miocene to pleistocene osmium isotopic records of the ...grupo179/pdf/kuroda 2016.pdf · miocene to...
TRANSCRIPT
Miocene to Pleistocene osmium isotopic recordsof the Mediterranean sedimentsJunichiro Kuroda1, Francisco J. Jiménez-Espejo1, Tatsuo Nozaki1, Rocco Gennari2, Stefano Lugli3,Vinicio Manzi2, Marco Roveri2, Rachel Flecker4, Francisco J. Sierro5, Toshihiro Yoshimura1,6,Katsuhiko Suzuki1, and Naohiko Ohkouchi1
1Japan Agency for Marine-Earth Science and Technology, Yokosuka, Japan, 2Department of Physics and Earth Sciences“Macedonio Melloni”, University of Parma, Parma, Italy, 3Dipartimento di Scienze della Terra, Universitá degli Studi diModena e Reggio Emilia, Modena, Italy, 4School of Geographical Sciences, Cabot Institute, University of Bristol, Bristol, UK,5Department of Geology, University of Salamanca, Salamanca, Spain, 6Atmosphere and Ocean Research Institute, Universityof Tokyo, Kashiwa, Japan
Abstract In the late Miocene the Mediterranean Sea experienced a salinity crisis and thick sequences ofevaporites precipitated across the deep and marginal basins. In this study we report Os isotopic recordsfrom Deep Sea Drilling Project and Ocean Drilling Project cores in the Mediterranean: the Balearic Sea(Site 372), the Tyrrhenian Sea (Site 654), the Ionian Basin (Site 374), and the Florence Rise (Sites 375–376), aswell as Integrated Ocean Drilling Project Site U1387 in Gulf of Cadiz, North Atlantic. Pliocene-Pleistocenesediments at all sites show 187Os/188Os values close to that of the coeval ocean water, indicating thatthe Mediterranean was connected to the North Atlantic. Evaporitic sediments deposited during the latestMiocene, however, have 187Os/188Os values significantly lower than coeval ocean water values. The offsetof the Mediterranean evaporite 187Os/188Os is attributed to limited exchange with the North Atlanticduring the Messinian salinity crisis. The source of unradiogenic Os is likely to be weathering of ultramaficrocks (ophiolites) cropping out in the Mediterranean’s drainage basins. Based on a box model we estimatedthe amount of unradiogenic Os and the Atlantic-Mediterranean exchange rate to explain this offset. Osisotopic ratios of the pre-evaporite sediments in the western Mediterranean are almost identical to that ofthe coeval ocean water. In contrast, equivalent sediments from the Florence Rise have significantly lower187Os/188Os values. The offset in the Os isotopic ratio on the Florence Rise is attributed either to limitedwater exchange between eastern and western Mediterranean or to local effects associated withexhumation of the Troodos ophiolites (Cyprus).
1. Introduction
At the end of the Miocene the Mediterranean Sea experienced an extraordinary event known as theMessinian salinity crisis (MSC) [see Hsü et al., 1973; Krijgsman et al., 1999; Rouchy and Caruso, 2006; CommissionInternationale pour l’Exploration Scientifique de la Méditerranée (CEISM), 2008; Ryan, 2009; Manzi et al., 2012;Roveri et al., 2014a, and references therein]. This saw the precipitation of more than 1 million km3 of evaporitesacross the Mediterranean basin formed in less than ~700 kyrs (Figure 1). A thick salt layer buried beneath theMediterranean sea floor has been identified on seismic data, and the topmost part of these deep basinalevaporites were sampled during the Deep Sea Drilling Project (DSDP) Legs 13 and 42A and Ocean DrillingProject (ODP) Legs 107, 160, and 161. The evaporite lithologies (gypsum, anhydrite, and halite) interbeddedwith dolomitic marls, combined with their substantial thickness preserved in the Mediterranean depths,require a significantly more restricted connection with the Atlantic Ocean [e.g., Meijer, 2006].
Changes in hydrological budget of the Mediterranean Sea, its connection with the Atlantic Ocean andmixingwith nonmarine waters have been examined using Sr isotopic records (87Sr/86Sr) from Miocene-Pliocenesediments in the Mediterranean basins [McKenzie et al., 1988; McCulloch and De Deckker, 1989; Müller et al.,1990; Müller and Mueller, 1991; Müller, 1993; Fortuin et al., 1995; Montanari et al., 1997; Flecker and Ellam,1999, 2006; Flecker et al., 2002; Sprovieri et al., 2003; Lugli et al., 2010; Roveri et al., 2014a]. These data showstrong divergence between Mediterranean and global ocean ratios both during the MSC and up to 3 millionyears prior to the first evaporite precipitates in the Mediterranean’s marginal basins. This indicates that theproportion of Atlantic water reaching the Mediterranean (and or its subbasins) was significantly reduced[Flecker et al., 2002; Topper et al., 2011].
KURODA ET AL. OS ISOTOPE RECORD OF MEDITERRANEAN 148
PUBLICATIONSPaleoceanography
RESEARCH ARTICLE10.1002/2015PA002853
Key Points:• Os isotope ratio of the MediterraneanSea sediments registers the history ofwater exchange
• Unradiogenic Os isotope ratios ofMessinian evaporite suggest limitedinflow from the Atlantic
• The unradiogenic Os may besupplied from ultramafic rocks inthe drainage basins
Supporting Information:• Table S1
Correspondence to:J. Kuroda,[email protected]
Citation:Kuroda, J., et al. (2016), Miocene toPleistocene osmium isotopic recordsof the Mediterranean sediments,Paleoceanography, 31, 148–166,doi:10.1002/2015PA002853.
Received 2 JUL 2015Accepted 15 DEC 2015Accepted article online 18 DEC 2015Published online 21 JAN 2016
©2015. American Geophysical Union.All Rights Reserved.
In this paper we investigate the hydrological history of the Mediterranean Sea from the middle Miocene to thePleistocene using the Os isotopic composition (187Os/188Os) of marine sediments as a potential tracer of hydrolo-gical connectivity between seas and oceans [Paquay and Ravizza, 2012; Du Vivier et al., 2014, 2015]. The187Os/188Os of seawater varies in response to input of Os from three major sources: (1) radiogenic Os(187Os/188Os ~1.4) from ancient continental crust, (2) unradiogenic Os (187Os/188Os ~0.12 to 0.15) from juvenilemantle via mid-ocean ridge and intraplate volcanisms, and (3) unradiogenic Os (187Os/188Os ~0.125) fromextraterrestrial source via partial dissolution of cosmic dust [Peucker-Ehrenbrink and Ravizza, 2000]. The averageresidence time of Os in the present ocean has been estimated about ~8kyr [Oxburgh, 2001] to 25–54kyr[Levasseur et al., 1999], which means that the isotopic composition of dissolved Os is nearly homogeneousthroughout the global ocean (present-day value of ~1.06±0.04) [Peucker-Ehrenbrink and Ravizza, 2000] but varieson the order of 104 years. The evolution of ocean water 187Os/188Os has been reconstructed throughout theNeogene based on Os isotopic compositions of marine sediments from the Atlantic [Reusch et al., 1998; Klemmet al., 2008] and Pacific [Pegram et al., 1992; Ravizza, 1993; Peucker-Ehrenbrink et al., 1995; Pegram and Turekian,1999; Reusch et al., 1998] oceans, which experienced a gradual increase from ~0.8 in the middle Miocene to~1.0 in the latest Pleistocene. In addition to the three sources, weathering of ultramafic rock bodies such as ophio-lite can also supply unradiogenic Os to the ocean. For example, Os concentrations of chromitites and peridotitesare typically on the order of 10 to 100ngg�1 and 1ngg�1, respectively, and their 187Os/188Os values are typically0.12–0.15 [Peucker-Ehrenbrink and Ravizza, 2000; Walker et al., 2002]. The Mediterranean’s tectonic history hasresulted in the formation and preservation of a substantial number of ophiolites around its margins (Figure 1),e.g., Cyprus, Turkey, Greece, and Albanian ophiolites in the eastern Mediterranean [Robertson, 1977, 2002, 2004;Koepke et al., 2002; Dilek and Furnes, 2009], Ligurian ophiolite [Rampone et al., 2005] in the Apennines in theTyrrhenian Sea and Adriatic Sea, and peridotite bodies in the Betic-Rif orogenic belt [Zeck, 1997] in the westernMediterranean. As a result, the Os isotopic composition of Mediterranean seawater should be sensitive to thesesources of unradiogenic Os during restriction and cessation of Mediterranean-Atlantic exchange.
We present Os isotopic data, together with major and trace element composition from four different typesof sediments: marl (or marlstone), mud (mudstone), gypsum (or anhydrite), and halite. These samples were
Figure 1. Map of the Mediterranean Sea with distributions of the Messinian evaporites (after Ryan [2009] and Manzi et al. [2012]) and locations of drill sites fromwhich materials used in this study were taken. Dark green areas indicate ophiolite exposures [Koepke et al., 2002; Robertson, 2004; Dilek and Furnes, 2009], andgray areas indicate Betic-Rif complex [Coward and Dietrich, 1989; Zeck, 1997].
Paleoceanography 10.1002/2015PA002853
KURODA ET AL. OS ISOTOPE RECORD OF MEDITERRANEAN 149
taken from Deep Sea Drilling Project (DSDP) and Ocean Drilling Program (ODP) sediment cores: Site 372 atthe margin of the Balearic Basin in the western Mediterranean, Site 654 at the Tyrrhenian Basin, Site 374 atthe Ionian Basin, and Sites 375 and 376 at the flank of the Florence Rise in the eastern Mediterranean(Figure 1). We also present Os isotope data measured on hemipelagic calcareous mudstone from SiteU1387C in the Gulf of Cadiz in the eastern North Atlantic, which were collected during the IntegratedOcean Drilling Project (IODP) Expedition 339 to investigate the evolution of the Mediterranean OutflowWater from the late Miocene to the Holocene. We discuss the hydrological and paleoceanographic implica-tions of these Os isotopic data and compare the Os isotopic data with the previously published Sr isotopicdata to reconstruct the hydrological evolution of the Mediterranean Sea and its connectivity with theAtlantic Ocean.
2. DSDP/ODP Core Samples2.1. Site 372
Sediment core was recovered at Site 372 on the deep part of the western Balearic Basin, westernMediterranean (40°01.86′N, 4°47.79′E, at 2699m water depth; Figure 1) [Hsü et al., 1978]. The top 150mcomprise Plio-Quaternary nannofossil marl with minor terrigenous sand-silt layers (Lithologic Unit I), whichunconformably overly a ~50m thick succession of primary cumulative and clastic gypsum [Lugli et al.,2015], and dolomitic marls (Lithologic Unit II) (Figure 2). Unit II at this site represents a pinch-out wedge ofa seismic unit that is ascribed to the “Upper Evaporite” seismic unit (sensu Lofi et al. [2011]). Conversely, basedon Sr isotope record Unit II could be separated in a lower portion belonging to stage 2 and an upper onebelonging to stage 3 of the MSC, 5.60–5.55 and 5.55–5.33Ma, respectively [Roveri et al., 2014b]. A significantstratigraphic hiatus at 200mbsf marks the base of the evaporites overlying Unit III, the latter consists of~270m thick succession of late Burdigalian to late Serravallian nannofossil marls. These sediments weredeposited in an open marine environment without coarse detrital input. Sediments below 468mbsf(Unit IV) show a sudden drop in carbonate content and an increase in the terrigenous component.Paleobathymetric estimates based on benthic foraminiferal assemblages indicate deposition at water depthsof ~900m during the early Burdigalian and in excess of 1500m during the late Burdigalian [Wright, 1978; Hsüet al., 1978].
