PALEOCEANOGRAPHY, VOL. 16, NO. 2, PAGES 164-178, APRIL 2001
Correlationof late Mioceneto early Pliocenesequences between the Mediterranean
and North
Atlantic
David A. HOdell and Jason H. Curtis
Department of Geological Sciences, Universityof Florida,Gainesville ß
Francisco
J. Sierro
Departmentode Geologia,Universidadde Salamanca,Salamanca,Spain
Maureen E. Raymo Departmentof Earth Sciences,Boston University, Boston
Abstract. Ocean Drilling Program (ODP) Site 982 in the North Atlantic contains a complete latest Miocene to early Pliocene section that was tuned to the astronomicaltimescale by correlating the record of gamma ray
attenuation (GRA) bulk densityto summerinsolationat 65øNand the benthic•5•80signalto orbitalobliquityfor the interval from 4.6 to 7.5 Ma. The astronomicaltuning of the Site 982 record permits a direct bed-to-bed correlationto the cyclostratigraphyof Messinian sectionsin the Mediterranean [Krijgsman et al., 1999a, 2001].
The benthic•5•80signalat Site 982 recordsa latestMioceneglacialperiodthat lastedfrom -6.26 to 5.50 Ma and consistedof 18 glacial-to-interglacialoscillationsthat were controlled by the 41-kyr cycle of obliquity. Although the intensification of glaciation at 6.26 Ma may have contributedto the restriction of the Mediterranean, it precededthe depositionalonsetof the lower evaporiteunit at 5.96 Ma by some300 kyr. The transitionfrom Stage TG12 to TGll at 5.5 Ma marks the end of the latest Miocene glacial period and precedesthe Miocene/Pliocene
boundary by 170 kyr. Althoughbenthic•5•80valuesare relativelylow and /5•80of bulk carbonate reachesa minimum at the Miocene/Pliocene boundary at 5.33 Ma, there is no single "event" that would indicate deglaciation and sea level rise as the cause of the reflooding of the Mediterranean. We conclude that glacioeustaticchangesalone were not responsiblefor either the start or end of evaporite deposition during the Messinian, suggestingthat tectonic or local climate changes in the Mediterranean region were the dominant cause(s).
1. Introduction
of glacioeustatic changes in theevolutionof the MSC (for a review, see Hilgen et al. [1999]).
Theroleofglacioeustatic andtectonic changes in the
isolation, desiccation, and reflooding of the Mediterranean
Correlation of sedimentary cyclesto astronomical changesin solar radiation has resulted in the development
during theMessinian haslongbeen debated (fora review, of veryaccurate timescales for the Neogene.Two see [Kastens 1992]). The marine oxygen isotope record provides a proxy for late Miocene temperature and
approacheshave been employed. Recovery of completeand continuous sectionsby the Ocean Drilling Program (ODP)
glacioeustatic changes thatcanbe compared withthe havepermitted orbital tuningof deep-sea sedimentary timing of events during the Messinian Salinity Crisis
sequences through the Miocene [Shackleton et al., 1990,
(MSC).Ourabilityto infercausal relationships between 1995a; Shackleton andCrowhurst, 1997; Tiedemann et al.,
sealevelandevaporite deposition is limited, however, by 1994].In theMediterranean, correlation ofsedimentary theaccuracy towhich thetiming ofthese events canbe cycles withorbital precession hasprovided anastronomical resolved.Recentadvances in the development of timescale forthePlio-Pleistocene [Hilgen, 1991] andthe astronomically tunedtimescales in Mediterranean and lateMiocene uptothebase oftheMessinian evaporites (for deep-sea sequences havespurred a reassessment oftherole a review, seeHilgen et al. [1999]).A gapexists in the astronomicaltuning of Mediterranean sequencesin the late Messinian because marine sedimentation was interrupted
•Supporting data areavailable electronically attheWorld Data by the isolation anddesiccation of the Mediterranean
Center-Afor Paleoclimatology, NOAA/NGDC, 325 Broadway, duringthe MSC. This "Messiniangap" has been filled Boulder, CO 80303 (email:
[email protected]. gov; URL:
partially by astronomical tuning of sedimentary cycles in
http://www.ngdc.noaa.gov/paleo/).
