Magnetization (NRM), three paleomagneticage marks could be identified for GeoB 2821 (Figure 2), Asimple age model generated by linear interpolation of thesereversal ages alreacly shows spectral characteristics 01'Milankoviteh cyclicity (100 anel 41 kyr cyeles), suggestingclimatic forcing. Correlation to a precisely dated paleoclimaterecord was necessary to improve the age-depth relation prior toorbital tuning, The 8 13C record of South Pacifie Ocean DrillingProgram (ODP) Site 806 (T. Bickert, unpublished manuscrJpt,1998) exhibits a eontinuously high pattern similarity with thesusceptibility reeord of GeoB 2821 and was therefore used asage referenee. The preclominant 40 kyr cyclieity of susceptibilitywas then synchronized with astronomically calculated obliquityvariations [Bcrgcr al1d LOIl/rc, 1991] as a target curve. A phaseshift of -30 0 (-3.4 kyr) was found by cross-spectral analysis ofmagnetic susceptibility and 8 I go for one of the SUSAS records[VOll Dobel1eck alld Schmiedcl', 1998] (8 180 data by Bickerl

[1992]). This value corresponds to those ealeulated by Imbrie el

al. [1993] for early proxy responses in the SouthernHemisphere, As the SPECMAP stack [Imbrie cl af, 1984] isassumed to lag obliquity by 7.9 (7,4 - 8,2) kyr, a net lag of 4.5kyr was applied for the tuning ofmagnetie susceptibility.

From the observation that faintly visible precessional peaksin the speetra gradually sharpened at both refinements of the agemodel the obliquity-tuned primary susceptibility record wasfiltered (15 - 47 kyr) to extract obliquity and precession cyclesand matched to a eomposed target curve of normalized obliquityand preeession.

2.3. Stratigraphy of GeoB 2822

According to Cl. mCllardii stratigraphy [.lahll, 1994] coreGeoB 2822 reaches zone V of Ericsoll alld Wolfill [1968]. Noisotope data were available to confinn these ages. Pa]eomagneticmeasurements showed a uniform normal polarity throughout theeore, The more detailed glacial-interglaeial cycles revealed bythe susceptibility record justify minor shifts of assumed stageboundaries ZN (isotope stages 1/2) and WIV (617), As in theother cores, kaolinite/chlorite and siltlclay (grain size) ratiosgenerally correlate with the glacial-interglacial susceptibility re-

72 Chapter 5

G menaJ(iii Susceplibilily (10-3 SI) K/C ratio (1180 (',\JnVS PDß)G

SiltlClay

200

300

~

~ 400

mou

500' "\_

·----S~

600 ~-~

700

800'

1.5 1.0 0.5 0.0

G mcnardii Susccptibility (10.3 SI) I<lC ratio SiltlCI8y ChRM lnclinalion n CI:RM Declirwtiofl n Polar;ty

0 200 400 600 SOO 07 0.6 0.5 OA 03 02 0.5 10 1.5 2.0 1.2 10 0.8 0.6 ·90 -45 45 ·90 90 180 270o· -=:::::::= ~ ~ ~ ~~-=--=--

~5-:~-

-~-=--=tr;eOB 2821 =:----~ .-<;.' .~

100 L~ J "~~~

.~~->

-=~j:c- -~~------=

==-----~ ~~ J~ (200~~~ C~ -<---~ -~ -> t-------~ )~ /5 '"S- C '( '"K

.c

=~.3 ~ §

? ~, m~ =- ~~-~ '-.~ '----...__._-~ ..... -

~ ~3 -~:~-=~ ro

~ __~~J- E

====3(\J

--=.._.____-.'1,.,

l:::J

1iil~__~ L_-

72

~~

f~ ~~ ~ -=;,

~ ~ -=r~ 1, ~ -4

700 ~~

800 ----- ~

G. menBrdii Susceptibilily (10-3 SI} K/C ratio (S01001hed x 3) SiitfClay Polarity

1000.L..--------.l....--------.l....--------.l....-------~

900

10 20 30 40 06 05 DA 0.3 0.2 01 0.0 0.2 DA 0.6 0.8...,

2·4

-;;~ 5

---==-

1.0 14 1.2 1.0 0.8 0.6 04

Terrigenous flux in the Rio Grande Rise area during the past 1500 ka 73

Table 2. Data of Cross Spectral Analysis

r'CI'SIiS RCI'!'!'scd ODP677 (yISO-33±8(13) 094 -31±10(15)

V!'!'SIIS R!'I'CI'scd S'PFCMAP Slock ö·ISO-5±3(5) 0')') -32±3(5)

100 kyr- 1 Bandk °Phi

093 -46± 12

T M,kyr kyr

500 50

220 ]()

400 3()

388 2.0

1525 50

500 50

41"X7

23 kyr I Bandk °Phi

08'\

11±14

-42±10(13)

41 kyr I Bandk °Phi

005

087

086

-6±7

16±15

-202(3 )

096

099

087

0.96

Kaolinite/eh lorite(Geoß 2821-1), 0-500 kaMaximum DDVostok ice eore[JoIl2!'1!'I n/., 1')')41%,NADW index[Rn)'/IIo!'1 nl., I(NO]Kaol in ite/eh lorite43° S [Di!'k/llllllll !'I o/., 1996]

Kaolinite/eh lori tc(GcoB 2821-1 ). 0-1500 kaKaolinite/eh lorite(GeoB 2821-1 ), 0-500 ka

Abbreviatiolls are k, cohereney; °Phi, phasc angle with 80% eonfldenee interval (in parentheses 05'%);1: maximum age used in the calculation; :.\1,

sall1ple interval. The 8 "0 SPECMAP stack was taken from l/IIb!'ie ('/ o/. [10841, ami the 8"0 record Oecan Drilling Program Site (ODP) 677 flOmSiloeklelol1 el o/. [1090].

cord with the exception of stage 6 (Figure 2). Still, individualmaxima in the susceptibility rccord within stage 6 cOITespond tomaxima in the precessional index around 133, 160, and 185 ka.Minima can be related to preeessional minima and peaks in thekaolinite/chlorite record. Nevertheless, absolute values for theminima in susceptibility are rather low for 21 glacial stage. Sincethis is 21 featurc unparalleled in the othcr COlTS it cannot bc satis­factorily explained by 21 common mechanism. Further detailedpaleomagnctic investigations would be required to investigatcthis local, timc-restricted phenomena. The final age models ofall corcs were obtained by linear interpolation between strati­graphie tie points (Figure 3).

3. Results and DisCllssion

3.1. Surface Distribution

Sinee kaolinite and chlorite are the proxy clay mineralscommonly associated with propagation of NADW and AABWwe use the kaolinite/chlorite ratio to delineate the extension ofthe deepwater masscs as demonstrated in the approach ofPetschick et al. [1996]. Additional sources of kaolinite shouldrefiect in exceptionally high kaolinite/chlorite ratios. Kao­linite/chlorite ratios of the surface sampIes are depicted in Figureland reveal some characteristic features of sediment input anddistribution. Extremely high values between 10 and 80 arerecorded off the mouth of the Rio Doce and confirm the role ofthis river as an important kaolinite source as already stated by

Figlll'c 2. (opposite) Parameters lIsed for the stratigraphie classifieation01' the COlTS. Age models tor cores GeoB 21 10 and GeoB 2821 are basedon shipboard counts ofGloborola/in menardii. Oxygen isotopes in coreGeoB 2110 were measured on the planctie toraminifera Globigeril1oidessncclIlz!er and some bcnthie Uvigeril1a peregrina. The stratigraphieJi-amework 01' GeoB 2821 is based on paleomagnetic analyses and orbitaltuning 01' the magnctic suseeptibility reeord. Also depicted arekaolinite­chlorite ratios and grain size (ratio 01' silt/elay weight percentages),whieh nuctllate in a glacial-interglacial pattern. Note that the reeords 01'

suseeptibilty and grain size are reversed for bettel' graphic eorrclation.