We chose nine samples from Site 372 (Table 1). Three marl samples were taken from the Pliocene interval(Unit I) between 136 and 144mbsf. A nannofossil marl sample was taken from the top of Unit II near the
Figure 2. Age, lithology, concentrations of Re (open) and Os (filled), andmeasured 187Os/188Os values (open) and initial 187Os/188Os values (filled) in the drilled coresamples from DSDP/ODP/IODP sites. (left to right) Site U1387, Gulf of Cadiz, Site 372, Balearic Sea, Site 654, Tyrrhenian Sea, Site 374, Ionian Basin, and Site 375and 376, Florence Rise. Age and lithology of Sites 372, 374, 375, and 376 are after Hsü et al. [1978], Site 654 are after Shipboard Scientific Party [1987], and Site U1387are after Expedition 339 Scientists [2013], respectively.
Paleoceanography 10.1002/2015PA002853
KURODA ET AL. OS ISOTOPE RECORD OF MEDITERRANEAN 150
Table
1.Dep
th,A
gean
dLitholog
iesof
Samples,and
Rean
dOsCon
centratio
nsan
d187Os/188Os(M
easuredan
dInitial)V
alue
sa
Leg
Site
Core
Section
Interval(cm)
Dep
th(m
bsf)
Age
Litholog
y[Os]
1σ[Re]
1σ
187Os/188Os
1σ
187Re
/188Os
1σ
187Os/188Os
1σTo
pBo
ttom
Top
Bottom
(Ma)
(pgg�
1)
(pgg�
1)
measured
measured
Initial
BalearicBa
sin,
Western
Mediterran
ean
42A
372
24W
9193
136.41
136.43
2.5
Marl
45.73
0.28
82.3
0.4
0.81
80.00
89.44
0.07
0.81
70.00
842
A37
23
2W
4446
142.44
142.46
3.1
Marl
37.42
0.25
639.9
2.6
0.90
80.00
990
.60.7
0.90
30.00
942
A37
23
3W
7274
144.22
144.24
3.8
Marl
85.6
0.5
971.2
2.5
0.94
30.00
760
.50.4
0.94
00.00
742
A37
24
1W
9193
150.91
150.93
5.2
Marl
57.2
0.4
263.2
0.5
0.81
80.00
924
.14
0.19
0.81
60.00
942
A37
28
1W
121
123
189.21
189.23
5.6
Gyp
sum
(laminated
)7.22
0.05
8.99
0.11
0.76
70.01
56.48
0.09
0.76
70.01
542
A37
28
2W
7779
190.27
190.29
5.6
Gyp
sum
(laminated
)4.64
0.04
14.3
0.4
0.75
30.01
716
.00.5
0.75
20.01
742
A37
210
2W
8283
.520
9.32
209.34
11.5
Marl
52.1
0.4
112.9
0.8
0.79
80.00
811
.34
0.12
0.79
60.00
842
A37
217
5W
4042
279.90
279.92
12.0
Marl
59.9
0.4
2092
.33.1
0.83
90.00
818
3.9
1.2
0.80
20.00
842
A37
233
1W
3739
463.87
463.89
17.6
Marl(stone
)62
.80.5
9508
121.12
20.01
282
26
0.88
10.01
2
Tyrrhenian
Sea
107
654A
16R
4W
4546
.514
1.75
141.77
3.2
Marl
84.8
0.9
129.91
0.29
0.84
80.01
08.07
0.08
0.84
80.01
010
765
4A
20R
2W
102
103.5
177.52
177.54
3.9
Marl
69.0
0.7
73.65
0.34
0.75
20.01
05.56
0.06
0.75
10.01
010
765
4A
25R
5W
1314
.522
9.23
229.25
4.3
Marl
154.3
2.3
144.2
0.4
0.88
80.01
04.94
0.07
0.88
70.01
010
765
4A
33R
1W
3031
.528
7.90
287.92
5.5
Gyp
sum
(gyp
siferous
mud
ston
e)18
5.4
2.2
1509
215
0.83
50.00
742
85
0.79
60.00
7
107
654A
41R
2W
6566
.536
0.85
360.87
7.0
Marl
107.4
1.3
2537
.51.3
0.74
10.00
912
2.9
1.5
0.72
70.00
910
765
4A
42R
2W
9495
.536
9.84
369.86
7.5
Marl
112.5
1.3
2789
.61.5
0.75
00.00
812
9.1
1.4
0.73
40.00
810
765
4A
45R
2W
1415
.539
8.04
398.06
7.9
Marl
106.7
1.2
2594
.31.4
0.66
50.00
812
5.3
1.4
0.64
80.00
8
Ionian
Basin
42A
374
52W
8990
.529
9.39
299.41
1.9
Marl
26.00
0.21
668.1
2.0
0.91
60.01
313
6.4
1.2
0.91
20.01
342
A37
47
2W
5153
342.01
342.03
3.2
Marl
23.4
0.4
55.43
0.22
0.92
50.03
212
.57
0.21
0.92
40.03
242
A37
48
3W
6667
.535
3.16
353.18
3.9
Marl
51.9
0.4
1278
50.88
60.00
813
0.2
1.0
0.87
70.00
842
A37
49
4W
4749
363.97
363.99
4.2
Marl
32.82
0.18
34.37
0.12
0.79
30.00
75.48
00.03
50.79
30.00
742
A37
410
1W
1314
.537
1.63
371.65
4.4
Marl
31.11
0.21
73.3
0.6
0.90
80.01
012
.50
0.14
0.90
70.01
042
A37
416
1W
120
122
407.70
407.72
5.4
Gyp
sum
(darkcolor)
8.56
0.06
1730
40.76
20.01
410
528
0.66
70.01
442
A37
420
1W
143
145
421.43
421.45
5.5
Gyp
sum
14.83
0.08
4241
130.75
40.01
014
8810
0.61
70.01
042
A37
422
2W
113
115
437.63
437.65
5.6
Halite
1.08
80.01
742
.72
0.20
0.47
00.02
219
6.2
3.1
0.45
10.02
242
A37
422
3W
1215
438.12
438.15
5.6
Halite
2.15
60.03
48.21
0.08
0.33
60.01
518
.80.4
0.33
40.01
5
Florence
Rise,Eastern
Mediterran
ean
42A
375
11W
8486
138.34
138.36
5.4
Gyp
sum
5.94
0.06
8.31
0.18
0.40
90.01
16.98
0.17
0.40
80.01
142
A37
52
4W
9910
119
3.89
193.91
5.5
Gyp
sum
(darkcolor)
10.38
0.09
1404
.72.6
0.70
50.01
270
06
0.64
10.01
242
A37
54
4W
4850
249.48
249.50
7.3
Marl(stone
)13
0.2
1.4
1196
.21.6
0.36
40.00
445
.60.5
0.35
90.00
442
A37
56
5W
8587
467.85
467.87
9.0
Marl(stone
)15
5.3
1.8
2297
40.39
90.00
473
.80.9
0.38
70.00
442
A37
59
3W
8789
654.37
654.39
13.5
Marl(stone
)63
.70.4
393.9
0.4
0.56
10.00
531
.46
0.21
0.55
40.00
5
Florence
Rise,Eastern
Mediterran
ean
42A
376
55W
140
142
41.90
41.92
4.5
Marl
35.53
0.32
172.1
0.5
0.86
50.01
225
.55
0.24
0.86
30.01
2
Paleoceanography 10.1002/2015PA002853
KURODA ET AL. OS ISOTOPE RECORD OF MEDITERRANEAN 151
boundary between the Messinian and Pliocene(~151mbsf) [Hsü et al., 1978]. Two samples oflaminated gypsum were taken from the lower partof Unit II (189–190mbsf), which would have beendeposited during the second stage of MSC(5.60–5.55Ma) [Roveri et al., 2014b]. Marl and marl-stone samples were taken from the middle MioceneUnit III (Serravallian) and early Miocene Unit IV(Burdigalian). The ages of the pre-Messinian sampleswere estimated using the sedimentation rates calcu-lated by Abdul et al. [2008].
2.2. Site 654A
This site is located on the upper Sardinian marginof the Tyrrhenian Sea (40°34.76′N, 10°41.80′E, at2208m water depth; Figure 1). The top 243m of thesedimentary succession (Unit I) is Plio-Pleistocene inage, and dominated by nannofossil ooze and marl,and minor terrigenous components, volcanic ash,sapropel layers, and a single interval of basalt(Figure 2) [Shipboard Scientific Party, 1987]. Its lower-most part, of Zanclean age, contains a foraminiferrecord that shows an oceanic Sr isotope signature[Müller and Mueller, 1991]. Lithologic Unit II (243–313mbsf) is Messinian in age and comprises primarycumulative and clastic gypsum [Lugli et al., 2015] andgypsiferous mudstone interbedded with calcareousmudstone, minor sandstone, breccia, dolostone, andanhydrite beds [Shipboard Scientific Party, 1987].Based on seismic, sedimentological and Sr isotopeinvestigations, this unit can be separated in an upperportion belonging to the stage 3 of the MSC, andlower portion ascribed to the stage 2 of MSC [Roveriet al., 2014b]. This unit is floored by an angularunconformity and thickens to the west. LithologicUnit III (313–349mbsf), which is dated as early to mid-dle Messinian, is dominated by laminated organiccarbon-rich mudstone (sapropel) and dolomitic mud-stone with minor volcanic ash. Because of theabsence of calcareous microfossils, this unit has beeninterpreted as a deepwater equivalent of MSC stage 1evaporites (5.97–5.60Ma) [Manzi et al., 2013; Roveriet al., 2014b]. A slumped interval is present at thebase of Unit III, separating it from the underlyinghomogenous nannofossil ooze/chalk (Unit IV, 349–404mbsf). This unit is the late Tortonian to earliestMessinian in age [Shipboard Scientific Party, 1987].Benthic foraminiferal assemblages in Unit IV suggestthat water depth in the Tortonian was shallowand became deeper into the Messinian [ShipboardScientific Party, 1987]. The Sr isotope values measuredon foraminifera which are slightly lower than coevalseawater values have been interpreted as beingcaused by fluvial runoff [Müller and Mueller, 1991] orTa
ble
1.(con
tinue
d)
Leg
Site
Core
Section
Interval(cm)
Dep
th(m
bsf)
Age
Litholog
y[Os]
1σ[Re]
1σ
187Os/188Os
1σ
187Re
/188Os
1σ
187Os/188Os
1σTo
pBo
ttom
Top
Bottom
(Ma)
(pgg�
1)
(pgg�
1)
measured
measured
Initial
42A
376
64W
3032
50.30
50.32
5.1
Marl
66.7
0.5
4954
70.75
60.00
938
6.6
3.1
0.72
30.01
042
A37
619
1W
140
143
170.40
170.43
5.5
Gyp
sum
5.92
0.07
47.59
0.23
0.54
80.01
540
.70.5
0.54
40.01
542
A37
620
1W
7375
185.73
185.75
5.6
Gyp
sum
6.15
0.06
579.6
1.1
0.69
90.01
548
65
0.65
40.01
542
A37
622
1W
6870
205.68
205.70
5.6
Halite
0.47
10.01
638
.37
0.12
0.46
10.03
340
314
0.42
40.03
442
A37
622
1W
140
142
206.40
206.42
5.6
Halite
0.27
50.01
214
.32
0.10
0.30
50.03
025
411
0.28
10.03
0
Gulfof
Cadiz,North
Atla
ntic
339U13
87C
32R
1W
7577
578.25
578.27
3.8
Calcareou
smud
ston
e(>
180μm
)25
9.4
3.3
1209
524
0.92
10.01
324
7.8
3.2
0.90
60.01
3
339U13
87C
51R
5W
4850
760.48
760.50
5.5
Calcareou
smud
ston
e(>
180μm
)13
9.0
1.5
1696
40.87
60.01
364
.40.7
0.87
00.01
3
339U13
87C
58R
5W
4951
827.69
827.71
5.6
Calcareou
smud
ston
e(>
180μm
)13
3421
1809
100.88
10.01
27.17
0.12
0.88
10.01
2
a Blank
contrib
utionwas
0.2–
2%forOsan
d0.03
–6%
forRe
inmarlsam
ples,0
.4–6
%forOsan
d0.04
–13%
forRe
ingy
psum
samples,and
9–47
%forOsan
d3–
11%
forRe
inha
litesamples,
respectiv
ely.
