Late Maastrichtian paleoclimatic and paleoceanographicchanges inferred from Sr/Ca ratio and stable isotopes

D. Stu«ben a, U. Kramar a;�, Z.A. Berner a, M. Meudt a, G. Keller b,S. Abramovich b, T. Adatte c, U. Hambach d, W. Stinnesbeck e

a Institut fu«r Mineralogie und Geochemie, Universita«t Karlsruhe, Kaiserstr. 12, 76128 Karlsruhe, Germanyb Department of Geosciences, Princeton University, Princeton, NJ 08544, USAc Institute de Geologie, 11 Rue Emile Argand, 2007 Neucha“tel, Switzerlandd Geowissenschaften, Universita«t Bayreuth, D-95440 Bayreuth, Germany

e Geologisches Institut, Universita«t Karlsruhe, Postfach 6980, 76128 Karlsruhe, Germany

Received 9 April 2002; accepted 4 June 2003

Abstract

Milankovitch-scale cycles can be recognized in high-resolution N13C, N18O, Sr/Ca, mineralogical, and magnetic

susceptibility data in hemipelagic sediments that span the last 700 kyr of the Maastrichtian at Elles, Tunisia. Oxygenisotope data reveal three cool periods between 65.50 and 65.55 Ma (21.5^23.5 m), 65.26 and 65.33 Ma (8^11 m), and65.04 and 65.12 Ma (1.5^4 m), and three warm periods between 65.33 and 65.38 Ma (12^16 m), 65.12 and 65.26 Ma(4^8 m), and 65.00 and 65.04 Ma (0^1.5 m). The cool periods are characterized by small surface-to-deep temperaturegradients that reflect intensive mixing of the water column. The surface-to-deep Sr/Ca gradient generally correlateswith the oscillating vT trend (temperature difference between surface and bottom waters). The carbon isotopecomposition of planktonic foraminifera indicates a continuous decrease in surface bioproductivity during LateMaastrichtian. Decreasing v

13C values (difference between the N13C values of surface and bottom dwelling

foraminifera) and the carbon isotope ratios of the planktonic species at the onset of gradual warming at 65.50 Mareflect a reduction in surface productivity as a result of decreased upwelling that accompanied global warming andpossibly increased atmospheric pCO2 related to Deccan Trap volcanism. Time series analysis applied to magneticsusceptibility, N

18O, and Sr/Ca data identifies the 20 kyr precession, 40 kyr obliquity, and 100 kyr eccentricityMilankovich cycles.? 2003 Elsevier B.V. All rights reserved.

Keywords: Late Maastrichtian; stable isotopes; Sr/Ca; paleoclimate

1. Introduction

The Maastrichtian climate was characterized by

long-term cooling followed by short-term warm-ing (65.4^65.2 Ma) and rapid cooling near the endof the Maastrichtian (Shackleton et al., 1984; Za-chos et al., 1985, 1986; Barrera and Huber, 1990;Stott and Kennett, 1990; Spicer and Cor¢eld,1992; D’Hondt and Lindinger, 1994; Barrera,1994; Barrera et al., 1997; Li and Keller,

0031-0182 / 03 / $ ^ see front matter ? 2003 Elsevier B.V. All rights reserved.doi:10.1016/S0031-0182(03)00499-1

* Corresponding author.E-mail addresses: doris.stueben@img.uni-karlsruhe.de

(D. Stu«ben), utz.kramar@img.uni-karlsruhe.de (U. Kramar).

PALAEO 3174 9-9-03

Palaeogeography, Palaeoclimatology, Palaeoecology 199 (2003) 107^127

www.elsevier.com/locate/palaeo

1998a,b, 1999). Salinity £uctuations indicate thatduring the short-term global warming, high-lati-tude deep-water production was signi¢cantly re-duced and warm saline deep waters, probablyoriginating in the shallow middle- and low-lati-tude regions of the Tethys, £ooded the ocean ba-sins (Li and Keller, 1998b). The rapid warmingmay have been caused by increased CO2 due tomajor Deccan Trap volcanism (Courtillot et al.,1996; Ho¡mann et al., 2000) and possibly an im-pact event, as suggested by the recent discovery ofglass spherule deposits in upper Maastrichtiansediments dated between 65.4 and 65.2 Ma (Stin-nesbeck et al., 2001; Keller et al., 2002). Majorfaunal changes across latitudes associated withthis greenhouse warming have been increasinglyrecognized by planktonic foraminiferal workers(Abramovich et al., 1998; Kucera and Malmgren,1998; Li and Keller, 1998a,b; Olsson et al., 2001;McLeod et al., 2001).Although these studies provide a tantalizing