DeMelo et al. [1975]. The highest values art": found on tht": sht":lfand upper slopt":. However, it is important to note that enoughriver suspension is carried through the water column to keepkaolinite/chlorite ratios above 10 at 4000 m watt":r depth. Evensouth of 25°, valut":s exceed 2 below 4000 m water depth. Below4000-4100 m in the Vema and Hunter Channt":ls and the abyssalplains of the Argentine and Brazil Basins, kaolinite/chloritt":ratios fluctuate from 0.5 to land confirm the concept of AABWas 21 source and carrier of chlorite [Biscaye 1965; .lalles, 1984;Petschick et al., 1996]. Some of the river suspension introducedby the Rio Doce is carrit":d south by the Brazil current.Scavenging and incorporation into fecal pellets are believed toremove most of the fine-grained material from the water COlUll1l1rather rapidly [Deuser et al., 1983]. Accul11ulations of kaolinite­rich sediments are found on the continental slopes off Cabo Frioand with decreasing values above and on the Sao Paulo andSantos Plateaus from 2500 to 3000 111, Since the rivers south ofCabo Frio contain relatively little kaolinite the Rio Doce has tobe regarded as the major source for these deposits [Jolles, 1984].

A characteristic enrichment of kaolinite was also found atintermediate depths (3000-4000 m) on the RGR by Chamley[1975] and JOlles [1984], with the maximum centered at 3300 to3600 m. Such an enrichment is confirmed by our sampIes, whichcover 21 depth range from 4500 to 2900 m water depth on thewestern flank of the RGR. A similar feature is observed Oll thetransect from the eastern slopt": of the RGR to the Hunter Chall­neI. East of the Hunter Channel toward the MAR kao­Iinite/ch lorite ratios are lower « I) at comparable water depths(Figure I). Only on the top of the ridge maximum values of 1.0are reached_ This has important implications concerning theorigin and propagation of kaolinite.

JOlles [1984] proposed two models to explain the kaoliniteenrichment at mid-depths of the RGR: (1) advective transport ofkaolinite from low latitudes with NADW and (2) advection ofkaolinite along isopyenals from resuspension of kaolinite-richdeposits on the Sao Paulo Plateau. Though he did not rule outthe NADW transport 1110del complt":tely, the isopycnal mixingmodel was bettel' suited to explain his observed elay mineral dis-

74 Chapter 5

Srnectite % IIlite% Kaolinite % Chlorite % Kaolinite/Chlorite ratio

6 82010o 10 20 30 404020 30403020

o ·-r·-~~--t~~~"';:!'~~~~~r~~~==>".bdT~~~~~~,_Iiiiiä""''''''":1

so

100 5

ro-"<lJ01<i

150

50

100

'"-"<lJ01<i

150

200

250-1...-------'--------'--------'--------'- -'

Terrigenous flux in tbe Rio Grande Rise area during the past 1500 ka 75

tribution. Kaolinite/chlorite ratios determined on our sedimentsurface sampie set yield arguments for both transport models.Thc symmetrieal nature 01' the kaolinitc cnrichment at middepths01' both sides 01' the RGR eould support thc eoncept of advectivetransport 01' kaolinite from low latitudes with NADW.Consequently, similar kaolinite/chlorite ratios should beexpected at comparable depths 01' the MAR. However, only aslight rise in kaolinite/chlorite values is recorded on the top,whieh may rcsult from minor adveetion 01' kaolinite from lowlatitudes. On the other hand, although the transport of kaoliniteto the western slope 01' the RGR along isopyenals tl'om the SaoPaulo 01' Santos Plateaus may explain the kaolinite enrichmenton the western side 01' the RGR, it is hard to imagine a mecha­nism that transports kaolinite-rieh suspensions around 01' acrossthe RGR to the "backside." 1'0 explain the symllletriealnature ofthe kaolinite enrichment on the RGR, we alternatively suggestthe injection 01' kaolinite-rich suspensions into intermediatedepths 01' the NADW off the mouth 01' the Rio Doce and short­distance transport southward instead 01' long-distance advcetionfrom lowcr latitudcs. This idca is supportcd by high kao­Iinite/ehloritc ratios down to 4000 III wafer depth off thc rivermouth, dClllonstrating that kaolinitc-rich suspensions are able torcach decpwatcr Icvels. Injeeted at 20 oS, these suspensionscould be advceted south with the NADW anel be e1epositeel onthe first obstaclc, whieh is thc RGR.

Our results suggest that a substantial part 01' the terrigenousseelimentation on the RGR originates from freshwater input anelsuspension supply by the Rio Doce as evidenced by the kaolinitecnriehment at intermcdiate depths 01' thc RGR. Thc qucstionarises of whcther tcmporal variations in kaolinite supplyprimarily record pulses 01' fluvial discharge 01' rather representvarying intensities of short-distance NADW flow. 1'0 aelelressthis problem and look at the temporal variations 01' the kao­linite/chlorite ratios, we investigateel three seeliment COlTS.

3.2. Temporal Variations 01' Clay Mineral Input

Core GeoB 2110 was taken from 3000 m wafer e1epth at thecontinental slope above the Santos Plateau. Kaolinite/chloriteratios in surface sediments reach a maximum at this e1epth (Fig­ure I) because 01' advection anel deposition 01' kaolinite from thenorth with the Brazil curren!. Significant fluctuations 01' chloriteflux can be ruled out 1'01' this site because the only potentialcarrier of chloritc, AABW, always stayeel weil below thcsce1epths in thc southwestern Atlantie [Cuny, 1996]. Sitc GeoB2110 shows highest kaolinite/chlorite ratios (4-8) in interglacialstages I, 5 and 7 and lowest ratios (± 1) in glacial sections (Fig­ure 3). This is still in aeeordance with the model 01' .Iones

[1984], who proposeel a southward transport 01' kaolinite fromthe mouth 01' the Rio Doee only during times 01' high sea level(interglaeials) anel a direet injeetion 01' the river load into inter­mediate and deepwater levels during times 01' low sea level

Figllre 3. (opposite) Relativc clay Illineral percentagcs and kao­linite/chlorite ratios of cores GeoB 2110, GeoB 2821, ami GeoB 2822versus agc. In ordcr to producc a maxilllulll glacial-intcrglacial contrastthe cutoff for the shading 01' kaolinite/chlorite ratios was set at anaverage valuc 01' 1.0 101' the cores within the North Atlantie Decp Water(NADW) (GcoB 2110 ane1 GeoB 2821) anel at 0.6 for the core within theAntarctic Bottom Water (AABW) (GeoB 2822). Isotope stages 1-7flmbric cl a/., 1984] are shaded.