Paleoceanography 10.1002/2015PA002853
KURODA ET AL. OS ISOTOPE RECORD OF MEDITERRANEAN 152
alteration of volcanic materials [Müller et al., 1990] impacting the Sr isotope ratio of a restricted marine basin.Unit V (404–418mbsf) is characterized by glauconitic sandstone and marl with calcareous bioclasts andthick-walled oyster shells at the bottom of this unit. The Sr isotopic values measured on foraminifers ofUnit V plots on the oceanic curve [Müller and Mueller, 1991]. The lowest part of the core (416–474mbsf)comprises gravel-bearing calcareous mudstones, underlain by conglomerate and pebbly mudstone (Unit VI).The age of this unit is not clear.
From Site 654A, we chose seven samples (Table 1). Three Pliocene samples of marl were taken from Unit I(between 141 and 229mbsf, 4.3 to 3.2Ma) [Shipboard Scientific Party, 1987]. One gypsiferous mudstonesample ofMSC (Messinian Salinity Crisis) stage 3 was taken fromUnit II (~288mbsf). We took threemarl samplesfrom Unit IV (361–398mbsf), ranging in age from the late Tortonian (~7.9Ma) to earliest Messinian (~7.0Ma).
2.3. Site 374
Site 374 is located in the deep part of the Ionian Basins (35°50.87′N, 18°11.78′E, at 4078m water depth;Figure 1) [Hsü et al., 1978]. The top 373m of the core are composed mainly of nannofossil marls and mudswith intercalations of distal turbidites and sapropels of Plio-Pleistocene age (Unit I), which overlies an 8.5mthick dolomitized pelagic marl of earliest Pliocene age (Unit II) (Figure 2). The underlying Messinian sedimentsinclude three subunits [Hsü et al., 1978], consisting, respectively, of homogeneous dolomitic mudstone(Subunit IIIa), an alternation of dolomitic mudstone and clastic/cumulative gypsum [Lugli et al., 2015] (SubunitIIIb), and massive anhydrite underlain by blocky halite with minor kainite, sulfoborite, sylvite, and bischovite atthe bottom of the hole (Subunit IIIc). Although the Messinian units have previously been correlated to the“Upper Evaporite” seismic unit, the two subunits are characterized by different Sr isotope ratios, which ledRoveri et al. [2014b] to assign subunits IIIa and IIIb to MSC stages 3 and 2, respectively.
At this site we selected nine samples (Table 1). Two samples were taken from Unit I: one from near the bound-ary between upper and lower Pliocene (342mbsf) and the other from the lower Pliocene (364mbsf). Twocrystalline gypsum samples and two halite samples were taken from Unit III between 408 and 438mbsf.
2.4. Site 375
Site 375 is on the Florence Rise in the eastern Mediterranean (34°45.74′N, 31°45.58′E, and 1900mwater depth;Figure 1). Hsü et al. [1978] identified eleven lithological units across two adjacent sites (375 and 376)(Figure 2). Not all these unites are recovered at Site 375 which includes nannofossil marls with sapropelinterlayers of Pliocene age (Unit III): marlstones with interbedded sand- and siltstones that are thought tobe latest Messinian age (Unit V) [Hsü et al., 1978]; clastic gypsum and dolomitic marlstones (Unit VI) whichare correlated to the upper Evaporites of MSC stage 3 (5.60–5.33Ma) [CIESM, 2008; Roveri et al., 2014a];marlstones with interbedded sand- and siltstones of the Tortonian age (Unit VII); Serravallian variegatedmarlstones and limestones (Unit VIII); dark nannofossil marlstones of Serravallian age (Unit IX); Langhianreddish-brown nannofossil marlstones with interbedded limestones (Unit X); and dark limestones and marl-stones of Burdigalian age (Unit XI).
Five samples were taken from Site 375 (Table 1). Two gypsum samples were from the Messinian interval ofUnit VI (138–194mbsf). Three marlstone samples were taken from the pre-Messinian intervals; two of them(~249 and ~468mbsf) are Tortonian and the oldest sample (654mbsf) is early Serravallian.
2.5. Site 376
Site 376 was drilled on the Antalya Basin flank of the Florence Rise (34°52.32′N, 31°48.45′E, and 2101m waterdepth; Figure 1). Sediments recovered from this site include Quaternary nannofossil marls with tephra andsapropel layers in top 38.4m (Unit I), a 4.2m thick nannofossil marl with minor sapropelic marl layers of latePliocene age (Unit II), a 5.5m thick nannofossil marl with minor sapropelic marl layers of early to late Plioceneage (Unit III), a 40 cm thick slumpedmarls of Miocene and early Pliocene age (Unit IV), and evaporites (Unit VI)at the bottom of the hole (140.5–216.5mbsf) (Figure 2) [Hsü et al., 1978]. Unit VI is divided into two subunits: a47.5m thick interval of clastic gypsum and dolomitic marlstone (Subunit VI-a) and the basal 28.5m interval ofentherolithic anhydrite and blocky halite [Lugli et al., 2015] (Subunit VI-b). Subunit VI-a at this site has beencorrelated with Unit VI at Site 375 [Hsü et al., 1978] and is thought to be equivalent to stage 3 UpperEvaporites in marginal basin settings [CIESM, 2008; Roveri et al., 2014b]. The Subunit VI-b is thought to havedeposited during MSC stage 2 based on its Sr isotopic record [Roveri et al., 2014b].
Paleoceanography 10.1002/2015PA002853
KURODA ET AL. OS ISOTOPE RECORD OF MEDITERRANEAN 153
Five samples from Site 376 were analyzed (Table 1). A calcareous marl sample (~50mbsf) was from Unit IV[Hsü et al., 1978]. Since this sample is likely to have been deposited after the MSC, we consider that itsage is somewhere between 5.3 and 5.0Ma. Two gypsum samples were taken from Subunit VI-a (170 and186mbsf), and two halite samples were taken from Subunit VI-b (~206mbsf).
2.6. Site U1387
Site U1387 is located in the Atlantic on the southern Iberian margin in the Gulf of Cadiz (36°48.32′N, 07°43.13′W,559m water depth; Figure 1). Site U1387 recovered a succession of contourite deposits developed along themiddle slope over the past ~5Ma under the direct influence of the Mediterranean outflow water that flowedout through the Strait of Gibraltar gateway [Expedition 339 Scientists, 2013; Hernández-Molina et al., 2014].
The sedimentary succession recovered at Site U1387 is divided into four lithologic units (Figure 2). APleistocene-Holocene sequence composes calcareous mud with bioclastic sandy and silty intervals. A signif-icant unconformity straddling 3.2 to 1.8Ma marks the boundary Units I and II at 450mbsf with dolomitehorizons. Unit II represents Pliocene sediments characterized by a clear cyclic arrangement of lighter coloredsandy sediments and darker colored muddy sediments. Sandy sediments are more abundant in Unit III(600–730mbsf) which is interpreted as the product of dense turbidity currents, debris flows and slump.Below ~680mbsf, a 50m thick section includes consolidated sandstone. This unit IV (730mbsf to thebottom), which is mainly the early Pliocene in age, but may start in the latest Miocene, comprises calcareousmud and ooze of Messinian age (<5.8Ma) [Expedition 339 Scientists, 2013].
We sampled mainly the calcareous mud sediments in the lower part of Hole U1387C (Table 1). A Pliocenecalcareous mud from the bottom of the Lithologic Unit II was analyzed (578mbsf) [Expedition 339 Scientists,2013]. Two other samples were taken from the bottom part of Hole U1387C (761 and 828mbsf). The age ofthose samples is most likely to be Messinian [Expedition 339 Scientists, 2013].
3. Analytical Methods
Mediterranean Sea samples were crushed to coarse fragments from which we handpicked fresh fragmentswithout veins, nodules, or weathered surfaces. The fresh fragments were ultrasonically washed with ~0.5MHNO3, rinsed 3–5 times with deionized water Milli-Q, dried and then crushed in an agate mortar.
The calcareous mud samples from Hole U1387C contained abundant terrigenous minerals (clay, quartz, feld-spars, and mica). We collected coarse fragments by sieving with 180μm mesh. Based on the microscopicobservation we confirmed that the coarse fractions were dominated by planktic and benthic foraminifers.This process could minimize potential contamination of Re and Os from terrigenous fraction so that hydro-genous Re and Os were extracted from the sediment to investigate hydrology of Mediterranean Sea.
Rhenium and osmium concentrations and isotope compositions weremeasured using themethod established byNozaki et al. [2012] and Tokumaru et al. [2015] which combines isotope dilution multicollector inductively coupledplasma-mass spectrometry (ID-MC-ICP-MS) with Carius tube digestion [Shirey and Walker, 1995]. Re and Os wereextracted from sediment core samples using inverse aqua regia digestion. After spiking the samples with solutionsenriched in 190Os and 185Re, ~1g of powdered sample was sealed in a Carius tube with 4mL of inverse aqua regiaand heated at 230°C for 24h. After centrifugation to remove residue, the leachate solution was diluted with 15mLof deionized Milli-Q water. Os isotope ratios were measured by a MC-ICP-MS (Themo Fisher Scientific NEPTUNE)combined with Ar gas sparging introduction [Hassler et al., 2000; Schoenberg et al., 2000]. The instrumental massfractionation of Os was corrected by means of the sample-calibrator bracketing method with a diluted Os stan-dard solution (Johnson Matthey Company Alfa Aesar ICP Os standard solution) containing ~50pg of total Os withthe 187Os/188Os value of 0.106838±0.000015 (1σ), whose valuewas certified by Negative Thermal IonizationMassSpectrometry [Nozaki et al., 2012]. After the Os analysis, Re was separated from the solution by anion-exchangecolumn purification [Morgan et al., 1991]. Re isotope ratios were analyzed using the NEPTUNE with solutionnebulization. As Re has only two isotopes and instrumental mass fractionation cannot be corrected internally,the measured Re isotope ratio was corrected by sample-calibrator bracketing using a Re standard solution(1000mg l�1, Merck Millipore Co. Ltd.) which was diluted to 85.8pgg�1 (measured raw 187Re/185Re valuesof 1.6908 to 1.6931), against an 187Re/185Re ratio of 1.6738 [Gramlich et al., 1973]. Total procedural blanksrange between 0.28 and 0.43pg for Os, between 2.0 and 3.0pg for Re, and between 0.128 and 0.193 for187Os/188Os, respectively. The Os procedural blanks are substantially smaller than the Os concentration in the
Paleoceanography 10.1002/2015PA002853
KURODA ET AL. OS ISOTOPE RECORD OF MEDITERRANEAN 154
samples analyzed. However, some samples contain extremely low quantities of Re, on the same order as the Reprocedural blank. The average 187Os/188Os value measured for the rock reference material JLs-1 (limestone),issued by the Geological Survey of Japan (GSJ), was 1.077±0.029 (2σ, n=6). We calculated the initial Os isotopicratios (187Os/188Osi) based on the measured 187Os/188Os and 187Re/188Os values, estimated age of each sample,and the decay constant of 187Re of 1.666×10�11 y�1 [Smoliar et al., 1996].