glimpse of the complexity of the climatic and en-vironmental changes during the last 500 kyr of theMaastrichtian, their detailed nature is still un-known. This is largely because time control af-forded by biostratigraphic and chronostrati-graphic studies are typically on the order ofhundreds of thousands of years, whereas oceanicprocesses operate on much shorter time scales ofthousands of years or less. High-resolution sam-pling at the 5^10 cm level can improve the tem-poral resolution to several thousand years, asshown by Li and Keller (1998a) at the mid-lati-tude South Atlantic deep-sea DSDP Site 525Aand by Abramovich and Keller (2002) for theElles section in Tunisia. However, these paleon-tological studies also revealed that the rate ofsedimentation was seven times higher in thelow latitude Tethyan continental shelf at Elles,than in the open ocean Site 525A. High-resolu-tion sample analysis of this expanded Tethyssection thus provides a unique opportunity torecover a very detailed climatic and environ-mental record that is not available in open oceansediments. This is the primary goal of thisstudy.In this study we evaluate the climatic and pale-

oceanographic history of the low-latitude Tethys

continental shelf of Tunisia for the last 700 kyr ofthe Maastrichtian based on stable isotopes (N13Cand N

18O), Sr/Ca ratios, magnetic susceptibility,and mineralogical studies. These extensive datasets provide information on paleoproductivity(N13C), sea surface and bottom temperatures(N18O and Sr/Ca ratios), and sea level £uctuations(mineralogy and Sr/Ca ratios). Correlationsamong these various data sets and the Milanko-vitch cycles provide new insights into the Tethyanpaleoenvironmental conditions during the termi-nal Maastrichtian.

2. Materials and methods

The Elles section is located 75 km southeast o¡the city of El Kef, Tunisia, and has been describedby Li et al. (2000) and Abramovich and Keller(2002) (Fig. 1). Elles outcrops span a continuoussuccession from the Campanian to the Eocene.The uppermost Maastrichtian portion examinedfor this study consists of 30 m, spanning fromchron 30N through 29R to the K/T boundary.Sediments comprise dark gray marls with somesandy to silty interlayers. Biostratigraphy is basedon the planktonic foraminiferal zonation of Abra-movich and Keller (2002) with zones CF1 andCF2 marking the last 300, respectively, 150 kyrof the Maastrichtian.Two hundred and forty samples were analyzed

at 20 cm intervals for this study. Sediments weredisaggregated in water and washed through a 63Wm sieve until clean sample residues were ob-tained. From each sample about 15 specimens ofthe benthic foraminifer Cibicidoides pseudoacutaand 30 specimens of the planktonic foraminiferRugoglobigerina rugosa (Fig. 2) were picked underthe microscope for stable isotope analyses. Thesame species were also picked for Sr/Ca ratioanalyses. Planktonic foraminiferal tests are wellpreserved, but show some recrystallization of theoriginal test calcite. Stable carbon and oxygenisotope data were obtained using a fully auto-mated preparation system (MultiCarb) connectedon-line to an isotope ratio mass spectrometer (Op-tima, Micromass Limited UK). All carbon andoxygen isotope values are reported relative to