100 kyr 41 kyr rkyr

,

19 kyr

CO 152.. 1 1

E: /\:::J

10'-U :! ~Q)0.. I 1I 1\Cf)

-0 5 T'/ '\Q)+-' Y 1\CUE JI 1\t5

~/1 1\

W 0 I +-1 I II--~-l

0 0.02 0.04 0.06

Frequency [1/kyr]

Figllrt' 4. Speetral analysis ofkaolinitclchlorite ratio orcorc CieoLJ 2110.In addition to speetral peaks near I(JO and 41 kyr, strang response topreecssional löreing hccomes visible in the modifJed pcriodogralllcalculatcd by Lomb-Scargle Fourier transrorlll using a Welch window.Thc cross dcpicls (, dB bandwidth (horizontal) ami 80'% cunlidcnceinterval (vertieal).

(glaeials). However, our surface data show high kao­linite/chlorite ratios e10wn to 4000 m wate I' e1epth also in recent(interglacial) sediments off the Rio Doce. Furthermore, if scalevel exerts a major control on the input of kaolinite tiom theRio Doce, minima in kaolinite/chlorite ratios could bc alsoexpcctcd at the e1eep sites GeoB 2822 and 2821 during highstands. Howevcr, both e1eep sites show high kaolinite!chloriteratios during intergiacials. Thereforc we hypothesi/e that apotential effect 01' sea levcl ehanges on the input of kaolinite toall our core sites is overridden by fluctuations in the totalamount 01' river e1ischarge.

Spectral analysis of the kaolinite/chlorite ratio 01' core GeoB21 10 (Figure 4) strengthens this assumption. Although the eom­bined ehronostratigraphy e1escribed above does not preserve theexact phase relation 01' different proxies, a Fourier transformyields information on speetral characteristics within the uncer­tainties 01' the dating proeedurc. The resulting spectru111 indicatesthat there is a strong response of the kaolinite/chlorite ratio toprecessional foreing. There is abundant evidence (summarizedby Delvfenocal [1995] and Gingeie el al. [1998]) that low-Iati­tude aridity-humielity cycles documenteel in eolian flux, mon­soon intensity, 01' river discharge respond to variations in low­latitude insolation, which in turn result from Earth' s orbitalprecession. Therefore we concluele that the kaolinite/chloriteratio at site GeoB 21 10 is significantly influenced by low-Iati­tude river discharge.

The spectrum also e10euments i00 and 41 kyr variance(Figure 4). The dual nature 01' high- and low-Iatituele forcing 01'low-Iatituele climate has been recognized by DeMenocal [1995]

in various sites around the African continen!. Mechanismstransporting high-Iatitude signals, namely the pronounced 100and 41 kyr cycles, to low latitueles are believed to be eoolingNorth Atlantic surface water anel fostering shifts in atmosphericcirculation anel migration 01' vegetation belts. Especially the 100kyr frequency 01' eeeentricity is prominent in many aridity­humielity reeorels [Paslourel el al., 1978; DeMenocal el al.,1993; Gingeie el al., 1998].

76 Chapter 5

Age [ka]

0 200 400 600 800 1000 1200 1400

3 1.8

f2 1.6 c:I'-c.o 00.- 14 ,

4 (\)0 ......0 o'~ 1.2 ~.

m G)CD

0 1.0 00.- 0]

0 5 N~ 0.8 (XlE.-- N0 --'oe

Io:J --'oe

vo 0.6

6

0 200 400 600 800 1000 1200 1400

0.5

o0.060.02 0.04

Frequency [1/kyr]

---A k __ I,-

- ----- - -----~ -A --~-

--I

I

I I AII I I I -I

:l~I I

I

I~I I

I I I

+- I I I

80%level

0.06 00.02 0.04

Frequency [1/kyr]

o

10 10m2.E::::J 0 0'-.-üQ)0.

Cf)

"0Q) -10

CDE

:;:::;(()

w-20

>,ücQ)'-Q)

..coU 0.5"0~cu::::J

g 0

Figure 5. Cross-spcctral analysis ofkaolinite/chlorite ratio 01' core GeoB 282 I (black lines) and the reversed8 180 recorcl 01' Ocean Drilling Program(OOP) Site 677 (white lincs) [Shacklelol1 el a/.. 1990]. While the past 500 kyr are dOlllinated by eccentricity-relatecl 100 kyr cycles (A), an increasedalllount ofvariance in the obliquity band near 41 kyr appears in the spectra 01' the 0-1500 ka records (B). Both time serics are intel1JOlated to 5 kyrsteps, the average time resolution 01' the kaolinite/chlorite recorcl 01' GeoB 2821. The analysis was performed using We!ch's [1967] overlapped seg­ment average (WOSA) procedure with two ovcrlapping segments and a Welch window. The sign ofthe 8 180 record ofOOP 677 was inverted to attainclirectly intelvretable phase relations. Crosses depict 6 clB bandwidth and 80% confidencc interval. The resu1ting negative phase angles shown inTable 2 indicate a lead with respcct to 8 180. High squarcd cohcrcl1cy is calculatecl in the 100 and 41 kyr band.

Tenigenous flux in tbe Rio Grande Rise area during tbe past 1500 ka 77

Core GeoB 2822 from the eastern terraee of the Vema Chan­nel (4300 m water depth) is believed to represent sedimentationwithin the AABW during interglaeial as weil as glaeial times.The adveetion of chlorite from the south is doeumented in thelowest kaolinite/chlorite ratios «1) of all investigated eores.Although of low amplitude, a signifieant glaeial-interglaeialpattern in the kaolinite/chlorite signal ean be reeognized, withhigher values in the interglacial seetions (Figure 3). Evidencefrom grain size and sediment texture from the area [Jolmsoll emdRaslllllssell, 1984; Masse el !l1., 1994] indicates a more vigorousnow of AABW at the transition from warm to cold periods andwithin glacials. !nereased adveetion of chlorite with intensifiedAABW activity dming glacials could explain the observednuctuations in kaolinite/chlorite ratios of eore GeoB 2822. Sincethe site is weil below the direet innuenee of NAOW the advee­tion of kaolinite from low latitudes is an unlikely source forGeoB 2822. Therefol'e we remain with the Rio Ooce as theprimaJ'y kaolinite souree also for this eore.

Cyelie variations are also reeorded in kaolinite/chlorite ratiosof COlT GeoB 2821 (Figure 3). High values correspond to lowsin the suseeptibility reeord (Figure 2) and thus interglacial stages(Figure 3). Thc clay mineral proxy nuetuates in tune with manypalcoclimate proxies dominated by high-Iatitude orbital foreing.A shift in the dominant perioel of variation from 41 to 100 kyr isvisible near 1000 ka (Figure 3).

Cross-speetral analysis of the kaolinite/chlorite ratio of coreGeoB 2821 was performed versus thc reversed SPECMAP 8 180stack and reversed 8 180 reeord of OOP 677 for the past 500 kyrand and versus the reversed 8 180 record of OOP 677 for the past1500 kyr. It appears that the elay mineral proxy is eoherent withglobal iee volume in the 41 and 100 kyr periods of obliquity aneleeeentrieity (Table 2 and Figures 5 and 6). Resolution and sedi­mentation rates of the eore are too low to investigate a preees­sion-related variability. Moreover, statistiea! data (Table 2) forthe complete kaolinite/chlorite reeord are difficult to interpretsinee signifieant variation in the 100 kyr banel oeeurs onlyduring the past 350 kyr, where 41 kyr eyeles, elominating theolder section, are redueed (Figure 5). As a consequenee of thisrefleetion of the mid-Pleistocene elimate transition [e.g., Rliddi­Illall el a!., 1989] we eoncentrate on the seetion from 0 to 500 kafor the !00 kyr eyclc anel make use of the whole reeord for the41 kyr period. As statistical data (Table 2) and phase wheels(Figure 6) indicate, maxima in GeoB 2821 kaolinite/chloriteratios lead maxima in NAOW nux by 5800 years in the 100 kyrcycle (0 to 500 ka) and 4 I00 years in the 41 kyr eycle (0 to 500and 0 to 1500 ka). These results are consistent with characteris­tic values of early proxy responses in the Southern Hemisphere[Imbrie el a!., 1993].