Major and trace element concentrations were measured by an ICP-MS (Agilent 7500ce). For the elementanalyses we used the same sample solutions that were used for Re and Os isotope analyses to investigatethe host minerals of Re and Os extracted with the inverse aqua regia digestion method. After drying, eachsample solution was dissolved with 3% HNO3. Analyses were generally within 3% of the accepted valuesof the GSJ reference materials, and the reproducibility of replicate analyses of the GSJ reference materialswas typically <5% RSD for major elements and <3% RSD for trace elements, respectively.
4. Results and Discussion4.1. Major and Trace Element Abundances
Marl samples contain 5.94 to 16.2wt % Ca (Table 2), which indicates that carbonate is the most abundantcomponent in the marl samples. Marl samples also contain 0.903–4.96wt % Al (Table 2), suggesting that
Table 2. Major Element Concentrations of Leachates Dissolved With the Inverse Aqua Regia Digestion Methoda
Leg Site Core Section
Interval (cm) Depth (mbsf)
Lithology
Na Mg Al K Ca Mn Fe
Top Bottom Top Bottom (wt %) (wt %) (wt %) (wt %) (wt %) (wt %) (wt %)
Balearic Basin, Western Mediterranean42A 372 2 4 W 91 93 136.41 136.43 Marl 0.631 0.669 2.70 1.05 12.3 0.0544 1.8942A 372 3 2 W 44 46 142.44 142.46 Marl 0.685 0.747 3.03 1.16 9.79 0.0518 1.6242A 372 3 3 W 72 74 144.22 144.24 Marl 0.663 0.875 3.15 1.26 8.21 0.0455 2.1742A 372 4 1 W 91 93 150.91 150.93 Marl 0.581 1.52 3.81 1.37 9.22 0.0649 2.2242A 372 8 1 W 121 123 189.21 189.23 Gypsum (laminated) 0.0610 0.128 0.235 0.107 1.50 0.00379 0.13142A 372 8 2 W 77 79 190.27 190.29 Gypsum (laminated) 0.0562 0.114 0.0671 0.0290 1.40 0.00204 0.053942A 372 10 2 W 82 83.5 209.32 209.34 Marl 0.778 1.17 3.85 1.49 7.91 0.0511 1.9742A 372 17 5 W 40 42 279.90 279.92 Marl 0.842 1.28 3.39 1.36 10.1 0.0466 2.0042A 372 33 1 W 37 39 463.87 463.89 Marl(stone) 1.011 1.24 2.32 1.04 13.4 0.0172 1.6042A 372 33 1 W 37 39 463.87 463.89 Marl(stone) 1.03 1.19 2.31 1.26 13.0 0.0173 1.55
Ionian Basin42A 374 5 2 W 89 90.5 299.39 299.41 Marl 0.632 1.05 1.53 0.801 16.2 0.109 1.3342A 374 7 2 W 51 53 342.01 342.03 Marl 0.373 1.25 1.36 0.540 12.2 0.0688 1.1642A 374 8 3 W 66 67.5 353.16 353.18 Marl 0.388 1.58 2.16 0.763 8.98 0.0788 1.1842A 374 9 4 W 47 49 363.97 363.99 Marl 0.317 1.56 0.903 0.378 16.2 0.0552 0.90942A 374 10 1 W 13 14.5 371.63 371.65 Marl 0.306 2.04 1.49 0.643 12.9 0.0486 1.3142A 374 16 1 W 120 122 407.70 407.72 Gypsum (dark color) 0.0286 0.0946 0.0321 0.0205 1.35 0.00119 0.042942A 374 20 1 W 143 145 421.43 421.45 Gypsum 0.0365 0.161 0.0232 0.0160 2.86 0.00192 0.055542A 374 22 2 W 113 115 437.63 437.65 Halite 35.9 0.520 0.0268 0.788 0.751 0.00113 0.032142A 374 22 3 W 12 15 438.12 438.15 Halite 38.9 0.430 0.0422 0.535 0.474 0.00158 0.0517
Florence Rise, Eastern Mediterranean42A 375 1 1 W 84 86 138.34 138.36 Gypsum 0.0498 0.0851 0.122 0.0358 2.00 0.00193 0.14242A 375 2 4 W 99 101 193.89 193.91 Gypsum (dark color) 0.0309 0.105 0.0453 0.0153 3.22 0.00284 0.097342A 375 4 4 W 48 50 249.48 249.50 Marl(stone) 0.565 2.47 3.95 1.09 8.03 0.0693 3.3842A 375 6 5 W 85 87 467.85 467.87 Marl(stone) 0.653 2.86 4.96 1.11 5.94 0.0604 4.0242A 375 9 3 W 87 89 654.37 654.39 Marl(stone) 0.289 1.33 2.48 0.758 15.3 0.272 2.27
Florence Rise, Eastern Mediterranean42A 376 5 5 W 140 142 41.90 41.92 Marl 0.760 1.13 2.744 0.744 12.7 0.154 2.39242A 376 6 4 W 30 32 50.30 50.32 Marl 0.431 1.24 1.176 0.420 12.5 0.112 1.12442A 376 6 4 W 30 32 50.30 50.32 Marl 0.498 1.48 1.267 0.493 13.4 0.128 1.12642A 376 19 1 W 140 143 170.40 170.43 Gypsum 0.0341 0.0793 0.0387 0.00944 2.20 0.00334 0.060642A 376 20 1 W 73 75 185.73 185.75 Gypsum 0.0430 0.158 0.0631 0.0166 3.09 0.00320 0.095242A 376 22 1 W 68 70 205.68 205.70 Halite 51.1 0.0104 0.0117 0.0111 0.341 n.d. 0.012142A 376 22 1 W 140 142 206.40 206.42 Halite 40.9 0.00715 0.00785 0.00731 0.0126 n.d. 0.00353
aAges of samples were calculated based on the Geologic Time Scale 2012 [Gradstein et al., 2012]. Note: n.d. not detected.
Paleoceanography 10.1002/2015PA002853
KURODA ET AL. OS ISOTOPE RECORD OF MEDITERRANEAN 155
aluminosilicate minerals (e.g., clay minerals) were partially dissolved by the inverse aqua regia digestionmethod. Concentration of Al in marl samples shows a negative correlation with Ca (Figure 3a). Al concentra-tion shows positive correlations with Fe (Figure 3b), as well as K, Sc, V, Cr, Zn, Rb, Pb, Th (Table S1 in thesupporting information), and rare earth elements (REEs), suggesting that these elements are contained withinthe aluminosilicate mineral phase. Marl samples contain 0.909–4.02wt % Fe, with the Fe/Al ratios of 0.51 to1.01 (Table 2). A compilation of the Fe/Al ratios of post-Archean average Australian shale (PAAS), NorthAmerican shale component (NASC), and upper continental crust (UCC) shows that average Fe/Al ratio ofthe terrigenous component should range between 0.44 and 0.49 [Gromet et al., 1984; Taylor andMcLennan, 1985, 1995], a range significantly lower than that displayed by the Fe/Al ratios of the marl samplesin this study (Figure 3b). This suggests that our marl samples may contain substantial amount of nonterrigen-ous Fe which is not associated with the detrital aluminosilicate component. The concentration of nonterri-genous Fe (Fenonterr) was calculated as follows:
Fenonterr ¼ Fetotal � Altotal � Fe=Alð ÞPAAS: (1)
(Fe/Al)PAAS is Fe/Al ratio of PAAS as a representative value for shale component. Even if Fe/Al ratios of NASC orUCC are used as the representative terrigenous composition, similar Fenonterr results are obtained. The non-terrigenous iron would exist as either oxide/hydroxide (magnetite, hematite, ferromanganese oxides, etc.) orsulfide (pyrite etc.). Because we did not use anoxic sediments (sapropel) for our analyses, and lithologicaldescriptions show that marl samples in this study were deposited under oxic condition, we interpret thatthe nonterrigenous Fe is mainly contained in oxide mineral phases.
Mediterranean marl samples show light REE-enriched patterns with slight negative Eu anomalies whennormalized by CI chondrite (Figure 4a). By contrast, they show relatively flat patterns with slight positiveanomalies of Eu when normalized to PAAS (Figure 4b). Although a negative Ce anomaly is visible in thedeepest marl sample from the pre-MSC interval at Site 372 in the Balearic Basin, no other samples show aclear Ce anomaly (Figures 4a and 4b). The relatively flat patterns of PAAS-normalized REE compositionssuggest that REEs in the marl samples, which were extracted by the inverse aqua regia digestion method,
Figure 3. Cross plots between (a) Al and Ca, (b) Al and Fe, (c) Al and Os, (d) total Fe and Os, and (e) nonterrigenous Fe(Fenonterr) and Os concentrations. Green: Site 372, Balearic Basin, blue: Site 374, Ionian Basin, orange: Site 375, FlorenceRise, and red: Site 376, Florence Rise. Square: marl, diamond: gypsum, and circle: halite. Element ratios of post-Archean averageAustralian shale (PAAS), North American shale component (NASC), and upper continental crust (UCC) are shown as solid lines.Major element data of PAAS, NASC, and UCC are after Taylor and McLennan [1985], Gromet et al. [1984], and Taylor andMcLennan [1995], respectively, and representative Os concentration of UCC is after Peucker-Ehrenbrink and Jahn [2001].
Paleoceanography 10.1002/2015PA002853
KURODA ET AL. OS ISOTOPE RECORD OF MEDITERRANEAN 156
are mainly from terrigenous sourceswith a minor contribution of REEs thatare slightly enriched in Sm, Eu, and Gd.
Gypsum leachate samples (Table 2) con-tain low abundances of major elementswhere the sum totals of Na, Mg, Al, K,Ca, and Fe are commonly less than4wt %. The low concentration of Casuggests that gypsum crystals were par-tially dissolved during digestion. Halitesamples from the Ionian Basin (Site374) commonly contain less Na thanthose from the Florence Rise (Site 376)(Table 2) but higher concentrations ofother elements.
4.2. Re and Os Contents and187Os/188Os Ratios
The Os concentrations of theMediterranean sediments differ betweenmarls and evaporites: the former rangefrom 23.4 to 155pgg�1 and the latterfrom 0.275 to 14.8 pgg�1 (Figure 2 andTable 1). A gypsiferous mudstone sam-ple in the Tyrrhenian Sea (Site 654A)has an exceptionally high Os concentra-tion of ~185pgg�1. Os concentrations
in the halite samples is particularly low (0.275 to 2.16pgg�1), indicating that halite is not an important hostmineral for Os. This results in larger analytical uncertainties in the Os isotopic values for halites has as a resultof error propagation from procedural blank. The coarse fraction extracted from calcareous mud samples fromthe Gulf of Cadiz (Site U1387) show significantly higher concentrations of Os (139 to 1334pgg�1) than samplesfrom the Mediterranean Sea (Figure 2 and Table 1).
Osmium concentrations show a positive correlation with Al (R2 ~ 0.75) and Fe (R2 ~ 0.85) but have significantlyhigher Os/Al and Os/Fe ratios than those of the upper continental crust (Al and Fe are after Taylor andMcLennan [1995], and Os is after Peucker-Ehrenbrink and Jahn [2001]) (Figures 3c and 3d). This indicatesthat the terrigenous fraction (mainly clay minerals) is not the major Os host and that other mineral phasesplay a substantial role in hosting Os. Os concentrations show a strong positive correlation with nonterrigen-ous Fe (Fenonterr and R2 ~ 0.75) (Figure 3e) and based on this observation we interpret that the Os extracted ismainly associated with nonterrigenous iron, e.g., ferromanganese oxides. Precipitation of ferromanganeseoxides is one of the most important mechanisms for removing Os dissolved in seawater and preservingthe seawater Os isotopic record [e.g., Peucker-Ehrenbrink and Ravizza, 2000; Klemm et al., 2005, 2008]. Thepositive correlation between Os and Fenonterr suggests that the Os extracted is largely hydrogenous in originand that the 187Os/188Os record of sediments therefore reflects that of the seawater from which the ironoxides precipitated.