PALAEO 3174 9-9-03

D. Stu«ben et al. / Palaeogeography, Palaeoclimatology, Palaeoecology 199 (2003) 107^127108

the V-PDB standard. External reproducibility wasbetter than 0.1x.Strontium and Ca contents in foraminiferal

tests were determined using total re£ection X-ray £uorescence (Atomika EXTRA II) on oneto six specimens (about 20 Wg) with no visibleFe-oxide coatings or sediment in¢llings of cham-bers. Foraminifera were cleaned with bi-distilledwater and transferred into PP-microcentrifugetubes. One hundred Wl of subboiled HNO3, spikedwith 0.2 ng Ga/Wl as internal standard was used todigest the foraminiferal tests. The tubes were cen-trifuged at approximately 10 000 RCF to concen-trate the sample and nitric acid in the cone-shapedbottom. Digestion was continued for 1 h in anultrasonic bath, followed by a ¢nal centrifugation.Fifty Wl of solution were transferred to a quartzsample carrier. After drying the samples were an-alyzed for 2000 s using a Mo tube (50 kV/38 mA)and ¢lter, and a W tube, respectively. Strontiumwas normalized to Ca, with the actual sample

weights calculated from the intensities of the Gaspikes.For magnetic susceptibility determinations non-

oriented samples with a spacing of 10 cm weretaken from the 30 m interval below the K/Tboundary. The rock was crushed, ¢lled in plasticboxes and ¢xed with cotton wool. The suscepti-bility was measured with a KLY-2 Kappabridge(AGICO, Brno, CR) and normalized to its mass,because of the unknown in situ density of thesediment (the resulting unit is 1036 SI/g). Assum-ing an average density of 2.4 g/cm3 for the marlsthe given mass susceptibility values have to bemultiplied by 24 in order to compare them toother data sets presented in original volume sus-ceptibility (Hambach et al., 2002).X-ray di¡raction analyses of the whole rock

were carried out for all the samples at the Geo-logical Institute of the University of Neucha“tel.Whole rock mineralogy and clay mineral contentswere measured with a SCINTAG XDS 2000 X-ray

Fig. 1. Location of the Elles section in northwestern Tunisia 75 km SE of the city of El Kef (modi¢ed after Burollet, 1967 andZaier et al., 1998), and schematic cross-section with paleogeographic positions of the Seldja, Elles, El Kef, and El Melah K^Tsections.

PALAEO 3174 9-9-03

D. Stu«ben et al. / Palaeogeography, Palaeoclimatology, Palaeoecology 199 (2003) 107^127 109

powder di¡ractometer following the proceduredescribed by Adatte et al. (1996).

3. Diagenesis

3.1. Diagenetic e¡ects on stable isotopes

Oxygen isotope records of well-preservedmonospeci¢c foraminifera are currently the mostaccurate and useful means of climate change re-construction. Although foraminifera are relativelywell preserved at Elles, diagenesis seems to havealtered the original signals, as indicated by scan-ning electron microscopy (SEM) images that re-veal minor recrystallization and test overgrowth(Fig. 2), and some tests show calcite cement in-¢llings. Similar diagenetic e¡ects have been dis-cussed by Mitchell et al. (1997), Pearson et al.(2001), Schrag et al. (1995), and Schrag (1999).

Diagenetic e¡ects at Elles are indicated by theoverall low N

18O values for planktonic andbenthic foraminifera. A shift in the same directioncan be due to burial diagenesis (Schrag et al.,1995), because recrystallization occurs at highertemperatures, and secondary carbonate with low-er N

18O values is formed in both planktonic andbenthic species. Besides a shift toward lower N18Ovalues due to the interaction with isotopicallylight meteoric waters, or equilibration at highertemperatures during burial, the e¡ect of early dia-genesis may also be a considerable factor (Schraget al., 1995). Due to a high-temperature gradientbetween surface and bottom waters in low lati-tudes, early diagenetic recrystallization in thecool water environment of the ocean £oor tendsto shift primary oxygen isotope values of (mainly)planktonic foraminifera towards higher valuesand carbon isotopes towards lower values, indi-cating cooler sea surface temperatures and lower

Fig. 2. SEM illustrations of foraminiferal species used for stable isotope analyses and Sr/Ca ratios. Scale bar represent 100 Wm.(1, 2) Rugoglobigerina rugosa ; (3, 4) Cibicidoides pseudoacuta.