NAOW production and propagation have varied in orbitaltime scales, though the relation to climate is not simple [Clirry,1996]. In the western Atlantie (Ceara Rise), only NAOW wasrecoreled in some peak intcrglacials, whereas in some glacials,southern source deepwater reached 3200 m water depth [ClIny,1996]. Investigations in the eastern part of the South Atlantic[Dieklllallll el al., 1996] have shown that the propagation andsouthwarel extension of northern source deepwater (NADW)were more pronounceel during interglacials, thus providing morekaolinite during warm periods. At first glance, kaolinite/chloriteratios of core GeoB 282 I fit nicely in the concept of alternatingintensities of interg!acial NADW [Diekmallll el al., 1996] and

glacial AABW [Masse el al., 1994] now. Within the error mar­gins (Table 2), phase angles of kaolinite/chlorite ratios in the100 kyr band of eeeentrieity are similar in the castern anelwestern Atlantie. The implieations for the RCiR are tllat eitherthe kaolinite/chlorite ratio is eontrolleel by ehanges in deepwaternux or that the high-Iatituele forcing meehanisms, whieh controleleepwater nux and low-Iatitude elimate in the eeeentricity baneI,do not show a signifieant time lag. However, there are strongarguments opposing an exelusively deepwater-controlled elaymineral supply for sites GeoB 2821 as weil as GeoB 2822.

First, AABW aetivity was strongly reduced after 350 [Masse'el al., 1994] 01' 275 ka [JOhllSOll alld Ra.s·1ll11,,·sell, \984], respee..tively, wh ich would lead to rising kaolinite/chlorite ratios f"rom350 ka to present. Instead, we find deereasing kao)inile!chloriteI'atios in COlT GeoB 2821.

Second, in the 41 kyr band, maxima of kaolinite/chloriteratios of eore GeoB 2821 signifieantly lead minima in ice vol­ume, maxima in 'Yr,NADW tlux [Raymo el al., 1990] andmaxima in kaolinite/chlorite ratios in the eastern South Atlantie[Dieklllallll el ClI., 1996] (statistical elata in Table 2). Moreover,the phase angle of kaolinite/chlorite ratios of core GeoB 2821 inthe 41 kyr perioel is elose to the maxima of aD (deuterium) inthe Vostok iee core [JOlizel el ClI., 1994]. Maxima in the aDreeorel inelicate perioels of inereased atmospheric temperatures inAntaretiea [Waelbroeck el al. , 1995], whieh are relateel to inten­sifieel evaporation anel preeipitation in lower latitudes. Conse­quently, we suggest that temporal ehanges in the kao­linite/chlorite ratio of the RGR eores may be rather interpretedas an atmospherie signal of preeipitation anel rivcr input than asa proxy of eleepwater fluetuations.

4.Paleoceanographic and PaleoclimaticImplications

Summarizing the evidenee from elay mineralogy, grain size,anel magnetie susceptibility, we finel that the flux of terrigenousmatter on the RGR shows eyclie variations at least in the 100anel 41 kyr periods. fnterglaeial stages are eharaeterized by thedeposition of fine-grained, kaolinite-rieh terrigenous matter oflow magnetie suseeptibility. They reeord perioels of inercaseelhumidity on the South Ameriean mainland and runoff of the RioDoee. High nuvial eliseharge was also reeoreleel for the Orinoeoanel Amazon Rivers eluring interglacials [B01vles ami Fleischer,1985]. C1apperlon [1993] reported more humid eonditions forinterglaeial stage 5 in Patagonia. The eoneept of increaseelhumielity in interglaeial perioels (100 kyr perioel) is in aeeor­danee with results from multiple investigations on the Afrieaneontinent (summary in DeMenocal [1995]). Runoff of majorAfriean rivers anel monsoon intensity fluetuates in the 100 kyrperiod of global iee volume and 23 kyr perioel of low-Iatitudeinsolation [PaSIOl/rel el al., 1978; Zachariasse el al., 1984;Rossignol Sirick, 1985; Pokras, 1987; Gillgele el al" 1998]. Ourdata from the RGR yield similar results for the South Amerieaneontinent at 20 0 S in the 100 and 41 kyr periods. In eore GeoB2 110 the resolution is sufficient to record fluetuations of kao­linite nux in the 23 kyr precessional band. They are believeel torepresent humidity anel discharge variations of the Rio Doeeinitiated by low-Iatitude insolation changes.

Minima in the preeessional index eorrespond to inereasedinput of kaolinite anel may be interpreted in terms of higher

78

1OO-kyr cycle

Max. Eccentricity

Max. Kaol./Chlor.43"S SE-Atlantie

I/

Max. ieeSPECMAP

Max. ieeSPECMAP

Chapter 5

41-kyr cycle

Max. Obliquity

Min. iee SPECMAP

.-i.~=:::::::;;;~~~IE~M'ax."%NADW" flux (41"N)-- Max. f<aol./Ch!or.

43°S SE-Atlantie

Figurr 6. Phase relallonships hctween kaolinite/chlorite ratios ofco]T Gcoll 2821 (0-)00 ka), 'X, N!\DW /lux IRo)'1I1O ('{ 111.. 1<J()Ojlaolinitc/chloritcratios 01' a core 43°S in the SE i\tJantic IDiekllloli1i cl al., 199ö!, ami thc maximum ol~D (deuterium) in the Vostok iee cme l.JolI~cI Cl 111., 1')9.\] in tlw100 ami 41 kyr pcriods 01' ccccntrieity and obliquily (statistical data sec Tablc 2).

humidity in the drainage area ofthe Rio Doce. It is interesting tonote that at about the same latitude west of the Andes in north­ern Chile, precession-related humidity/aridity cyclcs are re­cordcd, which show exactly the opposite pattern [Lall/)'. cl al.1998]. There, maxima in the precessional index correspond tomore humid eonditions in the sediment record. Inereased pre­cipitation there is related to a shift in the position ofthe SOllthernwesterlies and indicates a nonsynchronous behavior of SouthAmerican palcoclimate on an orbital timescale.

Glacial stages are characterized by maxima in magnetic sus­ceptibility, illite, and grain size (more silt) and lows in kaolinitesupply. Cold and dry conditions have been reported for theheadwaters of the Rio Doce during the last glacial [Be/ding andLiehle, 1997]. The lack of dilution by fine-grained river suspen­sions. a more vigorous AABW flow, and the increased input ofsilt-sized dust, rich in magnetic partielcs and i1lite, may havecombined to form the terrigenous signal of the cold periods.Previous studies have shown that major fluctuations in dust fluxappeal' to be of global signifieanee with maximum inputs duringcold and arid periods [elell/ens and Prell, 1990; DeAlenoeal 1'1

al., 1993]. A potential souree area for dust in the southwesternAtlantie is Patagonia. Patagonian dust is derived from loessdeposits, whieh contain a high pereentage of titanomagnetite andillite [Teruggi, 1957; Bonorino, 1966; Zarale and Balsi, 1993].Although insignifieant for the total terrigenous mass balance, theinput of Patagonian dust in cold periods eould seleetivelyenhanee the suseeptibility signal preformed by glaeial carbonatedissolution events. lt mayaiso eontribute to an illite enriehmentof glaeial eore seetions.