preevaporiticdepositsin Spain, Italy, and Greece
Copyright 2001 byAmerican Geophysical Union
[Krijgsman et al.,1999a, 1999b; Hilgen andKrijgsman, 1999; Sierro et al., in press], and by correlation of evaporitic
Papernumber1999PA000487 0883-8305/01/1999PA000487512.00
and terrestrial depositsin the Sorbas Basin of southern Spain[Krijgsmanet al., 1999b;Sierro et al., 2001]. 164
HODELL ET AL.' ATLANTIC-MEDITERRANEAN CORRELATIONS
165
In the deep sea, only two sedimentarysequenceshave been continuouslyrecoveredacrossthe Miocene/Pliocene boundary and tuned astronomically. Shackleton et al. [1995a] developeda tuned timescale for the last 6 million yearsat ODP Site 846 in the easternequatorialPacific. At
bath of orthophosphoric acid at 90øC using a Micromass (formerlyVG)Isocarb preparationsystem. Isotoperatiosof purified CO2 gas were measuredonline using Micromass Prism mass spectrometersat MIT and UF, respectively. Oxygen and carbon isotopeswere also measuredon bulk Site 926 on the Ceara Rise in the South Atlantic the carbonate by grinding dried sediment to a homogenous sedimentaryrecord was orbitally tuned from 14 to 5 Ma in powder and reacting untreated samplesin orthophosphoric the Miocene [Shackletonand Crowhurst, 1997]. During acid at 70øC using a Finnigan-MAT Kiel III carbonate ODP Leg 162 to the North Atlantic, Site 982 (57ø30.992'N, preparationdevice. EvolvedCO2 gas was measuredonline 15ø52.001'W)was drilled in shallow water (1134 m)on the with a Finnigan-MAT 252 mass spectrometerat UF. All RockallPlateau. Continuityof the compositesedimentary isotope results are reported in standard delta notation sequence was documented for the upper 255 meters below seafloor (mbsf), corresponding to -7.5 Ma in the late Miocene, by correlation of core-logging signals [Shipboard Scientific Party, 1996]. Site 982 is therefore a good candidate for developing an orbitally tuned chronology for the late Miocene and early Pliocene. We constructedan orbitally tuned timescale for Site 982 by correlating the record of gamma ray attenuation (GRA) bulk density to summer insolation at 65øN and the benthic
3
2.6 ++
2.4
•5•80signalto orbitalobliquityfor the intervalfrom 4.6 to 7.5 Ma. The astronomical timescale permits a direct correlation between proxy records in Site 982 and Messinian sedimentary cycles documented in the Mediterranean [Krijgsman et al., 1999a, 1999b; Hilgen and Krijgsman, 1999; Sierro et al., 2001]. The "tuned" time scale of Site 982 is used to reconstruct changes in late Miocene climate and global ice volume and assessthe role they played in the isolation of the Mediterranean,
2.2 2 1.8 1.6
deposition of the evaporitesl andthe reflooding of the basin at the Miocene/Pliocene boundary.
1.6
1.8
2
2.2
2.4
2.6
2.8
3
(•180C. kullenbergi
2. Methods
Cores within the Site 982 composite section were sampled at a constant interval of either 5 or 10 cm dependingupon sedimentationrate, resulting in an average temporal sampling spacing of-2500 years. Oxygen and carbon isotope ratios were measured on the benthic foraminifers Planulina wuellerstorfi and Cibicidoides kullenbergi. Previous studies have reported that the stable isotopic ratios of these two species are indistinguishable within analytical error. Measurementsof paired analyses of both species from the same samplesin Site 982 show a 1:1
correspondence for •80, but •3C valuesof C. kullenbergi divergefromP. wuellerstorfi at lower •]3C values(Figure
I::15 0
'
_.e 1 + +
0.5
1), suggesting that C. kullenbergi may sometimes live infaunally. For sampleswhere paired analyses were made, oxygen isotope values were averaged for each sample, which significantlyimproved the signal to noise ratio in the
upperMiocenesequence. For carbon isotopes,only •5•3C values measuredon P. wuellerstorfi were used. Samples between 122 and 138 meters composite depth (mcd) were analyzed for stable isotopes at Massachusetts Institute of Technology (MIT) and samplesbetween 138 and 285 mcd were measured at the University of Florida (UF).
o
-0.