PALAEO 3174 9-9-03

D. Stu«ben et al. / Palaeogeography, Palaeoclimatology, Palaeoecology 199 (2003) 107^127110

productivity compared to actual values (Schrag etal., 1992, 1995; Jenkyns et al., 1994; Mitchell etal., 1997; Sakai and Kano, 2001; Pearson et al.,2001). However, the shallow paleowater depth,and hence small temperature gradient betweensurface and bottom water, indicates that the e¡ectof early recrystallization is presumably low andtherefore has not to be taken into account atElles. Today Elles is located at 36‡N, but at theend of the Cretaceous the position of Elles was inlow latitudes (close to the tropic of Cancer, Sco-tese and Sager, 1988) on the African continentalshelf at an estimated paleowater depth of aboutV150 m.At Elles, the N

18O/N13C plot shows a signi¢cantpositive correlation at the 99.95% level (r=0.76;175 sample pairs; error probability P6 0.005) forthe planktonic foraminifer Rugoglobigerina rugo-sa, but no signi¢cant correlation (r=0.09; 156

sample pairs; P=0.259) for the benthic foramin-ifer Cibicidoides pseudoacuta (Fig. 3). This rela-tionship is of particular interest to interpret theisotope records and discuss the possible alterationof primary isotope signals. There are severalmechanisms which could lead to a correlation be-tween N

18O and N13C:

(a) Based on experimental results, Spero et al.(1997) concluded that the covariance betweenN13C and N

18O in foraminifera can be a conse-quence of changes in CO23

3 content (i.e. in alka-linity) of seawater. However, the slopes of theregression lines in their study are much lower(0.29^0.33) than that observed in our data set(about 3.5), which rules out such an interpreta-tion.(b) Changes in salinity may also lead to cova-

riance between the N18O and N13C (e.g. Khim and

Park, 2000). The oxygen isotope composition of

Fig. 3. Plot of N18O versus N13C of benthic and planktonic foraminifera from the Elles section. Note that the correlation betweenthese data is weak for benthic (r=0.09; P=0.259) compared with planktonic foraminifera (r=0.76; P6 0.005). Due to the smalldi¡erence between N

13C of diagenetic calcite and N13C of benthic foraminiferal tests the slope of the resulting mixing is small and

the regression is masked by the non-diagenetic variation of the N13C values. The variation range of N

18O for the benthic(V2.5x) is only slightly lower than for the planktonic foraminiferal tests (V3x), suggesting at most only minor di¡erentialdiagenesis for pairs of benthic and planktonic foraminifera.

PALAEO 3174 9-9-03

D. Stu«ben et al. / Palaeogeography, Palaeoclimatology, Palaeoecology 199 (2003) 107^127 111

seawater is related to salinity either as a conse-quence of mixing between watermasses with dif-ferent salinities and isotopic compositions, or ofchanges in the precipitation^evaporation cycle(e.g. Wol¡ et al., 1999; Pierre, 1999; Rohlingand de Rijk, 1999). However, a correlation be-tween N

18O and N13C of foraminiferal carbonate

is to be expected only if salinity changes are dueto a higher input of continental run-o¡ depletedin both 18O and 13C. Simple mass and isotopebudget calculations (Martin and Letolle, 1979;Be¤dard et al., 1981), using rough end memberestimates, indicate that the observed correlationis unlikely due to mixing. If we assume that lateCretaceous seawater di¡ers in salinity and isotopecomposition by only a few x from actual values(e.g. O’Neil, 1987), and that fresh water input

(run-o¡/precipitation) for latitudes correspondingto the paleogeographic position of the study area(20‡N) has typical values (salinity 1^2 psu; N18Oabout 35 to 37x), then a shift in N

18O of about3^4x (as observed in our data set) would implysalinities between 7 and 20 psu. This correspondsto a fresh water contribution of about 45^80%.Similar conclusions may be drawn from theN18O/salinity relationship in waters of moderntropical/subtropical oceans where the slope ofthe regression line is in the range of 0.18^0.25(Schmidt, 1999; Wol¡ et al., 1999). Similar values(0.25^0.27) are reported for the Mediterraneanbasin, with a climate regime in which evaporationis in excess to fresh water input (Pierre, 1999).Based on these data, the observed shift of 3^

4x in N18O at Elles would correspond to a drop

Fig. 4. Plot of N18O versus Sr/Ca ratio of benthic and planktonic foraminifera from the Elles section. The cluster of N18O and Sr/Ca suggests minor di¡erential diagenetic alteration e¡ects.