It appears that the eomposition of the terrigenous matter onthe Brazilian slope and also on the RGR responds to ehanges inhumidity/aridity on the adjacent South Ameriean eontinentrather than to variations in deepwater adveetion. The dual influ-

ence of high- and low-Iatitude foreing on the palcoelimate in theBrazilian hinterland is believed to result in periodieal disehargeof the Rio Doee. Consequently, the supply and deposition ofkaolinite on the slope and the RGR are reeorded for the majorMilankovitch periodieities.

Multiple potential sources complieate the interpretation ofthemain elay mineral eomponents smeetite and illite. Nevertheless.individual peaks in the downeore reeord of smectite in eoreGeoB 2821 are fi'equently assoeiated with maxima in kao­linite/chlorite ratios and thus with warm and humid periods,whereas i1lite shows a reciproeal pattern. More evident is a long­term trend in smeetite and illite percentages (Figure 3). Illitepereentages inerease while smeetite percentages, as weil asaverage kaolinite/chlorite ratios, decrease from 1500 ka topresent. If the paleoelimatie interpretation of the elay mineralproxies is eorreet, a trend towarcl more arid and glaeial eondi­tions during the Pleistoeene is evident. These findings are eon­sistent with results from the Afriean eontinent [Del'vfenoeal.1995]. Here proxies for Afriean aridity reeonstruetecl fromvarious ODP sites around the eontinent show a gradual trendtoward more arid and cooler eonditions from the Plioeene topresent with prominent shifts arouncl 2800, 1700, ancl 1000 ka[DeMenoeal, 1995].

5. Conclusions

Clay mineral analyses revealed that the Rio Doee (Brazil) is amajor souree of kaolinite for marine sediments at 20o-30oS inthe western Atlantie. Charaeteristie enriehments of kaolinite areobserved at intermecliate depths of the eontinental slope and onthe flanks of the RGR at 3000-4000 m water depth.

The supply of kaolinite with NADW is minor as evideneeclby the eomparison of kaolinite/chlorite ratios from the MAR and

Tenigenous Dux in the Rio Grande Rise area during the past 1500 ka 79

thc RGR, Conscqucntly, pattcrns of tcrrigcnous scdimcntation in

this part of thc Atlantic rccord climatic conditions on thc South

Amcrican hintcrland.

Low-Iatitudc insolation fostcrs pcrioelical elischargc of thc

Rio Docc in thc prcccssional 23 kyr baneI, as cvidcnccd in kao­

linitc/chloritc ratios on thc Brazilian slopc, Cyclic variations of

kaolinitc/chlorite ratios on thc slopc anel in a corc from thc

wcstcrn flank of thc RGR arc cohercnt with global icc volumc in

thc 41 anel 100 kyr pcrioels, Thcy are also bclicvcd to rccord

fluctuations in thc dischargc of thc Rio Doce and mirror humid­

ity conelitions on the adjaccnt South Amcrican hintcrland, Hu­

miel pcrioels arc cocval with warm intcrglacial phascs, whcrcas

ariel pcriods correspond to colel, glacial stagcs, Thc elualmodc of

high- anel low-latitude foreing of low latituele clilllate is consis,

tent with silllilar findings li'om thc African contincn! A long­

time elccrcasc in smcctitc content and kaolinite/chlorite rarios

from 1500 ka to present is bclieved to elocul11cnt a trcnd toward

more arid and cooler climatc conelitions for subtropical southcrn

latitudcs of South America,

Acknowlcdgmcnts, The dcdication and etrort of crew and mastcr o!"R/V Meleol' is greatly acknowledged. We are grate!"ul to r:. Bassinot andD, Dobson 101' construetive reviews allCl hclpl'ul commenls. All thc dalaused in this manuseript are arehived in the inl'ormatiou systemPANGAEA/SEPAN (www.pangaea.de). F, Sehmleder was supported bythe Deutsche Forsehungsgeilleinschat't in the framcwmk o!"Ciroduierteu­kolleg 221, This is contribution 249 of spccial research p1"O.leet SFB 261.

References

Balsam, W I,., B, I" Otto-Bliesner, ami 13, c:.Deaton, Modem ami Last ()Iaeial Ivlaximumeolian sedimcntation pattems in the AtlanticOcean intcrprcted ti'om sediment iron oxidecontent, l'al!'Oceanogmphy, I IJ, 493-507,1995,

ßehling, H., and lvI. Lichte, Evidence of dry amicold climatic conelitions at glaeial times intropical southcastern Brazil, Qllal, Res., 48,348-358, 19'J7

Berger, A" ami M,F, Loutre, Insolation valuesfor the cllmate of the last 10 million years,Qllal, Sei, !?e\', IIJ, 297-317,1991.

Biekert, 1'., Rekonstruktion der spätquartären130denwasscrzirkulation im östlichenSüdatlantik übel' stabile Isotopc benthischerForaminiferen, l3er. {'achhereich Geowiss.Univ, l31'ell/en, 27,205, 1992,

Biscaye, PE., Mineralogy and sedimentation ofreeent deep-sea clay in the Atlantie Oeeanand adjacent seas ancl oceans, GSA 131111.,76,803-832, 1%5,

Bleil, U" anel Cruise Pal'ticipants, Report andpreliminary results ofMeleol' cruisc 23/2 Riode Janeiro - Recife, 27,2,1993 - 19,3,1993,Bel', Fachbereich Geowiss, Univ. l31'ell1ell,43, 133, 1993,

Bleil, LJ" and Cruise l'articipants, Report andpreliminary reslJits of Meleol' eruise 29/2Montevideo - Rio de Janeiro, 15,7,1994 ­08,08,1994, Bel', Fachbereich Geowiss,Ulliv, l31'ell/en, 59, 153, 1994,

Bonorino, F, Ci., Soil clay mineralogy of thePampa Plains, Argentia, .J. Sedilllelll, Res"36,1026-1035,1966,

Bowles, F, A., and P, Fleischer, Orinoeo andAmazon River sediment input to the easternCaribbean Basin, M(/I' Cleol., 68, 53-72,1985

Boyle, E, A" Caelmium: Chemieal tracer ofdcep-water paleoeeanography, Paleoceallog­mphy,3,471-489,1988,

Boyle, E. A" A comparison of carbon isotopcsallCl cadmium in the modern and glacialmaximum ocean: Can we aCCOlull for theeliserepaneies? Cal'boll Crelillg ill Ihe CllacialOceall: COl1slmil1ls 011 Ihe ()ceal1 's Roll' il1Global Chal1ge, NA 1'0 ASI 5'el' C. vol. I 17,edited by 11., Zahn et al., pp. 105-144,Springer, New York, 1994.