Marl samples from the Mediterranean Sea have Re concentrations ranging from 34.3 to 9508 pg g�1 (Figure 2and Table 1). Pre-Messinianmarine sediments (Sites 372, 654A, and 375) show particularly high Re concentra-tions compared with post-Messinian samples. Re concentrations in Messinian gypsum and gypsiferousmudstone samples also vary widely from ~8.31 to ~15092 pg g�1. Gypsum samples from the westernMediterranean Sea (Site 372) contain less Re (8.99 to 14.3 pg g�1) compared with those from the othereastern basins. Halite samples are commonly depleted in Re (8.21 to 42.7 pg g�1) relative to the othersediments. The coarse fraction of calcareous mud samples from the Gulf of Cadiz (Site U1387) containshigh concentrations of Re (~1696 to 12095 pg g�1). No clear correlation is seen between Os and Re
Figure 4. Rare earth element patterns of marl samples. REE patternsnormalized by (a) CI chondrite and (b) PAAS.
Paleoceanography 10.1002/2015PA002853
KURODA ET AL. OS ISOTOPE RECORD OF MEDITERRANEAN 157
concentrations (Figure 5a). In contrastto Os, Re concentrations or Re/Al ratiosshow no clear positive correlation withthose of major elements (e.g., Mg, K,Ca, and Fe). This implies either that themajor host of Re may be organic matter,or Re is contained in multiple hosts.
The marl samples show high measured187Os/188Os values (>0.7) relative toevaporite samples, except for threesamples from the pre-MSC intervals atSite 375, which have much less radio-genic values (<0.6) (Figures 2 and 5b).Evaporite samples (gypsum and halite)are generally characterized by low Osconcentrations and low measured187Os/188Os values (Figure 5b).Although the Os concentration in thegypsum samples is low, the relativecontribution of the Os blank was gener-ally only 2–6% of the Os extracted,suggesting that the quantity of Osextracted is large enough to promptdiscussion about paleoenvironmentalchanges. This is not true of the verylow Os concentrations and unradio-genic 187Os/188Os values in the halitesamples (Figure 5b and Table 1), wherethe contribution of procedural blankOs is not negligible (>10%). In thispaper, therefore, we discuss thetemporal variations of 187Os/188Os ofthe Mediterranean seawater as deducedfrom Os isotopic analysis of marl andgypsum samples, while Os isotopiccomposition of halite is used ascomplementary information.
4.3. Hydrological Change of the Mediterranean Sea
The ocean water 187Os/188Os values which have been reconstructed based on Os isotopic compositions ofsediments from the Atlantic [Reusch et al., 1998; Klemm et al., 2008] and Pacific [Pegram et al., 1992;Ravizza, 1993; Peucker-Ehrenbrink et al., 1995; Pegram and Turekian, 1999; Reusch et al., 1998] oceans experi-enced a gradual increase from ~0.8 in the middle Miocene to ~1.06 in Holocene (Figure 6a). Here we discussthe history of exchange between the Mediterranean Sea and the North Atlantic by synthesizing the187Os/188Osi data we have from all the different Mediterranean sites and comparing them with the oceanwater record for the Miocene-Pleistocene (Figure 6).4.3.1. Miocene Before the MSC (>5.97Ma)Marl samples from the pre-MSC interval (Burdigalian to Serravallian) at Site 372, Balearic Basin, are character-ized by radiogenic 187Os/188Osi values (Figure 6b), which are close to the 187Os/188Os values of the coevalglobal ocean water. This suggests that seawater in the western Mediterranean was well mixed with theNorth Atlantic during early to middle Miocene. Two of the three pre-MSC samples from the Tyrrhenian Seaalso plot near the range of the coeval ocean values (Figure 6b) suggesting that the Tyrrhenian Sea was alsoconnected with the western Mediterranean prior to the MSC. As an exception, the deepest sample at Site 654(~8Ma) has a particularly low 187Os/188Osi value of ~0.65 (Figure 6b). A similar trend has been reported for
Figure 5. Cross plot between (a) Os and Re concentrations and (b) Osconcentration and measured 187Os/188Os values. Green: Site 372, BalearicBasin, light blue: Site 654, Tyrrhenian Sea, blue: Site 374, Ionian Basin,orange: Site 375, Florence Rise, and red: Site 376, Florence Rise. Square:marl,diamond: gypsum, and circle: halite. Open circle is calcareousmud from SiteU1387, Gulf of Cadiz, North Atlantic.
Paleoceanography 10.1002/2015PA002853
KURODA ET AL. OS ISOTOPE RECORD OF MEDITERRANEAN 158
87Sr/86Sr records of foraminifers from Sites 652 to 654, with slightly unradiogenic values relative to coevalocean water during the pre-MSC interval (7 to 6Ma) [Müller et al., 1990]. This has been attributed to theincreased supply of unradiogenic Sr due to alteration of volcanic materials [Müller et al., 1990] or hydrother-mal activity [Flecker et al., 2002] associated with the Tyrrhenian seafloor rifting started around 10–9Ma [e.g.,Kastens and Mascle, 1990]. The 187Os/188Os value of Tyrrhenian seawater at ~8Ma may therefore have beenlowered by the same process as Sr.
In contrast to the western Mediterranean basins, marlstone samples in the pre-MSC interval (Serravallianthrough latest Tortonian) at Site 375, Florence Rise in the eastern Mediterranean, are characterized by muchless radiogenic isotopic compositions showing a gradual decrease in 187Os/188Osi values from 0.55 to 0.36(Figure 6b). These can be explained by two possible mechanisms: (1) progressive decrease in connectionbetween the eastern and western Mediterranean and (2) a local effect due to an increasing supply ofunradiogenic Os in combination with a quick removal of Os from the seawater.
In either scenario, we attribute the lower 187Os/188Osi values of the Florence Rise sediment (Figures 6b) to thesupply of unradiogenic Os from the ultramafic rocks including ophiolite complexes. The distribution ofophiolite bodies is particularly abundant around the eastern Mediterranean (Figure 1), e.g., Troodosophiolites in Cyprus, Western Hellenic ophiolites in Greece, and Mirdita ophiolites in Albania [Robertson,1977, 2002, 2004; Koepke et al., 2002; Dilek and Furnes, 2009]. This contrasts the western Mediterranean wherea smaller area of ultramafic rocks including the Ligurian ophiolite in the northwestern Italy and Corsica
Figure 6. Secular variation of (a) 187Os/188Os values of the global oceans and (b) the Mediterranean Sea and the Gulf ofCadiz during the last 20Ma. Gray band indicates global seawater 187Os/188Os values from literatures [Pegram et al.,1992; Ravizza, 1993; Peucker-Ehrenbrink et al., 1995; Reusch et al., 1998; Pegram and Turekian, 1999; Klemm et al., 2008]. Thegray band does not include the analytical uncertainties. The yellow vertical band shows the duration of Messinian SalinityCrisis (MSC). Plots are same with Figure 5.
Paleoceanography 10.1002/2015PA002853
KURODA ET AL. OS ISOTOPE RECORD OF MEDITERRANEAN 159
[Rampone et al., 2005], and peridotite bodies along the Betic-Rif orogenic belt around the Alboran Sea[Coward and Dietrich, 1989; Zeck, 1997] are exposed (Figure 1).
The first scenario regards the unradiogenic Os isotopic compositions of the pre-MSC sediment in Site 375 asa phenomenon affecting the entire eastern Mediterranean and attributes it to reduced exchange betweenthe eastern and western basins. If this were the case, it means that the connection between the eastern andwestern Mediterranean was restricted more than 8 million years before the MSC. Based on Sr isotopicrecord, Flecker and Ellam [1999] suggested restriction in the southern Turkey and the Adriatic Sea at around8Ma, 2Ma before the MSC (Figure 7a). On the other hand, the Sr isotopic records from the centralMediterranean basins such as Pissouri section in southwestern Cyprus (7.5–6.0Ma), Gavdos section inCrete (9.8–6.0Ma), and Gibliscemi section in Sicily (9.8–7.4Ma) show 87Sr/86Sr values almost identical to thatof the coeval ocean water (Figure 7a) [Flecker and Ellam, 2006]. This observation suggests that the centralpart of the eastern Mediterranean was sufficiently connected with the western Mediterranean and theAtlantic and might be apparently inconsistent with our first scenario. However, Os isotopic compositioncould be more sensitive to change in the hydrologic budget than Sr, as the Os concentration ratio betweenglobal mean river water and global ocean water (1:1) [Peucker-Ehrenbrink and Ravizza, 2000] is much higherthan that of Sr (1:20) [e.g., Palmer and Edmond, 1992], which causes a large difference in mean residencetime of Sr (106 years) and Os (104 years) in seawater. We speculate that this ratio can be also applied tothe Mediterranean Sea and therefore that when the gateway between the western and easternMediterranean is restricted, Os isotopic composition in the eastern Mediterranean could change more easilythan that of Sr.
Figure 7. Secular variations of (a) Sr isotopic ratios of carbonate and (b) Os isotopic ratios of sediments from theMediterranean Sea between 10 and 4.5 Ma. Sr isotope records from pre-evaporite marginal subbasins, Messinian SalinityCrisis evaporites and brackish-water deposits, and Pliocene foraminifera [McKenzie et al., 1988; McCulloch and De Deckker,1989; Müller et al., 1990; Müller and Mueller, 1991; Müller, 1993; Fortuin et al., 1995; Montanari et al., 1997; Flecker and Ellam,1999, 2006; Flecker et al., 2002; Sprovieri et al., 2003; Roveri et al., 2014a] are plotted against the ocean Sr isotope record[McArthur et al., 2001]. Error bar of oceanic Sr isotopic values (Figure 7a) are shown as a gray interval. The yellow verticalband shows the duration of MSC.
Paleoceanography 10.1002/2015PA002853
KURODA ET AL. OS ISOTOPE RECORD OF MEDITERRANEAN 160
The second scenario attributes the low 187Os/188Osi values of the pre-MSC sediments at Site 375 to localeffects, i.e., the incorporation of unradiogenic Os supplied from the adjacent Troodos ophiolite in Cyprus[Robertson, 1977, 2004], despite a strong connection between the eastern and western basins. This scenariorequires reducing conditions in the water column so that Os reduces to insoluble tetravalent state and can bequickly removed. This could result in heterogeneous 187Os/188Os values of seawater from place to place.Consequently, anoxic or dysoxic conditions are necessary to explain the second scenario. Regular, episodicdeposition of organic-rich sediments (sapropels) have been reported throughout Mediterranean over the last10Ma [e.g., Hsü et al., 1978; Rossignol-Strick et al., 1982; Hilgen, 1991; Cramp and O’Sullivan, 1999; De Langeet al., 2008; Rohling et al., 2015] supporting the idea that the dysoxic or anoxic conditions may have resultedin periodic rapid and enhanced removal of Os from water column.