PALAEO 3174 9-9-03

D. Stu«ben et al. / Palaeogeography, Palaeoclimatology, Palaeoecology 199 (2003) 107^127112

in salinity of 12^22 psu. Such brackish conditionsare ruled out by the presence of normal tropicalto subtropical open ocean planktonic foraminifer-al assemblages (Abramovich and Keller, 2002). Inaddition, the slope of the N

18O/N13C regressionline is much higher than that reported for corre-lations relating on mixing (e.g. 0.53^0.57, Khimand Park, 2000). Nevertheless, it cannot be com-pletely ruled out that simultaneous decreases inN13C and N

18O of the Elles section are at leastpartly related to increased fresh water input andassociated terrigenous debris. This may have beenthe case at the intervals around 7 and 22.5 mdiscussed below.(c) Meteoric groundwaters are characterized by

lower N18O (V310x in low latitudes; Faure,

1986) and N13C values compared to seawater.

Thus, in addition to a decrease in initial N18O

values, varying degrees of cementation and/or re-crystallization due to meteoric waters will alsoproduce a correlation between the N

18O andN13C, and the samples therefore plot on a mixingline which links the initial N18O and N

13C values offoraminifera with that of the diagenetic cement(Jenkyns et al., 1994; Mitchell et al., 1997; Sakaiand Kano, 2001). In contrast, the addition of aconstant amount of diagenetic calcite to eachsample merely shifts all samples to more negativepositions in Fig. 3, but will not induce a lineararray. In this case, the oxygen isotope values can-not be interpreted in terms of absolute tempera-ture, but the relative variations between them re-£ect primary signals, even if attenuated.Fig. 3 shows that planktonic foraminifera at

Elles narrowly cluster along a linear regressionline, suggesting recrystallization, test overgrowth,and/or cement in¢llings. The greater variation inthe N

18O and N13C signals along the mixing line

for planktonic species apparently suggests a high-er degree of alteration or diagenetic in¢lling thanin benthic foraminifera. It is therefore likely thatat least the unusually low N

18O values of theplanktonic foraminifera are a result of diageneticalteration due to interaction with meteoric waters,and/or recrystallization due to burial and associ-ated thermal e¡ects. The variability in isotopicvalues is likely a function of the ratio of originaltest carbonate to diagenetic carbonate in the form

of chamber in¢lling. Based on the regression linebetween N

18O and N13C, the diagenetic cement in

the Elles pro¢le is estimated to average at V38to 39x for N

18O and V0.3 to 0.7x for N13C

(Fig. 3).The benthic foraminiferal tests at Elles show no

signi¢cant correlation between N18O and N

13C.However, the lack of correlation not compulsorilymeans that the benthic formaminifers are not af-fected by diagenetic overprint. Mitchell et al.(1997) observed that benthic foraminiferal testsare often less a¡ected by diagenetic overprintsthan planktonic foraminiferal tests. Althoughthis statement may be generally valid, it is veryunlikely that in the scale of a single sample,planktonic and benthic formaminifers were af-fected to signi¢cantly di¡erent degrees by diagen-esis. It is more likely that due to the small di¡er-ence between N

13C of the diagenetic calcite andN13C of the benthic foraminiferal tests the slopeof the resulting mixing is small and the regressionis masked by the non-diagenetic variation of theN13C values. The variation range of N18O for thebenthic foraminiferal tests (V2.5x) is onlyslightly lower than for the planktonics (V3x),indicating no signi¢cant di¡erence in the degree ofdiagenetically induced changes.It can be concluded that equal degrees of alter-

ation would reduce di¡erences in both N18O

(vN18O) and N13C values (vN13C) between pairs

of planktonic and benthic species, whereas stron-ger diagenesis of planktonic species compared tothe benthic species would reduce vN

13C andslightly increase vN

18O (see the insert in Fig. 3).N18O values, therefore, cannot be used for abso-lute temperature calculations, though estimates oftemperature di¡erences and variations in surface-to-deep gradients should be valid.