Chamley, H., Inlluence des eourants profonds aularge du bresil sur la seelimentation argeleuserecent, lXI' COl1gl', 1111, Sedill1el1lol. Nice,Abslmcls, 13-19,1975,

Clapperton, e. M" Quatemm)' Cleologr alldGeoll1olphology 0/ SOlItli Alllerica, 779 pp"Elsevier, New York, 1993,

Clemens, S, e., and W, L. Prell, Late Pleisto­eene variability of Arabian Sea summcr mon-

soon winds and continental aridity: Eolianrccords !"rolll the lithogcnic cOlllponcnt o!"deep-sea sedilllcnts, hdeocelll1ogmpliy, 5,109-145,1990

Curry,W, 13., Latc Quaternary dcep cireulationin the western equawrial Atlantic, Tlle SoulliAIlalllic: Preselli al1d Pasl ('iI'Clilaliollcdited by Ci Wel'cr et al., pp. 577-598:SprinGer-Verlaa New York 1996

Curry, \V, H, J c', Duplcssy: L. D Labeyrie,allCl N, ,I, Shaekleton, Changes in thedistribution of Ol3C o!" deep water LC02between thc last glaciation and the Holocene,Paleoceallogmpliy, 3, J 17-341, 1988,

Dalllllth,.I. E" Late Quaternary sedimentation inthe western eqllatorial Atlantic, Geol. Soc.;/111, Bull" 88,695-710,1977,

DeMclo, U., C, P, SlIllllllerhayes, allCl J P, Fllis,Salvador to Vitor!a, southcastern 13razil, IV,COIIIl'ib, 5'edillleI1I01., 4,78-116, 1975,

DeMenocal, P., Plio-Pleistocene Ati-iean el i­mate, Sciel1ce, 270,53-59, 1995 ,

DeMenoeal, P, 13" W, F, Rudcliman, and E M,Pokras, Intluences 01' high- anel low-Iatitudeprocesses on Ati'iean terrestrial climate:Pleistocene eolian records from equatorialAtlantie Ocean drilling program site 663,Poleoceallogmphy, 88, 209-242, 1993,

Deuser, W. G., P, G, Brcwcr, T. D. Jickells, andR, F, Commeau, l3iological control 01' thercmoval 01' abiogenic partieles from the sur­face oeean, Sciellce, 219, 388-391, 1983,

Diekmann, 1'3., R, Petsehick, 1', X. Gingeie, D,K, Fütterer, A, Abelmann, R, Gersonde, U,Brathauer, and A. Maekensen, Clav mineralIluetuations in late Quaternary sed-iments ofthe southeastern South Atlantie: ImplicationsJor past changes of deepwater advcction, nieSoulh Allolilic: Preselli ami Pasl Cil'Clilolioll,edited by G. Wefer et al., pp. 621-644,Springer-Verlag Ncw York, 1996,

Duplcssy, ,I, c., N, .I. Shaekleton, 11., G. Fair­banks, I... D" Labeyrie, D, Oppo, and N,Kallel, Deep water source variations duringthe last elimatic cycle allCl their impact on theglobal deep water cireulation, Paleo­ceonogmphy, 3, 343-360, 1988,

Ericson, D, B" and G, Woll in, Pleistoceneclimates and chronology in deep-sea sedi­ments, Sc/enee, 162,1227-1234,1968,

Gingeie, F, X" P .I. Müller, allCl R, 11., Schnei­der, Orbital forcing of ti'eshwater input in theCongo-Fan area: Clay mineral cvidenee ti'omthe last 200 kyr, Poleogeogr, Paleoclimalol,Poleoecol., 138, 17-26, 1998,

Ciri1'1'in, ,I, ,I" H, Windom, anel E, D, Goldberg,The distribution of elay minerals in theWorld Ocean, Deep Sea Res, Oceanogr,Absll'" 15,433-459, 1968,

Heath, G, R" and N, G, Pisias, A method tor thequantitative estimation of elay minerals in

North Pacific dcep-sea sediments, C/m's C/oJ'AIincl',,27,175-184,1'J79, ,.

Imbrie, J, .I. D, Ilavs, D, G, Martinson, ;\,Mclntyre, anel AC', Mix, The orbital thcorv01' Pleistoccne climatc: Support ('rom a re'­vised ehronology of the marine 0 180 reeord,in Milollkovilch ami Clilllflle, Pari I, editcdby ABerger ct al., pp, 269-305, D, Reielel,Norwell, Mass" 1984.

Imbrie .1., et al., On the slructure allCl origin 01'major glaciation cyeles, 2, Thc 100,OOO-yeareyelc, Poleoceollogmphy, 8,699-735, 1993,

.lahn, B" Menardii stratigraphy, in Repol'1 IIlIdPrelill1inorr ReslIlls 01' Meleol' Cmise 29/2MOlllevidc() - Rio dc' JOlleil'o, 15 7.1994 ­08,08.1994" edited by U, Bleil anel CruiseParticipants, Ber, Fachbereich Geowiss,Ulliv, I3rell/ell, 59, 153, 1994.

,lohnson, D. A" and K. A. I,asmussen, LateCenozoic turbidite anel contourite depositionin the southern Brazil Basin, Mol', Geol., 58,225-262, 1984,

,Iones, G, A" Adveclive transport of clay miner­als in thc region of the Rio (jrande Rise,lviaI', Geol., 58,187-212,1984,

,Iones, G. A" and D, A, ,lohnson, DisplacedAntaretie diatoms in Vema Chal111el sedi­ments: Late Pleistoeene/Holocene lluctua­tions in AABW 1l0w, MIII', Geol, 58, 59-88,1984

Jouzel, .1., C. Lorius, .I, R, Petit, c:. Ritz, M.Stievenard, P. Yiou, N, I. Barkov, V, M,Kotlyakov, and V, Lipenkov, The climatierecord from Antarctic ice now extcnds baekto 220 kyr BI', in LOllg-lenll Climillic Vllrill­liollS, NATO ASI SeI', , I, vol. 22, edited by ,1.­e. Duplessy anel M ,-1'. Spyrielakis, pp. 213­237, Springer, New York, 1994,

Lamy, F" D, Hebbeln, anel G, Wcfcr, LatcQuaternary preeessional eyeles of terrigenoussediment input off the northern ehieo, Chile(27,5°S) and paleoclimatie implications,Paleogeogr, Paleoclimalol, Paleoecol., 141,233-251, 1998.

Lomb, N,R" Least-squares ti-equeney analysis 01'unequally spaeeel elata,Asll'Ophys, S'pllce Sci"39,447-462,1976

Maekensen, A" D, K. Fütterer, H, Grobe, and G,Schmiedl, 13enthie foraminiferal assemblagesti'om the eastern South Atlantic Polar Frontregion between 35° and 57°S: Distribution,ccology and tossilization potential, Mal',Micl'Opaleolllol., 22,33-69, 1993,

Mackensen, A" H, Grobe, I-I.-W, I-lubberten, andG. Kul111, 13enthic foraminiferal assemblagesand the ol3C-signal in the Atlantic sector 01'the Southern Oeean, Carboll ()lC!ing ill IheGlacial Ocellll: Conslminls Oll Ihe Ocean '5

Roll' in Global Chllnge, NATO ASI SeI', C.vol. 117, editeel by 11., Zahn et al., pp, 105­144, Springer, New York, 1994,

80

Masse, L., l-C. Faugercs, M. 13crnat, 11. Pujos,ancl M.-L. Mezerais, 11 600,000-year recorcl01' IIntarctic Botlom Water activity in1'errecl/i'om sediment textures ami struclures in asediment core /i'om the Southern Brazil Ba­sin, Paleoceal/ogmphy. 9, 1017-1026, 1994.