-0.5
0
0.5
1
1.5
2
(•13C C. kullenbergi
Benthicforaminiferaltestswere cleanedin an ultrasonic Figure1. Oxygen andcarbon isotope measurements of specimens bath to remove fine-grained particles and soaked in either
of P. wuellerstorfiand C. kullenbergifrom the same samples. Note
thatthe trendfollowsa 1:1 relationshipfor /5•80but divergesfrom
15%H202orroastedin a vacuumto removeorganicmatter.
1:1 at lower/513C values,suggesting that C. kullenbergi may
The foraminiferal
sometimeslive infaunally.
calcite was reacted in a common acid
166
HODELL ET AL.'
ATLANTIC-MEDITERRANEAN
CORRELATIONS
Table 1. Depth-AgePoints Used for Initial Chronology and AdjustedAges of These Events After Tuninga
solution of La90 (1,1), and the astronomically tuned chronologyis consistentwith Hilgen et al. [1995] in the late
Datum
Miocene [Shackleton and Crowhurst, 1997]. Recognition in Site 982 of the coiling change in
Depth
Age
Tuned age
(mcd)
(Ma)
(Ma)
Neogloboquadrina acostaensis (from sinistral to dextral; Figure 2)is an important datum because it has been found
MIS
NS6
128.09
4.713
4.778
in
MIS
Si2
132.17
4.792
4.821
other
North
Atlantic
sites
in
the
middle
of
Chron
MIS
Si4
136.55
4.871
4.854
MIS
Si6
139.35
4.915
4.897
MIS
T8
156.32
5.211
5.218
MIS
TG12
172.78
5.540
5.506
MIS TG20
183.00
5.753
5.757
C3An. lr and is synchronous with the same event in the Mediterranean [Hooper and Weaver, 1987; Hodell et al., 1994; Sierro et al., 2001]. The coiling change has been dated at 6.36 Ma throughout the Mediterranean and occurs 17 precessional cycles below onset of evaporite deposition [Hilgen and Krijgsman, 1999]. The replacement of the
Event 4b
211.78
6.360
6.342
Globorotaliamenardii group by the Globorotaliarnioturnida
Event 3 c
248.29
7.240
7.246
Event 1d
257.94
7.512
--
group (event 3 of Sierro et al. [1993]) occurs near the Tortonian/Messinian boundary and is dated at 7.24 Ma throughout the Mediterranean. Last, the LaO of G.
aMarine isotopestages(MIS) designationsfollow Shackleton et al. [1995b]. The ages for TG20 and TG12 were derived from Site 926 [Shackleton and Crowhurst, 1997], whereas all others were taken from Site 846 [Shackleton et al., 1995b] and adjusted by adding41 kyr corresponding to one obliquity cycle. Foraminiferal events are described by Sierro et al. [1993], and the astronomical ages were derived from Hilgen et al. [1995].
(•180(Cibicidoides) 2
1.8 •
120
I
•
•
•
I
2.2 t
•
•
bFirstcommonoccurrence of dextralformsof Neogloboquadrina
aLast occurrence of G. menardii sinistral
2.6
2.8
•-NS6
acostaensis
eReplacement of the Globorotalia menardii group by the Globorotaliamiotumida group
2.4
.
-
140
_
_
t--"---
relative to Vienna Pee Dee Belemnite (VPDB) [Coplen,
1996]. • Analyticalprecisionfor all isotopeanalyseswas
160
better than _+0.1%o. Weight percent carbonate was measuredby coulometric titration using a UIC Inc. Model 5240 Total Inorganic Carbon (TIC) preparation system and
Model 5011 coulometertenglemannet al., 1985]. GRA
no data
• 180
TG20
bulk densitymeasurements weremadeaboardLeg 177 using a GEOTEK multi-sensor core logger [Shipboard Scientific Party, 1996].