Since the Florence Rise is located near Cyprus (Figure 1), the second scenario is explained by supply ofunradiogenic Os from the adjacent Troodos ophiolite complex [Robertson, 1977; Robertson et al., 1995;Kinnaird et al., 2011]. Although it is not evident whether the Troodos ophiolite complex has been alreadyexposed on land in the late Miocene, the presence of (1) ophiolite-derived clastic materials associated withthe evaporites at Site 375 and (2) ophiolite pebbles in the latest Messinian conglomerates in the PissouriBasin [Rouchy et al., 2001] may support our speculation. Furthermore, the trend of progressive decrease inOs isotopic ratios since the Serravallian onward might be coherent with the progressive shallowing of thecircum-Troodos area during Tortonian [Manzi et al., 2015]. The second scenario may be preferable giventhe oceanographic background of the eastern Mediterranean in which the bottom water frequently becameanoxic to precondition to create heterogeneous Os isotopic values of sediments place by place, byshortening residence time of reduced Os. However, our current data are insufficient to conclude which ofthe two contrasting hypotheses is a more probable explanation for the unradiogenic Os isotopic values inthe pre-MSC interval of the Florence Rise sediment. Pre-MSC Os isotopic records at subprecessionaltimescales and from other sites in the eastern Mediterranean would help to answer this.4.3.2. Messinian Salinity Crisis (5.97–5.33Ma)Messinian gypsum and halite samples have significantly lower 187Os/188Osi values than the range for coevalocean water (~0.80 to 0.96) (Figures 6b and 7b). In contrast, the coarse fractions of two Messinian samples(mainly foraminifers) from the Gulf of Cadiz (Site U1387) have 187Os/188Osi values within error of coeval oceanwater. The significant difference in 187Os/188Osi values between the Gulf of Cadiz and the Mediterraneanbasins suggests reduced connectivity between the North Atlantic and the Mediterranean Sea during theMSC. Gypsum samples from the western Mediterranean (Site 372) and Tyrrhenian Sea (Site 654A) have higherratios (0.75 to 0.80) relative to those from Site 374 in the Ionian Basin (0.62 and 0.67) and Sites 375 and 376 onthe Florence Rise (0.41 to 0.65) (Table 1 and Figures 6b and 7b). This west-east gradient in Os isotope ratiosmay be attributable to restricted exchange between the east and west Mediterranean basins. As discussedabove, we infer that the influx of unradiogenic Os was much more significant in the eastern Mediterraneanbecause of the distribution of ultramafic rocks in the drainage basins (Figure 1). The lower 187Os/188Osi valuesof evaporite samples from the eastern Mediterranean including the Ionian Sea (Site 374) and the FlorenceRise (Sites 375 and 376) could therefore be explained by the combination of the peculiarity of the bedrockaround the eastern Mediterranean Sea and the restricted exchange between the eastern and westernMediterranean basins.
A further possibility could be related to the mechanism of formation of the evaporite. The precipitation ofevaporites in the deep basins might have occurred from saturated brines formed in shallower areas andsubsequently sunk to the basin bottom through density currents [Roveri et al., 2014c]. In the LevantineBasin, such brines formed close to a large ophiolite complex of the Troodos massif [Manzi et al., 2015] couldhave supplied less radiogenic Os values like those found in Sites 375 and 376. Site 374 is located from theopposite side of the eastern Mediterranean in the Messina abyssal basin plain, the deepest portion ofthe Mediterranean in modern and probably also during the Messinian. Ophiolite-derived detritus have notbeen found around Site 374, but unradiogenic Os could have been provided to this site by density currentsformed elsewhere where ophiolitic rocks expose (such as in the northern Calabrian Arc and in the Hellenids)(Figure 1).
The temporal evolution of Mediterranean 187Os/188Osi data across the MSC (Figure 7b) resembles of theequivalent 87Sr/86Sr records [McKenzie et al., 1988; McCulloch and De Deckker, 1989; Müller et al., 1990;
Paleoceanography 10.1002/2015PA002853
KURODA ET AL. OS ISOTOPE RECORD OF MEDITERRANEAN 161
Müller and Mueller, 1991;Müller, 1993; Fortuin et al., 1995;Montanari et al., 1997; Flecker and Ellam, 1999, 2006;Flecker et al., 2002; Sprovieri et al., 2003; Roveri et al., 2014a], which also shows divergence away from coevalocean water toward lower, more unradiogenic values during the MSC (Figure 7a). The unradiogenic source ofSr has been interpreted as Mesozoic limestone bodies that prevail in the circum-Mediterranean drainageareas plus the very low Sr isotope ratios supplied by the Nile from its basaltic catchment in the Ethiopia[Brass, 1976; Albarede and Michard, 1987]. Unlike the Os isotopic data, no clear discrepancy can be seen forthe Sr isotopic records between the eastern and the western Mediterranean basins during the MSC(Figure 7a). The difference between Os and Sr isotopic records may be attributed to the long residence timeof Sr (~106 years) in seawater compared to Os (104 years) due to the difference in river water/seawater ratiosbetween Sr and Os. The clear discrepancy of our 187Os/188Os data between eastern and western basins(Figure 7b) provides new evidence to suggest limited water exchange between the eastern and westernMediterranean basins during the MSC.
The 87Sr/86Sr data of the MSC evaporites increasingly diverge from coeval seawater from Lower Gypsum toUpper Gypsum (Figure 7a) [Topper et al., 2011; Roveri et al., 2014a]. This cannot be seen in the 187Os/188Osidata because our data set is missing samples from the first phase of MSC (Figure 7b). As a result of the muchshorter residence time of Os in the water column, we anticipate that the Lower Gypsum would show187Os/188Os that evolve away from coeval seawater values much earlier than the Sr data during the first phaseof MSC. The only place where this interval was recovered by the previous drilling projects is Site 378 [Roveriet al., 2014b; Lugli et al., 2015]. Here unfortunately no pre-MSC deposits have been recovered; thus, this sitehas not been considered in this study aimed to reconstruct the long-term hydrological evolution of theMediterranean and not specifically focusing on the MSC for which topic a further study including a higheramount of samples also coming from the onshore successions is needed.4.3.3. Pliocene-Pleistocene (<5.33Ma)The 187Os/188Osi values of marl samples mainly plot within the range of the isotopic values of coeval oceanwater (Figure 6b). This suggests that the Mediterranean seawater was well mixed with North AtlanticOcean water during the Pliocene-Pleistocene. Exceptionally, the earliest Pliocene marl sample from theFlorence Rise (Site 376) shows an unradiogenic 187Os/188Osi value (~0.72) (Figure 6b). This might be alsorelated to supply of unradiogenic Os from ultramafic rocks associated with weathering of ophiolite complex,
Figure 8. (a) Osmium box model through the Mediterranean Sea and (b) a diagram of contour of Os isotopic ratio ofMediterranean Sea (RMed) as a function of Os isotopic ratio of river (Rriv) and a ratio of influx from Atlantic Ocean andriver (f = FA-M/Friv). Rriv value is explained by mixing between radiogenic Os from upper continental crust (187Os/188Os~1.4) and unradiogenic Os from ultramafic rock (ophiolite, ~0.13). The fraction of unradiogenic Os is shown as percentagein the top horizontal axis of Figure 8b. Two examples with Rriv values of 0.3 (white star) and 0.2 (gray star) are shown todecrease the RMed value from 0.85 to 0.4 by decreasing the f value. Detail parameters are given in Table 3.
Paleoceanography 10.1002/2015PA002853
KURODA ET AL. OS ISOTOPE RECORD OF MEDITERRANEAN 162
as ophiolite pebbles have been found in sediments from Cyprus deposited at the latest stage of Messinian[Rouchy et al., 2001].
4.4. Estimation of Unradiogenic Os
Our data show a clear decrease in 187Os/188Osi values in Mediterranean sediments during MSC. Based on abox model (Figure 8a) we estimate roughly the amount of unradiogenic Os to account for this decrease(Table 3). We assumed a similar parameters used by Flecker et al. [2002] and Topper et al. [2011] for thiscalculation. Although our 187Os/188Os data show different values between the eastern and westernMediterranean during MSC (Figures 6b and 7b), we regard the entire Mediterranean Sea as one box becauseour main purpose here is a first order estimation of unradiogenic Os required to decrease the Mediterraneanseawater 187Os/188Os (RMed) to the values measured rather than reproducing the isotopic discrepancybetween the subbasins. The RMed value varies as a function of a mean 187Os/188Os value of river inflow toMediterranean (Rriv), and an f value, which is a ratio of Os influxes from Atlantic and river (f= FA-M/Friv)(Figure 8b). The Rriv value depends on the mixing rate between radiogenic Os from continental crust (~1.4)and unradiogenic Os from ultramafic rocks (ophiolite, ~0.13) [Peucker-Ehrenbrink and Ravizza, 2000]. Forexample, if we assume 0.3 as the average Rriv value, which means ~87% of total riverine input comes fromultramafic source, a decrease in RMed value from 0.85 to 0.40 can be explained by a drop of f value from 18to 0.21 (white stars in Figure 8b). If the Rriv value was 0.2 (~94% of riverine Os is from ultramafic source), thisisotopic change is explained by a drop of f value from 21 to 0.42 (gray stars in Figure 8b). In either case, the fvalue should drop two orders of magnitude. If we assume that riverine Os flux (Friv) did not changesignificantly, Atlantic input (FA-M) would have to reduce during MSC down to 1–2% of the pre-MSC conditionin order to decrease the RMed value.
Present FA-M value (232 kgOs y�1) is estimated from water influx (2.268 × 1013m3 y�1) [Bryden et al., 1994;Flecker et al., 2002] and Os concentration of ocean water (10 pg kg�1) [Peucker-Ehrenbrink and Ravizza,2000]. If the FA-M value was same for the pre-MSC, the f values of 18 (Rriv of 0.3) and 21 (Rriv of 0.2) correspondto Friv values of 13 and 11 kgOs y�1, respectively. These values correspond to a supply rate of ~10 kg y�1
of unradiogenic Os from ultramafic rock source (Table 3). If the ultramafic bedrock has an average density
Table 3. Parameters Used in Box Model Calculations
Code Detail Value Reference and Note
Water fuxes (m3 y�1)Water inflow from Atlantic to Mediterranean at present 2.268 × 1013 Bryden et al. [1994] and Flecker et al. [2002]
Os concentration (pg kg�1)
[Os]ocean Ocean water at preset 10 Peucker-Ehrenbrink and Ravizza [2000]
Os fluxes (kg y�1)
FA-M Os inflow from Atlantic before and after MSC 232 Assumption: water inflow rate and [Os]oceanat present
FA-M Os inflow from Atlantic during multichannel seismic 2.64–4.46 Calculated to decrease RMed to the observed value(0.40), with an assumption that Friv value did
not change through time.
Friv Average Os influx from river 10.7–12.7 Prescribed to keep a steady state with Rrivvalues of 0.2–0.3
Volume and amountVolume of seawater in Mediterranean at present (km3) 3.75 × 106
MMed Initial amount of Os in Mediterranean (kg) 38400 Assumption: water volume and [Os]oceanat present
187Os/188Os valuesRocean Late Miocene ocean water 0.88 Peucker-Ehrenbrink and Ravizza [2000]
Rriv Average river water to Mediterranean Sea 0.2–0.3 Mixing between upper continental crust (1.4)and ultramafic rocks (0.13). See Figure 9b
(marked as stars).
RMed Mediterranean water before and after MSC 0.85 Prescribed to keep a steady state
RMed Mediterranean water during MSC 0.40 From our data (lowest value amonggypsum samples)
Paleoceanography 10.1002/2015PA002853
KURODA ET AL. OS ISOTOPE RECORD OF MEDITERRANEAN 163
of 3.2 g cm�1 and average Os concentration of 10 ngg�1, this flux corresponds to an erosion of~3.2 × 105m3 y�1 of ultramafic rock body. Given the general erosion rate of 10–300mmky�1 under thepresent Mediterranean climate [Einsele, 1992], an area of ~1000 to 32,000 km2 of ultramafic rock exposureis required to account for the erosion rate of 3.2 × 105m3 y�1. We think this range of exposure area is areasonable number. In total, about 200 km3 of ultramafic rock body would be eroded during 600 kyr of MSC.