Mulitza, S., Stratigraphy, in Reporl ami Prelill1i­1/(11)' l?eslIlls ol Mcleor Cmise 23/2 I?io deJal/ciro - I?cci/i', 27.2.1993 - 19.3.1993,editecl by U. Bbl, ami Cruise Participants,Ber. Fachbercich Geowiss. Ul1iv.Brell1en, 43,133,1993

Oppo, D. W., ami Y. Rosenthai, Cd/Ca ehangesin a deep Cape 13asin eore over the past730,000 years: Response 01' circumpolarclccpwatcr variability to Northcrn Hemi­sphere icc sheet mclting'" Paleoceal/ogm­phy, 9, 661-675, 1994.

Pastouret, L., 1-1. Chamley, G. Delibrias, l C.Duplcssy, allCl l Thiecle, Latc Quaternaryelimatic changes in western tropical A/i'icadeduced 1'rom dcep-sca sedimentation otlNiger delta, Oceallo/. //cla, 1,217-232, 1978.

Pätzold, l, ami Cruise Partieipants, Berichte underste Ergebnissc über die MClcor-FahrtM 15/2 Rio de .1aneiro - Vitoria, 18.1.1991 ­7.2.1991, Ber. Fachbereich Geowiss. UI/iv.Brell1cl/, /7,46, 1993.

Pctcrson, R. G., and L. Stramma, Uppcr-Ievelcireulation in the South Atlantie Oeean,ProgI'. Oceal/ogr.. 26, 1-81, 1990.

Petschick, R., G. Kuhn, allC] F. X. Gingeie, Claymineral distribution in surf~lcC sediments 01'the South At1antic: Sourccs, transport andrelation to oceanography, Mal' Geo/., 130,203-229, 1996.

Pokras, E. M., Diatom record 01' late Quatcrnaryclimatie change in the eastern equatorialAtlantic and tropieal Al1'ica, Poleoccol/ogm­plly, 2, 273-286, 1987.

Raymo, M. E., W F Ruddiman, N. .I. Shackle­ton, and D. W. Oppo, I:volution 01' Atlantic­Paeitie ü13c: gradients ovcr the last 2.5 ll1.y.,Eonh PIonei. Sci. Lcll., 97,353-368,1990.

Robinson, S, G., Applications 1'01' whole-coremagnetic susceptibility mcasurcmcnts 01'

Chapter 5

dcep-sea sediments, Proe. ()cean /)ril/. Pro­gralll Sci. Res"lls, I 15,737-771, 1990.

Rossignol-Striek, M., Mediterranean Quaternarysapropels: IIn immcdiate response 01' theAfriean monsoon to variation 01' insolation,Poleogeogr. Palcoelill1olo/. Paleoecol., 49,237-263, 1985.

Ruddiman, W.F., M.E. Raymo, D.G. Martinson,13.M Clement, ami .I. Baekman, Plcistoeeneevolution: Northern Hemisphere ice sheetsami North Atlantle oeean, Paleoceooogra­plly, 4, 353-412, 1989.

Seargle, l[)., Studies in astronomical time seriesanalysis, 11, Statistical aspects 01' speetralanalysis 01' uncvenly spaced data, Asrropllys.J, 263, 835-853, 1982

Scargle, .1.0., Studies in astronomical time scriesanalysis, 111, Fourier translorms, autoeorrcla­tion funetions, and cross-eorrelation func­tions 01' uncvcnly spaced data, ASlropllys. J,343, 847-887, 1989

Schmiedl, Ci., and A. Mackcnscn, Late Quatcr­nary palcoproduetivity ami decp water cir­eulation in the eastern South Atlantie Oeean:Evidenec li'om bcnthie foraminifera, Poleo­gcogr. Polcoclill1a!ol. Paleoeco!., 13IJ, 43-80,1997.

Schulz, M., ami K. Statlcgger, Speetrllm: Spee­tral analysis 01' uncvenly spaced palcocli­matic time scries, COIlljJ. Geosci., 23, 929­945,1997.

Shaekleton, N. .1., A. Berger, amI W. R. Peltier,An alternative astronomieal ca libration 01' thelower Pleistoecnc timcscale based on OJ)Psite 677, hans. R. Soc. Edinb"rgll Eorlll Sci.,81,251-261,1990

Teruggi, M. E., The nature ami origin 01' Argen­tine Locss, J Scdilllcnlo/. Pelm/., 27/3,322­332, 1957

von Dobeneek, 1'., and F. Schmiedcl', Using rockmagnctie proxy rccords 1'01' orbital tuning andextenclecl time series analysis into the superami sub-Milankoviteh bands, Proxies in Pa­Icoccanograplly, cdited by G. Fischer and G.Werer, Springer, New York, in press, 1998.

Waelbroeck, c., .I . .Iouzel, L. L.abcyric, C.Lorius, M. Labraeheric, M. Sticvcnard, andN. l. 13arkov, A cOlllparison o1'the Vostok ieedeuterium record and serics jj'om SOLilhernOecan eore MD 88-770 ()Ver the last two gla­cial-interglaeial cyeles, Clim. Dyn., 12, 113­123,1995

Wcfcr, G., and Cruisc Participants, Rcport am!preliminary results 01' Meleor-eruise M34/3Walvis Bay - Recife, 21.02. - 17.03.19%,Bcr. Fachbereich Geowiss. Univ. Brelllen.79, 168, 1996.

Zachariasse, W, .1., R. R. Schmidt, ane] R. J W.Van Lcuwen, Distribution 01' 1'oraminiferaanel calcareous nannoplankton 111 Qualernarysedimcnts 01' the castern Angola Basin in re­sponse to climatic and oceanlc fluelualions,Nelli. J. Sea /(es., 17, 250-275, 1984.

Zarate, M., ami A. Balsi, Late Plcistocene­I-lolocene eolian c1cJlosits 01' the soulhcmBuenos Aires province, Argentiml: 1\ pre­liminary model, QIIIII. 1111., 17, 15-20, 1')93.

F. X. Gingeie, Baltlc Sea Research Institllte,Secstralle 15, D-I8119 Wal'l1c11lüncle, (lermany.(c-mail: jj·anz.gingele@io-warnemucnde,dc) .

R. Petschick, Fachbereich Geowissenschaf­ten, Univcrsitiit Frankfurt Scnckcnberganlage32-34, D-60325 Frankfurt, Germany.

C. Rühlcmann, F. Schmieder, anel T. V[)II

Dobcneck, Fachbereich Geowissenschaftcn,Universität Bremen, Bibliotheksstrasse, D­28334 Bremen, Gcrmany.

(Received April 9, 1998;revised October 23, 1998;aeeepteel Oetober 27, 1998.)

Summary and perspectives

Two closely related main objectives delineate the

framework ofthis thesis:

• Establishing representative age models for secii­

ment series from the South At1antic Ocean based

on high-resolution magnetic measurements.

• Extraeting paleoeeanographie and paleoelimatie

information from the elimatically eontrolled

magnetic signals by means of specially adapted

statistical ana1yses.

The rcsults achieved substantiate magnetie cyclo­

stratigraphy as a potentially very powerful tool for

detai1ed, accurate age modeling. Proceeding from

magnetostratigraphic or oxygen isotope dating high­

resolution ehronostratigraphies were established by

orbital tuning of susceptibi1ity records. In the sub­

tropica1 and western tropica1 South At1antic Oeean

as in many other open ocean environments these

signals reileet the varying ratio of climatically mo­

du1ated terrigenous and biogenic sediment accumu­

1atiol1.