'"
.s::
• 200
3. Chronologyand AstronomicalTuning No polarity reversal stratigraphy is available for the upper Miocene to lower Pliocene section of Site 982 because magnetization intensities were low and within the
noise level of the shipboard pass-through cryogenic magnetometer[Shipboard Scientific Party, 1996]. An initial chronology for Site 982 was constructedusing prominent marine isotopestages(Table 1 and Figure 2)[Shackleton et al., 1995] and three planktic foraminiferal events that have been calibrated previously to the astronomical time scale (Table 1) [Sierro et al., 2001; Hilgen et al., 1995]. For Pliocene isotope stages we used the published ages of Shackletonet al. [1995b]adjustedby 41 kyr, corresponding to one obliquity cycle [Lourens et al., 1996]. Because
Shackleton et al. [1995a]usedtheastronomical solution of
220
.
1stsin -dex coilingchange
240
0
50
100
% N. acostaensis (doxtraO Figure 2. (left) Coiling ratio of Neogloboquadrina acostaensis expressedas percent dextral to total specimens. The first coiling
change fromdominantly sinistral to dextral forms occurs at 211.78
Berger and Loutre [1991] and obtained a much younger
meters compositedepth (mcd) and has been dated in the
estimate for thebaseof theGilbertthandidKrijgsman et al.
Mediterranean at 6.36Ma [HilgenandKrijgsman, 1999].(right) Three-pointrunning mean of the benthic oxygen isotopic signal
[1999a], weused theages forlateMiocene isotope stagesshowing prominent isotope stages identified byShackleton etal.
TG20andTG12fromSite926[Shackleton andCrowhurst, [1995b] andused in constructing thepreliminary agemodel (see 1997]. The tuning of Site 926 signals used the orbital
Table1).
HODELL
ET AL.:
ATLANTIC-MEDITERRANEAN
benthic8]80and GRA bulk densitysignalswere tuned to the La90 u,]) astronomicalsolution[Laskaret al., 1993]by filteringthebenthic8•80 signalat 41 kyr and matchingit to
o
obliquity and filtering the GRA signal at 21 kyr and matching it to summer insolation at 65øN (Figures 4a-4c). The tuning was done iteratively in order to obtain the best
10 •
fit for both •]80 and GRA signals. Previousstudieshave shownthebenthic•80 signalto be dominatedby the 41-kyr
•'
120 140 160 180 200 220 240 260 •
obliquity cycle in the late Miocene and Pliocene [Hodell et
Depth (mcd)
al., 1994; Shackleton et al., 1995b; Shackleton and Hall, 1997; Tiedemann et al., 1994; Tiedemann and Franz, 1987].
Figure 3. Age-depthplot of biostratigraphicand isotopiccontrol points(solid circles) used to constructpreliminary age model of Site 982 and age-depthpointers(crosses) derivedafter tuning the record.
Interval
sedimentation
rates are shown for the
The GRA bulk density record in Site 982 is correlated with wt% carbonatecontent (Figure 5). In general, higher bulk density and carbonate content coincide with lower
initial
values of bulk carbonateand benthic •80, indicating
calibrationpoints(solid squares)and tunedrecord (solidline).
28/
'
' I
•
.
, •
167
menardii (sinistral) occurs just below the base of the compositesection at Site 982 and is datedat 7.51 Ma. As a starting point for astronomical tuning, depths (mcd) in Site 982 were converted to age by linearly interpolating between age-depth pairs in Table 1. Sedimentation rates average4.7 cm/kyr in the late Miocene and early Pliocene, and interval sedimentation rates of the tuned record vary by an orderof magnitudefrom --1 to 10 cm/kyr (Figure 3). The
-20 ::!
0
CORRELATIONS
'
'
'
Si6
:'
.•
'
:
.
I:
'
it
I
'
,
/ 21
PlioceneMio-
.:
....
•.
..
1.8 1.6
540
23
25 , ....