5. Conclusions
We investigate the hydrological history of the Mediterranean from the middle Miocene to Pliocene and itsconnection with the Atlantic using Os isotopic compositions of marine sediments recovered from five sitesin the Mediterranean Sea (Site 372 in Balearic Sea in the western Mediterranean, Site 654 in the TyrrhenianSea, Site 374 in the Ionian Sea, and Sites 375 and 376 in Florence Rise in the eastern Mediterranean) and asite in the Gulf of Cadiz (Site U1387). Significant offsets in 187Os/188Osi from the coeval ocean water valuesare observed in the eastern Mediterranean before the MSC. These offsets are attributed either to limitedexchange between the eastern and western Mediterranean basins or to the local impact of the exhumationof ophiolite rocks like the Troodos massif in Cyprus. Slight offsets seen in the pre-MSC sediments from theTyrrhenian Sea may reflect its opening during the late Miocene. Our 187Os/188Osi data for the MSC evaporitesshow significant divergence from coeval ocean water values, suggesting significant restriction and/orisolation of the entire Mediterranean. Based on a simple box model we estimate that exchange rate of Osbetween Atlantic and Mediterranean decreased by about 2 orders of magnitude relative to the pre-MSCcondition, with a supply rate of unradiogenic Os of ~10 kg y�1. Post-MSC sediments (Pliocene-Pleistocene)have Os isotopic ratios close to coeval ocean water values, suggesting that Mediterranean-Atlantic exchangewas reestablished at or shortly after the Mio-Pliocene boundary.
ReferencesAbdul, A. H., A. Di Stefano, L. M. Foresi, F. J. Hilgen, S. M. Iaccarino, K. F. Kuiper, F. Lirer, G. Salvatorini, and E. Turco (2008), Integrated
stratigraphy and40Ar/
39Ar chronology of early Middle Miocene sediments from DSDP Leg 42A, Site 372 (Western Mediterranean),
Palaeogeogr. Palaeoclimatol. Palaeoecol., 257, 123–138, doi:10.1016/j.palaeo.2007.09.013.Albarede, F., and A. Michard (1987), Evidence for slowly changing
87Sr/
86Sr in runoff from freshwater limestones of southern France, Chem.
Geol., 64, 55–65.Brass, G. W. (1976), The variation of the marine
87Sr/
86Sr ratio during Phanerozoic time: Interpretation using a flux model, Geochim.
Cosmochim. Acta, 40, 721–730.Bryden, H. L., J. Candela, and T. H. Kinder (1994), Exchange through the Strait of Gibraltar, Prog. Oceanogr., 3, 201–248.Commission Internationale pour l’Exploration Scientifique de la Méditerranée (CIESM) (2008), The Messinian Salinity Crisis From Mega-
Deposits to Microbiology—A Consensus Report, CIESM Workshop Monogr., vol. 33, edited by F. Briand, 168 pp., CIESM, Monaco.Coward, M., and D. Dietrich (1989), Alpine tectonics—An overview, in Alpine Tectonics, edited by M. Coward, D. Dietrich, and R. G. Park, Geol.
Soc. Spec. Publ., 45, 1–29.Cramp, A., and G. O’Sullivan (1999), Neogene sapropels in the Mediterranean: A review, Mar. Geol., 153, 11–28.De Lange, G. J., J. Thomson, A. Reitz, C. P. Slomp, M. S. Principato, E. Erba, and C. Corselli (2008), Synchronous basin-wide formation and
redox-controlled preservation of a Mediterranean sapropel, Nat. Geosci., 1, 606–610, doi:10.1038/ngeo283.Dilek, Y., and H. Furnes (2009), Structure and geochemistry of Tethyan ophiolites and their petrogenesis in subduction rollback systems,
Lithos, 113, 1–20, doi:10.1016/j.lithos.2009.04.022.Du Vivier, A. D. C., D. Selby, B. B. Sageman, I. Jarvis, D. R. Gröcke, and S. Voigt (2014), Marine
187Os/
188Os isotope stratigraphy reveals the
interaction of volcanism and ocean circulation during Oceanic Anoxic Event 2, Earth Planet. Sci. Lett., 389, 23–33.Du Vivier, A. D. C., D. Selby, D. J. Condon, R. Takashima, and H. Nishi (2015), Pacific
187Os/
188Os isotope chemistry and U–Pb geochronology:
Synchroneity of global Os isotope change across OAE 2, Earth Planet. Sci. Lett., 428, 204–216.Einsele, G. (1992), Sedimentary Basins Evolution, Facies, and Sediment Budget, 629 pp., Springer, Berlin, Heidelberg.Expedition 339 Scientists (2013), Site U1387, in Proceedings of the Integrated Ocean Drilling Program, vol. 339, edited by D. A. V. Stow et al.,
Integr. Ocean Drilling Prog. Manage. Int., Inc, Tokyo, doi:10.2204/iodp.proc.339.105.2013.Flecker, R., and R. M. Ellam (1999), Distinguishing climatic and tectonic signals in the sedimentary successions of marginal
basins using Sr isotopes: An example from the Messinian salinity crisis, Eastern Mediterranean, J. Geol. Soc. London, 156,847–854.
Flecker, R., and R. M. Ellam (2006), Identifying Late Miocene episodes of connection and isolation in the Mediterranean–Paratethyan realmusing Sr isotopes, Sediment. Geol., 188–189, 189–203, doi:10.1016/j.sedgeo.2006.03.005.
Flecker, R., S. de Villiers, and R. M. Ellam (2002), Modelling the effect of evaporation on the salinity–87Sr/
86Sr relationship in modern and
ancient marginal-marine systems: The Mediterranean Messinian Salinity Crisis, Earth Planet. Sci. Lett., 203, 221–233.Fortuin, A. R., J. M. D. Kelling, and T. B. Roep (1995), The enigmatic Messinian–Pliocene section of Cuevas del Almanzora (Vera basin, SE Spain)
revisited—Erosional features and strontium isotope ages, Sediment. Geol., 97, 177–201.Gradstein, F., J. Ogg, M. Schmitz, and G. Ogg (2012), The Geologic Time Scale 2012, Elsevier, Amsterdam.Gramlich, J. W., T. J. Murphy, E. L. Garner, and W. R. Shields (1973), Absolute isotopic abundance ratio and atomic weight of a reference
sample of rhenium, J. Res. Natl. Bur. Stand., 77A, 691–698.Gromet, L. P., R. F. Dymek, L. A. Haskin, and R. L. Korotev (1984), The “North American shale composite”: Its compilation, major and trace
element characteristics, Geochim. Cosmochim. Acta, 48, 2469–2482.
Paleoceanography 10.1002/2015PA002853
KURODA ET AL. OS ISOTOPE RECORD OF MEDITERRANEAN 164
AcknowledgmentsSediment core samples at Site U1387Cwere taken during cruise of D/V JOIDESResolution during IODP Exp. 339(sample request number 1394). IODPprovided sediment samples takenby the previous DSDP and ODPexpeditions (sample request numbers22458 and 22458B). Supporting traceelement data are included in Table S1.Any data reported in this paper may beobtained by contacting J.K. (email:[email protected]). This study wasfinancially supported by Grants-in-Aidfor Scientific Research from the JapanSociety for the Promotion of Science(KAKENHI Grant-in-Aid for YoungScientists (B) 23740387 and KAKENHIScientific Research (C) 25400505). Wehave benefited from discussions withH. Kawahata, Y. Isaji, G. de Lange,and G. Aloisi onto the draft of thismanuscript. We thank K. Ishikawa,Y. Nakamura, and H. Yamamoto forsample preparation, and R. Senda andY. Ohtsuki for analytical assistance ofmeasurement with TIMS and ICP-MS.
Hassler, D. R., B. Peucker-Ehrenbrink, and G. Ravizza (2000), Determination of Os isotope composition by sparging OsO4 in a magnetic sectorICP-MS, Chem. Geol., 166, 1–14, doi:10.1016/S0009-2541(99)00180-1.
Hernández-Molina, F. J., et al. (2014), Onset of Mediterranean outflow into the North Atlantic, Science, 344, 1244–1250, doi:10.1126/science.1251306.
Hilgen, F. J. (1991), Astronomical calibration of Gauss to Matuyama sapropels in the Mediterranean and implication for the geomagneticPolarity Time Scale, Earth Planet. Sci. Lett., 104, 226–244.
Hsü, K., et al. (1978), Initial Reports of the Deep Sea Drilling Project, vol. 42, Part 1, U.S. Gov. Print. Off., Washington.Hsü, K. J., W. B. F. Ryan, and M. B. Cita (1973), Late Miocene desiccation of the Mediterranean, Nature, 242, 240–244.Kastens, K. A., and J. Mascle (1990), The geological evolution of the Tyrrhenian Sea: An introduction to the scientific results of ODP Leg 107,
in Proceedings of the Ocean Drilling Program, Sci. Results, vol. 107, pp. 3–26, Ocean Drilling Program, College Station, Tex.Kinnaird, T. C., A. H. F. Robertson, and A. Morris (2011), Timing of uplift of the Troodos Massif (Cyprus) constrained by sedimentary and
magnetic polarity evidence, J. Geol. Soc. London, 168, 457–470.Klemm, V., S. Levasseur, M. Frank, J. R. Hein, and A. N. Halliday (2005), Osmium isotope stratigraphy of a marine ferromanganese crust,
Earth Planet. Sci. Lett., 238(1–2), 42–48, doi:10.1016/j.epsl.2005.07.016.Klemm, V., M. Frank, S. Levasseur, A. N. Halliday, and J. R. Hein (2008), Seawater osmium isotope evidence for a middle Miocene flood basalt
event in ferromanganese crust records, Earth Planet. Sci. Lett., 273, 175–183, doi:10.1016/j.epsl.2008.06.028.Koepke, J., E. Seidel, and H. Kreuzer (2002), Ophiolites on the Southern Aegean islands Crete, Karpathos and Rhodes: Composition,
geochronology and position within the ophiolite belts of the Eastern Mediterranean, Lithos, 65, 183–203.Krijgsman, W., F. J. Hilgen, I. Raffi, F. J. Sierro, and D. S. Wilson (1999), Chronology, causes and progression of the Messinian salinity crisis,
Nature, 400, 652–655.Levasseur, S., J.-L. Birck, and C. J. Allègre (1999), The osmium riverine flux and the oceanic mass balance of osmium, Earth Planet. Sci. Lett., 174,
7–23.Lofi, J., F. Sage, J. Déverchère, L. Loncke, A. Maillard, V. Gaullier, I. Thinon, H. Gillet, P. Guennoc, and C. Gorini (2011), Refining our knowledge of
the Messinian salinity crisis records in the offshore domain through multi-site seismic analysis, Bull. Soc. Géol. France, 182, 163–180.Lugli, S., V. Manzi, M. Roveri, and B. C. Schreiber (2010), The Primary Lower Gypsum in the Mediterranean: A new facies interpretation for the
first stage of the Messinian salinity crisis, Palaeogeogr. Palaeoclimatol. Palaeoecol., 297, 83–99, doi:10.1016/j.palaeo.2010.07.017.Lugli, S., V. Manzi, M. Roveri, and B. C. Schreiber (2015), The deep record of the Messinian salinity crisis: Evidence of a non-desiccated
Mediterranean Sea, Palaeogeogr. Palaeoclimatol. Palaeoecol., 297, 83–99, doi:10.1016/j.palaeo.2015.05.017.Manzi, V., R. Gennari, S. Lugli, M. Roveri, N. Scafetta, and B. C. Schreiber (2012), High-frequency cyclicity in the Mediterranean Messinian
evaporites: Evidence for solar–lunar climate forcing, J. Sediment. Res., 82, 991–1005, doi:10.2110/jsr.2012.81.Manzi, V., R. Gennari, F. Hilgen, W. Krijgsman, S. Lugli, M. Roveri, and F. J. Sierro (2013), Age refinement of the Messinian salinity crisis onset in
the Mediterranean, Terra Nova, 25, 315–322, doi:10.1111/ter.12038.Manzi, V., S. Lugli, M. Roveri, F. Dela Pierre, R. Gennari, F. Lozar, M. Natalicchio, B. C. Schreiber, M. Taviani, and E. Turco (2015), The Messinian
salinity crisis in Cyprus: A further step towards a new stratigraphic framework for Eastern Mediterranean, Basin Res., doi:10.1111/bre.12107.