3.2. Diagenetic e¡ects on Sr/Ca ratios

Sr/Ca ratios of foraminiferal tests may also bea¡ected by diagenetic alteration. Strontium up-take by foraminifera and incorporation in theirtests is known to be higher than Sr concentrationsof abiotic calcite (Elder¢eld et al., 1996). Duringdiagenetic recrystallization, Sr is released from theforaminiferal tests (Richter and Liang, 1993).

PALAEO 3174 9-9-03

D. Stu«ben et al. / Palaeogeography, Palaeoclimatology, Palaeoecology 199 (2003) 107^127 113

Without transport, the Sr2þ concentration in thesolution rapidly builds up to a point that furtherreaction could not lower the Sr/Ca of the solid(Baker et al., 1982). Thus, where only minoramounts of Ca are added to or minor amountsof Sr are removed from the system, diagenetice¡ects will not produce signi¢cant changes inthe Sr/Ca ratios.If temperature is the major control, foraminif-

eral N18O and foraminiferal Sr/Ca ratios will becorrelated negatively. A strong shift in N

18O andloss of Sr by the same diagenetic process will pro-duce a positive correlation. However, a moderatediagenetically induced shift in both or in N

18Oalone will shift, weaken, or even destroy an exist-ing negative correlation between N

18O and Sr/Ca.Di¡erential diagenesis (i.e. di¡erent degrees of al-teration of benthic and planktonic forms withinone sample) results in an overestimation of tem-perature gradients from N

18O and an underestima-tion from Sr/Ca ratios. Thus, the temperature gra-dients estimated from N

18O and those estimatedfrom Sr/Ca ratios will diverge. Non-di¡erentialdiagenesis will reduce temperature gradientsfrom N

18O, but temperature gradients obtainedfrom Sr/Ca ratios remain nearly unchanged. AtElles, N18O versus Sr/Ca shows ‘remainders’ of anegative correlation and similar variation rangesof the N18O and Sr/Ca values for both benthic andplanktonic foraminiferal tests. This indicates thatpairs of benthic and planktonic foraminiferal testswere not seriously a¡ected by di¡erential diage-netic alteration (Fig. 4).

4. Results

4.1. Stable isotopes

N13C isotopes: Planktonic N

13C values at Ellesshow a progressive decrease through the upperMaastrichtian beginning in the lower part ofzone CF2 and continuing into CF1, whereasbenthic N

13C values show a corresponding in-crease, thus reducing the surface-to-deep gradient.Are these long-term trends primary, or do theyre£ect an increasing diagenetic overprint fromthe bottom to the top of the section? Since the

di¡erence in N18O for planktonic and benthic fo-

raminifera is nearly constant across the entire pro-¢le, an increasing diagenetic overprint up-sectionand serious e¡ects of di¡erential diagenesis seemimprobable. The coupled but opposite changes inthe isotopic records of the planktonic and benthicspecies suggest a continuous decrease in surfacebioproductivity over the last 400 kyr (zones CF1and CF2) of the Maastrichtian (Fig. 5). This isalso indicated by the surface-to-deep N

13C gra-dient, which shows a gradual decrease in zonesCF2 and CF1 (Fig. 6).The decreasing trend in bioproductivity is inter-

rupted by several shorter excursions of higher am-plitude in N

13C values and increased v13C gra-

dients. The ¢rst conspicuous deviation in benthicvalues (0.7x negative shift) and increased gra-dient occur between 21.5 and 23.5 m across theCF2/CF3 boundary (Figs. 5 and 6). The suddendrop in N

13C values is likely related to terrestrialorganic matter in£ux, rather than bioproductivity,as a result of erosion and a lower sea level, as alsoobserved in other Tunisian sections (Li et al.,2000). The second deviation is a positive excur-sion in planktonic foraminifera and increasedv13C gradient between 2 and 4 m below the K/Tboundary, followed by a strongly reduced v