•
'
'
'
•
'
L6
• 500
• 440 400
ß
4ioo
4øo
sdoo
søo
sso
Age (rs) Figure4. Benthic•]•O signaltunedto obliquityand gammaray attenuation(GRA)bulk densityto summerinsolation at 65øNusingthe La90(•,•)astronomical solutionof Lask• et al. [!993]. TheGRA bulk densityrecordis normalizedto unit variance, and eccentricity values have been multiplied by 100 for illustrationpurposes. Prominent oxygen isotope stagesare identified in Figures 4a •d 4b accordingto Shackleton et al. [1995b]. UA16 to UA34, Y1 to Y14, and SI to SIII refer to cycle numbers in the upper Abad, Yes•es, and Sotbas Members of the Sorbas Basin of southern Spain [Sie•o et al., 2001; Kfijgsman et al., 2•1].
168
2.8
TG20
21
2.6
- 22
2.4
23.•
2.2
24
1.8 25
1.6
i
i
i
i
i
....
540
520
•
'• i 6.,
u,
4 _.
õ 500
• 480
'F•460
EE 440
on,•t.o.f • '•'~i -2•
420
4oo
5400 55'00 56'00 57'00 58'00 59'00 60'00 6100 62'00 6300 6400 Age (kyrs)
2.8 I ' ' ' " ' .,' 'Messinia 'Torton ,21 2.6
22
•": ir"r !,'.i ,,, ;•..,• a.i • :,;•..'L•;• ;:;!;i..'• ;•..': •:..'•b %,
, ,
•
,J
.: '
,;
:
' :'
'
', t :, t, ,; I I, '
2
24
1.8
25
1.6
........... i
i
540-
520••. ,i, 460:.• i: :, ,:. 500-
;:4 ,: i: ,, 480- ,,:, ;l:;.,,:' !
,',l!i:•. ::,,,,,: ::iii"
'|
440-
420400 6400
,
66'00
'
'68'00
'
7000
Age (kyrs) Figure 4. (continued)
'
72'00
7400
HODELL ET AL.: ATLANTIC-MEDITERRANEAN
CORRELATIONS
169
90
G• ,=.. .
!,
:.::. :1' :.
:i: :: :: •:
. 'o
, ....:
:;
:" '
{• GRAbulk density (raw)
1 I
160
85
170
I
I
180
190
I
200
210
Depth (mcd) Figure 5. Comparisonof GRA bulk density signal measured on whole-cores using a GEOTEK multisensor logger before (dashed line) and after (solid line) filtering to remove a 1.5-m cycle due to section breaks. The units of the filtered GRA bulk density signal are arbitrary and have been scaled to match the amplitude of the weight percent carbonate record. Note the general correspondencebetween carbonate content and GRA bulk density.
warmer surface and deep waters. We use this relationship as justification for tuning the GRA bulk density record to summer insolation at 65øN. Baumann and Huber [1999] adopteda similar approachfor the Plio-Pleistoceneof Site
fact, the correlation of the GRA bulk density and insolation recordswould be no different if we had used the filtered versus unfiltered GRA bulk density record. We used the GRA bulk density signal for tuning instead of the percent
982 by tuning the unfiltered carbonaterecord to highlatitude summer insolation. In addition, Beaufort and Aubry [1990] found major fluctuations in the nannofossil assemblagesin the late Miocene of nearby Deep Sea
CaCO3 recordbecausecore loggingwas done at continuous 1-cm increments as opposedto 5-cm spacing of percent CaCO3 measurements.In addition, the two recordsare not identical in that the GRA bulk density signal has more
Drilling Project (DSDP)Site 552 on the Rockall Bank (56øN,
variation at precessional frequencies than does percent
10øS)that were dominatedby the 20-and 100-kyrcyclesof
CaCO3(Figure6).