McArthur, J. M., R. J. Howarth, and T. R. Bailey (2001), Strontium isotope stratigraphy: LOWESS version 3: Best fit to the marine Sr-isotopecurve for 0–509 Ma and accompanying look-up table for deriving numerical age, J. Geol., 109, 155–170.
McCulloch, M. T., and P. De Deckker (1989), Sr isotopic constraints on the evolution of the Mediterranean Basin during the Messinian“salinity crisis”, Nature, 342, 62–65.
McKenzie, J. A., D. A. Hodell, P. A. Mueller, and D. W. Mueller (1988), Application of strontium isotopes to late Miocene-early Pliocenestratigraphy, Geology, 16, 1022–1025.
Meijer, P. T. (2006), A box model of the blocked-outflow scenario for the Messinian Salinity Crisis, Earth Planet. Sci. Lett., 248, 486–494,doi:10.1016/j.epsl.2006.06.013.
Montanari, A., et al. (1997), Integrated stratigraphy of the Middle to Upper Miocene pelagic sequence of the Conero Riviera (Marche Region,Italy), in Miocene Stratigraphy: An Integrated Approach, edited by A. Montanari, G. S. Odin, and R. Coccioni, pp. 409–450, Elsevier,Amsterdam.
Morgan, J. W., D. W. Golightly, and A. F. Dorrzapf Jr. (1991), Methods for the separation of rhenium, osmium and molybdenum applicable toisotope geochemistry, Talanta, 38, 259–265.
Müller, D. W. (1993), Pliocene transgression in the western Mediterranean Sea: Strontium isotopes from Cuevas del Almanzora (SE Spain),Paleoceanography, 8, 127–134.
Müller, D. W., and P. A. Mueller (1991), Origin and age of the Mediterranean Messinian evaporites: Implications form Sr isotopes,Earth Planet. Sci. Lett., 107, 1–12.
Müller, D. W., P. A. Mueller, and J. A. McKenzie (1990), Strontium isotopic ratios as fluid tracers in Messinian evaporites of the Tyrrhenian sea(western Mediterranean sea), in Proceedings of the Ocean Drilling Program, Sci. Results, vol. 107, pp. 603–614, Ocean Drilling Program,College Station, Tex.
Nozaki, T., K. Suzuki, G. Ravizza, J.-I. Kimura, and Q. Chang (2012), A method for rapid determination of Re and Os isotopecompositions using ID-MC-ICP-MS combined with the sparging method, Geostand. Geoanal. Res., 36(2), 131–148, doi:10.1111/j.1751-908X.2011.00125.x.
Oxburgh, R. (2001), Residence time of osmium in the oceans, Geochem. Geophys. Geosyst., 2(6), 1018, doi:10.1029/2000GC000104.Palmer, M. R., and J. M. Edmond (1992), Controls over the strontium isotope composition of river water, Geochim. Cosmochim. Acta, 56,
2099–2111.Paquay, F. S., and G. Ravizza (2012), Heterogeneous seawater
187Os/
188Os during the Late Pleistocene glaciations, Earth Planet. Sci. Lett.,
349–350, 126–138.Pegram, W. J., and K. K. Turekian (1999), The osmium isotopic composition change of Cenozoic sea water as inferred from a deep-sea core
corrected for meteoritic contributions, Geochim. Cosmochim. Acta, 63(23–24), 4053–4058.Pegram, W. J., S. Krishnaswami, G. E. Ravizza, and K. K. Turekian (1992), The record of seawater
187Os/
186Os variation through the Cenozoic,
Earth Planet. Sci. Lett., 113, 569–576.Peucker-Ehrenbrink, B., and B.-M. Jahn (2001), Rhenium-osmium isotope systematics and platinum group element concentrations: Loess and
the upper continental crust, Geochem. Geophys. Geosyst., 2(10), 1061, doi:10.1029/2001GC000172.Peucker-Ehrenbrink, B., and G. Ravizza (2000), The marine osmium isotope record, Terra Nova, 12, 205–219.Peucker-Ehrenbrink, B., G. Ravizza, and A. W. Hofmann (1995), The marine
187Os/
186Os record of the past 80 million years, Earth Planet. Sci.
Lett., 130, 155–167, doi:10.1016/0012-821X(95)00003-U.
Paleoceanography 10.1002/2015PA002853
KURODA ET AL. OS ISOTOPE RECORD OF MEDITERRANEAN 165
Rampone, E., A. Romairone, W. Abouchami, G. B. Piccardo, and A. W. Hofmann (2005), Chronology, petrology, and isotope geochemistry ofthe Erro-Tobbio peridotites (Ligurian Alps, Italy): Records of Late Palaeozoic lithospheric extension, J. Petrol., 46, 799–827, doi:10.1093/petrology/egi001.
Ravizza, G. (1993), Variations of the187
Os/188
Os ratio of seawater over the past 28 million years as inferred from metalliferous carbonates,Earth Planet. Sci. Lett., 118, 335–348.
Reusch, D. N., G. Ravizza, K. A. Maasch, and J. D. Wright (1998), Miocene seawater187
Os/188
Os ratios inferred from metalliferous carbonates,Earth Planet. Sci. Lett., 160, 163–178.
Robertson, A. (2004), Development of concepts concerning the genesis and emplacement of Tethyan ophiolites in the EasternMediterranean and Oman regions, Earth Sci. Rev., 66, 331–387.
Robertson, A. H. F. (1977), Tertiary uplift history of the Troodos massif, Cyprus, Geol. Soc. Am. Bull., 88, 1763–1772.Robertson, A. H. F. (2002), Overview of the genesis and emplacement of Mesozoic ophiolites in the Eastern Mediterranean Tethyan region,
Lithos, 65, 1–67.Robertson, A. H. F., S. Eaton, E. J. Follows, and A. S. Payne (1995), Depositional processes and basin analysis of Messinian evaporites in Cyprus,
Terra Nova, 7, 233–253.Rohling, E. J., G. Marinoa, and K. M. Grant (2015), Mediterranean climate and oceanography, and the periodic development of anoxic events
(sapropels), Earth Sci. Rev., 143, 62–97.Rossignol-Strick, M., V. Nesteroff, P. Olive, and C. Vergnaud-Grazzini (1982), After the deluge; Mediterranean stagnation and sapropel formation,
Nature, 295, 105–110.Rouchy, J. M., and A. Caruso (2006), The Messinian salinity crisis in the Mediterranean basin: A reassessment of the data and an integrated
scenario, Sediment. Geol., 188–189, 35–67, doi:10.1016/j.sedgeo.2006.02.005.Rouchy, J. M., F. Orszag-Sperber, M. M. Blanc-Valleron, C. Pierre, M. Rivière, N. Combourieu-Nebout, and I. Panayides (2001),
Paleoenvironmental changes at the Messinian-Pliocene boundary in the eastern Mediterranean: Southern Cyprus basins, Sediment. Geol.,145, 93–117.
Roveri, M., et al. (2014a), The Messinian Salinity Crisis: Past and future of a great challenge for marine sciences, Mar. Geol., 352, 25–58.Roveri, M., S. Lugli, V. Manzi, R. Gennari, and B. C. Schreiber (2014b), High-resolution strontium isotope stratigraphy of the Messinian deep
Mediterranean basins: Implications for marginal to central basins correlation, Mar. Geol., 349, 113–125.Roveri, M., V. Manzi, A. Bergamasco, F. Falcieri, R. Gennari, and S. Lugli (2014c), Dense shelf water cascading and Messinian canyons: A new
scenario for the Mediterranean salinity crisis, Am. J. Sci., 314, 751–784, doi:10.2475/05.2014.03.Ryan, W. B. F. (2009), Decoding the Mediterranean salinity crisis, Sedimentology, 56, 95–136, doi:10.1111/j.1365-3091.2008.01031.x.Schoenberg, R., T. F. Nagler, and J. D. Kramers (2000), Precise Os isotope ratio and Re–Os isotope dilution measurements down to the
picogram level using multicollector inductively coupled plasma mass spectrometry, Int. J. Mass Spectrom., 197, 85–95, doi:10.1016/S1387-3806(99)00215-8.
Shipboard Scientific Party (1987), Site 654, in Proceedings of the Ocean Drilling Program, Initial Rep., vol. 107, edited by K. A. Kastenset al., pp. 747–875, Ocean Drilling Program, College Station, Tex.
Shirey, S. B., and R. J. Walker (1995), Carius tube digestion for low-blank rhenium-osmium analysis, Anal. Chem., 67, 2136–2141.Smoliar, M. I., R. J. Walker, and J. W. Morgan (1996), Re–Os isotope constraints on the age of Group IIA, IIIA, IVA, and IVB iron meteorites,
Science, 271, 1099–1102.Sprovieri, M., M. Barbieri, A. Bellanca, and R. Neri (2003), Astronomical tuning of the Tortonian
87Sr/
86Sr curve in the Mediterranean basin,
Terra Nova, 15, 29–35.Taylor, S. R., and S. McLennan (1985), The Continental Crust: Its Composition and Evolution, 312 pp., Blackwell, Oxford.Taylor, S. R., and S. McLennan (1995), The geochemical evolution of the continental crust, Rev. Geophys., 33, 241–265.Tokumaru, A., T. Nozaki, K. Suzuki, K. T. Goto, Q. Chang, J.-I. Kimura, Y. Takaya, Y. Kato, A. Usui, and T. Urabe (2015), Re-Os isotope geochemistry
in the surface layers of ferromanganese crusts from the Takuyo Daigo Seamount, northwestern Pacific Ocean, Geochem. J., 49, 233–241,doi:10.2343/geochemj.2.0352.
Topper, R. P. M., R. Flecker, P. T. Meijer, and M. J. R. Wortel (2011), A box model of the Late Miocene Mediterranean Sea: Implications fromcombined
87Sr/
86Sr and salinity data, Paleoceanography, 26, PA3223, doi:10.1029/2010PA002063.
Walker, R. J., H. M. Prichard, A. Ishiwatari, and M. Pimentel (2002), The osmium isotopic composition of convecting upper mantle deducedfrom ophiolite chromites, Geochim. Cosmochim. Acta, 66, 329–345.
Wright, R. (1978), Neogene paleobathymetry of the Mediterranean based on benthic foraminifers from DSDP leg 42A, in Initial Reports of theDeep Sea Drilling Project, vol. 42, edited by K. J. Hsü et al., pp. 837–846, U.S. Gov. Print. Off., Washington.
Zeck, H. P. (1997), Mantle peridotites outlining the Gibraltar Arc—Centrifugal extensional allochthons derived from the earlier Alpine,westward subducted nappe pile, Tectonophysics, 281, 195–207.
Paleoceanography 10.1002/2015PA002853
KURODA ET AL. OS ISOTOPE RECORD OF MEDITERRANEAN 166