Whi1e the SUSAS reeords are essentially do­

minated by carbonate dissolution cycles due to en­

hanced iniluence of southem source waters during

glacia1s, the CEARIS series primari1y mitTor sea­

level re1ated ehanges in tenigenous input. These

regional depositiona1 settings had to be taken into

account in the tuning procedure as they affect tbe

pbase relations between driving orbital variations

and environmenta1 magnetic response and therefore

tbe definition of suitab1e target signals.

Age contro1 is a crucia1 precondition for every

pa1eoceanographic study. Consequent1y, tbe mag­

netic cyc10stratigraphies presented 1ed to severa1

interdisciplinary co-operations. Tbey enab1ed a

temporal analysis ofc1ay mineral input during PIeis­

tocene times recorded in a SUSAS core from tbe

western Rio Grande Rise. In this area surfaee

sampIes identify the Rio Doce as a major source of

kaolinite. Cyclic variations of kaolinite/chlorite

ratios in the 41 and 100 kyr bands were shown to

be eoherent with global iee vo1ume. They are

interpreted to originate from increased humidity in

the Soutb American hinterland and enhanced runoff

of the Rio Doce during interg1aeials. Due to 10w

sedimentation rates, the strong preeessional compo­

nent which must be expected under these eonditions

as a result oflow-latitude insolation changes is not

preserved in this core. However, for sediments from

the Santos Plateau high-resolution susceptibility

measurements permitted the refinement of an

oxygen isotope stratigraphy and the identification

of a strong 23 kyr periodicity in kaolinite/chlorite

ratios. This finding validates the link between low­

1atitude climate changes and kaolinite input.

Extended time series ana1yses ofmagnetic sus­

ceptibility focused on climate variability in the fre­

quency bands above and be10w the main Mi1anko­

vitch frequencies. Higher sedimentation rates a1­

lowed a detai1ed investigation ofsub-Mi1ankovitch

phenomena for sediments from the Ceara Rise.

Using techniques ofbandpass filtering, spectra1 and

bispectra1 analysis high frequency pattems could

be related to periodic harmonics and combination

tones of orbital frequencies as wel1 as to millennial

Dansgaard-Oeschger cycles. The direct Iinkage of

tenigenous sedimentation to tropica1 climate mir­

rored in enviroill11enta1magnetic data opens promis­

ing perspectives for sub-Mi1ankovitch reconstruc­

tions in these latitudes.

Magnetic cyclostratigraphy also enabled detailed

age modeling for the low sedimentation rate depo­

sits from the oligotrophie 'ocean desert' ofthe sub­

tropical South Atlantic yielding comp1ete Pleisto­

cene time series. The main, stil1 enigmatic feature

of this interval of Earth's climate history is the

change from variations with aperiod of 41 kyr to

the late Pleistocene 100 kyr ice age cycles in the

course of the mid-Pleistocene climate transition

82 Chapter 6

(MPT). Visualized by evolutionary spectral analysis

the changing fi'equency content ofthe SUSAS cores

reproduce in great detail the Pleistocene climate

evolution known from benthic oxygen isotopes.

Compared to 21 stationary ETP model the mag­

netic proxy signals document an alternating ex­

change of spectral energy between the 41 and

100 kyr bands. Preceded by the occurrence of 21

single 100 kyr cycle at about 1150 ka, the major

shift towards reduced response to obliquity forcing

occurred at about 650 ka, synchronous with the

onset of 100 kyr cyclicity. These results substantiate

threshold models which explain the late Quaternary

100 kyr cycles by postulating 21 decreasing atmo­

spheric pCO? level.

In addition, magnctic mineral concentrations re­

corded an important paleoceanographic feature not

reported from oxygen isotopes and hence obviously

not directly related to global ice volume. In the mid­

die Pleistocene average susceptibilities are enhanced

by 40%, a consequence ofreduced carbonate depo­

sition as magnetite accumulation remained constant.

Temporal analysis of the SUSAS west-east asym­

metry permits to define a MPT interim state of in­

creased carbonate dissolution, most likely resulting

from enhanced influence of southern source deep

waters. This interval exactly fills the time span bet­

ween the first occurrence oflarger glacial ice shields

at about 920 ka and the onset ofthe 100 kyr cyclic­

ity.

At the end of the MPT interim state, a terminal

event documented by various unusual lithologies

was observed in several cores. This '530 ka event'

is presumably related to strong changes in the ocean

circulation system. Aglobai character ofthis event

is suggested as it occurs coincidentally with other

unusual climate excursions which hint at extremely

warm and humid climates in Asia and Africa.

These new insights to the nature of the MPT were

essentially drawn from a rock magnetic view on

Quaternary marine sediment series. Magnetic cyclo­

stratigraphy and advanced signal analyses were thc

tools to decipher the sediment history ofa key region

for the understanding of the global thermohaline

circulation which, due to very low sedimentation

rates, had been largely excluded from carlier

paleoceanographic studies.

Danksagung

Herrn Professor Dr. Ulrich Bleil danke ich für die Vergabe und Be­treuung der Dissertation, sein stetes, anregend kritisches Interesseam Fortgang der Arbeiten und ganz besonders für das mir in schwie­rigen Zeiten entgegengebrachte Verständnis.

Für die freundliche Übernahme des zweiten Gutachtens danke ichHerrn Professor Dr. Gerold Wefer.

Die intensive Zusammenarbeit mit Dr. Tilo von Dobeneck hat mirgroße Freude bereitet. Gut, daß wir die gleiche 'Wellenlänge' habenund so manches Mal zur richtigen Zeit 'in Phase' gedacht und ge­handelt haben! Danke, Tilo!

Für die nette Arbeitsatmosphäre geht ein dickes 'Dankeschön' anmeine Kolleginnen und Kollegen in der Marinen Geophysik, an dieTechniker Liane Brück, Heike Piero und Christian Hilgenfeldt, un­sere Mädels Katharina Däumler und Andrea Schmidt, den unver­gessenen 'Ehemaligen' Dr. Harald 'Harri' Petennann, Dr. ThomasFrederichs, Dr. Karl Fabian und natürlich an meinen liebenZimmerkollegen Jens Funk. Ein 'Danke' auch an Professor Dr. Volk­hard Spieß und seine Arbeitsgruppe für die gute und produktiveNachbarschaft, an Dr. Franz GingeIe für die gute Zusammenarbeitam Rio Grande Rise und an Dr. Torsten Bickert für seine steteDiskussionsbereitschaft und auch dafür, daß er mir zum Teilunveröffentlichte Daten zur Verfügung gestellt hat. Stellvertretendfür all' die anderen netten Leute im Fachbereich Geowissenschaftender Universität Bremen, die das Arbeiten hier im Hause und aufdem Schiff so angenehm gestalten, möchte ich an dieser Stelle der'guten Seele' des SFB's, Gisela Boelen, danken, die wir alle kennenund lieben und die hier insbesondere für die jungen Leute (undzum Glück auch für die nicht mehr ganz so jungen) so viel tut.

Petra, Dirk und Sebastian Rahe haben mir manch' erholsame Stundein der Heide geschenkt und waren immer für mich da, wenn mirder Himmel auf den Kopf zu fallen drohte. Einen ganz lieben Dankdafür!

Nicht zuletzt möchte ich von Herzen den Menschen danken, die inden Jahren meiner Dissertation mit mir zusammen gewohnt undgelebt haben: Eleonora Uliana, Andrea Spies, Andre Janke, BerndLaser und Markus J osten.