orbitalprecession and eccentricity,respectively.Sprovieri The accuracyof the tuning variesin differentparts of et al. [1999] tuned the planktic oxygen isotope and the recordas the amplitudeand frequencyof the oxygen nannofossilabundance variations in Site 552 to insolation
isotopeandGRA density sigfialschange. For example, the
usingLa90(•.•),but continuityof the sectioncannot be assured because it wasrecovered in a singlehole. The powerspectrumof the GRA bulk densitysignalin
tuningbetween-5.2 Ma (isotope stageT8)and 6.4 Ma (N. acostaensis coiling change)is exceptionallygoodbecause the GRA signalcontainsa strongprecession and obliquity
the depthdomain revealed a peak at 1.5 m that is not seen
component(Figures 4b and 6a). In addition, the benthic
in the powerspectrumof percentCaCO3. This cycle may
5•80signalis alsostrongin this interval and dominatedby
be an artifact of the fact that each 9.5-m ODP core is cut into six 1.5-m sections. Partially filled core liners can produce low GRA bulk density values near the tops and bottomsof sections when cores are analyzed with the GEOTEK multi sensorlogger. Although this artifact was corrected for on board ship by eliminating data from the first and last 5 cm of each section [ShipboardScientific Party, 1996], this treatmentapparentlydid not eliminate the effect entirely. To removethe 1.5-m cycle from the record, the GRA bulk density record was filtered in the depth domain at 1.5 m, and the filtered signal was subtracted before converting the signalto an age scale. The corrected GRA bulk density signal was then correlated to summer insolationat 65øN. Filtering the 1.5-m cycle from the GRA
obliquity(Figure6b). This critical periodincludesthe MSC that is the focusof this paper.
Continuity of the compositesection was confirmed for the upper 255 mcd of Site 982 only [ShipboardScientific Party, 1996], and below this level the section was recovered in a singlehole (982B). The chronologybelow255 mcd was derivedby correlatingthe benthic carbon isotoperecordsof Hole 982B and the Sa16 Briqueterie Section, northwestern Morocco [Hodell et al., 1994]. For this purpose, the paleomagneticreversalstratigraphyof the Sa16Section was recalculated using the ages given by Krijgsman et al. [1999a] to the base of Chron C3An.2n and Hilgen et al. [1995] thereafter. The carbon isotopic records of Site 982 and Sa16are easily correlatedbetween -9 and 7 becauseof
bulk densitysignaldoesnot alter the relationshipbetween a maximumin 513Cin Chron C4n followedby the late the GRA bulk density and CaCO3 signals (Figure 5).
In
Miocene carbon shift [Hodell et al., 1994].
170
HODELL
ET AL.'
ATLANTIC-MEDITERRANEAN
CORRELATIONS
t 1004il
7oll I :
: bulk .GRA density I ,400j i
50 - '
'
60-
20
'
2319
b
%CaCO,
*
,_ 300
. .
! 100
10 0
0
0
i 0.02: 0.04
0.06
0.08
o.1
0
:0.02.'
requ,ncy
0.04
'i
0.06
0.08
o.
frequ,ncy
2
,,enth,c,"'o,,enth,c 1.
1 0.
0.,5 ' 0
: 0.02'
-
i 0.04
0.0•
0.08
0.,•
0
0.02,
0.04, , 0.0•
0.08
o.
frequency
frequency
,
10
•
f
Bulk
Bulk
4•
3-
2_
1
•,1i•;;;•,,11
o
0.02
0.04
0.06
frequency
iii
0.08
,,11111111111
O.1
i
0.02
i
0.04
'
'
'
i
,
0.06
", .... '
i ',lrl-
0.08
-•! .... 1',,1
0.1
frequency
Figure 6. Spectralanalysisof proxy signalsin Site 982 for the interval between 5.21 and 6.36 Ma, i.e., betweenthe coiling direction change in N. acostaensis and marine isotopestage(MIS)T8. All signalswere linearly detrended, normalized to unit variance, and re-sampled at a constant time interval of 1 kyr. Spectra were calculated by Analyseties[Paillard et al., 1996]using the Blackman-Tukeymethodwith one-thirdlag and Barlett window.
HODELL ET AL.: ATLANTIC-MEDITERRANEAN
2
,
•
•
t
I
•
t
•
•
I
[
•
t
•
ß Pliocen•e .......
I
i
t
CORRELATIONS
•
•
171
I
3
Miocene. . .
1.5
2.5
• 1 •3C
2
•