13Cgradient in the interval just below the K/T bound-ary. In the shallow shelf environment of the upperMaastrichtian in Tunisia these carbon isotopevariations may be related to a combination ofchanges in bioproductivity, sea level, or secularvariations in seawater carbon isotopes.Comparison of the deep-sea record with that of

the Tunisian sections suggests the presence of re-gional signals. During late Campanian the deep-sea N

13C record shows relatively low values be-tween 0.5x and 1.0x, as also observed at Elles,and minimum values at the Campanian/Maas-trichtian transition (not observed at Elles, thoughpresent at El Kef) coincident with maximum glob-al cooling (Barrera, 1994, Barrera et al., 1997; Liand Keller, 1998a,b). However, deep-sea signalsand Tunisian continental shelf signals divergedduring the Maastrichtian. In the deep sea, N13Cvalues increased by 1.5^2x in bottom and sur-face waters in the early Maastrichtian and re-mained relatively high through the upper Maas-

PALAEO 3174 9-9-03

D. Stu«ben et al. / Palaeogeography, Palaeoclimatology, Palaeoecology 199 (2003) 107^127114

Fig. 5. Late Maastrichtian litholog, N18O, N13C, and Sr/Ca pro¢les of the Elles section for benthic (Cibicidoides pseudoacuta) and planktonic (Rugoglobigerina rugo-sa) foraminifera (¢ve-point smoothed data). Dashed arrows indicate the general trend in N

13C (p=planktonic; b=benthic). The cross-over of Sr/Ca ratios ofbenthic and planktonic foraminifera is caused by a general inter-species shift of Sr/Ca levels due to di¡ering vital e¡ects of benthic and planktonic foraminifera.Sea level curve after Li et al. (1999). Magnetostratigraphy after Harland et al. (1990).

PALAEO3174

9-9-03

D.Stu« ben

etal./P

alaeogeography,Palaeoclim

atology,Palaeoecology

199(2003)

107^127115

PALAEO 3174 9-9-03

D. Stu«ben et al. / Palaeogeography, Palaeoclimatology, Palaeoecology 199 (2003) 107^127116

trichtian (Li et al., 2000). But in Tunisian sectionsMaastrichtian N

13C values in surface and deepwaters are variable and generally 2^3x lowerrelative to the open ocean (D’Hondt and Lin-dinger, 1994; Barrera, 1994). Similarly low N

13Cvalues have been observed in the upper Maas-trichtian of Egypt (Keller et al., 2002). This sug-gests a primarily regional record and re£ects arelatively constant terrestrial organic in£ux oflighter carbon.

N18O isotopes: N

18O values of planktonic andbenthic foraminifera display roughly similartrends in low-frequency cycles with varying gra-dients, but exhibit convergent and divergent epi-sodes in the high-frequency range (Fig. 5). The¢rst convergent episode and minimum surface-to-deep gradient coincide with the benthic N

13Cexcursions between 21.5 and 3.5 m and zoneCF2/CF3 transition. Benthic and planktonicN18O covary and change towards more positivevalues in this interval, though with a signi¢cantlymore positive shift in planktonic values. Thissuggests climatic cooling of both surface andbottom waters in this relatively shallow continen-tal shelf environment. Generally lighter N18O val-ues in zone CF2 suggest a period of warmingthat culminates between 11 and 16 m. A secondminimum in the v

18O gradient associated withmore positive planktonic N

18O values, indicatescooling between 8 and 11 m at the CF1/CF2zone boundary. Between 4 and 8 m generallymore negative planktonic N

18O values suggestwarming. This interval is followed by strongly os-cillating and generally more positive values be-tween 2 and 4 m below the K/T boundary, whichsuggests unstable climatic and environmental con-ditions. In the last 2 m below the K/T boundary,benthic and planktonic values covary, followed byan abrupt increase at the K/T boundary (see Stu«-ben et al., 2002, for details of the K/T boundaryevent).

4.2. Sr/Ca results

The incorporation of Sr into foraminiferal cal-cite is controlled by the Sr/Ca ratio of seawater,temperature, salinity, pH, pressure, and vital ef-fects (Lea et al., 1999; Rathburn and De Deckker,1997; Rosenthal et al., 1997). Sea level £uctua-tions a¡ect Sr £ux into the oceans. At times ofsubsiding sea levels, Sr-rich aragonite from con-tinental shelfs exposed to weathering alters to cal-cite; because calcite can incorporate less Sr thanaragonite, the released Sr is relatively rapidlytransported into the oceans (90% of Sr in shelfcarbonate within less than 100 000 years; Stolland Schrag, 1996, 1998). Therefore, sea level £uc-tuations are characterized by variations in the Srcontent of sediment layers and the foraminiferaincorporate Sr into their shells in equilibriumwith seawater (e.g. Graham et al., 1982; Stolland Schrag, 1996, 2001). The sea level dependencyof the Sr/Ca ratio can be overprinted by the e¡ectof changing temperature, salinity, pH, (Lea et al.,1999; Malone and Baker, 1999) and pressure(Rosenthal et al., 1997; V100 Wmol/mol per1000 m), as well as by non-equilibrium vital ef-fects. Laboratory experiments revealed that the Srdistribution coe⁄cient between calcite and solu-tion increases with temperature (Malone andBaker, 1999). From live cultures Lea et al.(1999) obtained positive slopes for the dependen-cies of Sr uptake in foraminiferal calcite with tem-perature (11^18 Wmol/mol per ‡C), salinity (10Wmol/mol per psu) and pH (69^148 Wmol/molper unit), whereas increasing pressure decreasedthe Sr uptake by 100 Wmol/mol per 1000 m(Rosenthal et al., 1997). The Elles section wasdeposited at middle to outer neritic depths atV100 m (Li et al., 1999; Abramovich et al.,2002). Due to low water depth during sedimentdeposition the pressure dependency can be ne-glected, but a decreasing sea level may have sig-

Fig. 6. Comparison of temperature di¡erences calculated from v18O and Sr/Ca (vSr/Ca) between planktonic and benthic forami-

nifera. vT18O is calculated after Erez and Luz (1992), based on the di¡erences of N18O. vTSr is calculated from vSr/Ca using cali-bration data from culture experiments (Lea et al., 1999). Dots are raw data, lines are ¢ve-point smoothed data. Magnetostratigra-phy after Harland et al. (1990).

PALAEO 3174 9-9-03

D. Stu«ben et al. / Palaeogeography, Palaeoclimatology, Palaeoecology 199 (2003) 107^127 117

Fig. 7. Late Maastrichtian litholog, mineralogical composition (phyllosilicates), and calcite/detritus ratio (detritus =quartz+phyllosilicates; magnetic susceptibility rawand ¢ve-point smoothed data) of the Elles section in comparison with the sea level curve of Li et al. (1999). Magnetostratigraphy after Harland et al. (1990).

PALAEO3174

9-9-03

D.Stu« ben

etal./P

alaeogeography,Palaeoclim

atology,Palaeoecology

199(2003)

107^127118

ni¢cantly increased riverine Sr £ux into the ocean.For this reason, one might argue for a more rapidresponse in seawater Sr/Ca ratios in shallow waterdepositional environments such as Elles than inthe deep sea, even though the long residencetime of Sr in the water column delays the responseto increasing sea levels.At Elles the Sr/Ca ratios of benthic and plank-

tonic foraminifera generally covary between ratiosof 1.6 and 1.8 mmol/mol near the base of the sec-tion and between 1.7 and 1.9 mmol/mol throughthe upper CF3 to CF2 intervals (Fig. 5). In CF1benthic Sr/Ca values gradually increase between1.8 and 2.0 mmol/mol with planktonic values sig-ni¢cantly higher (up to 2.2 mmol/mol) andstrongly £uctuating between 3 and 5 m belowthe K/T boundary. In the 2 m below the K/Tboundary Sr/Ca values of both planktonic andbenthic foraminifera reach peak values and thendecline near the boundary. Across the entire sec-tion Sr/Ca ratios of benthic and planktonic fora-minifera increase to the top. If this general trendis mainly temperature controlled, then N

18O valuesshould display a steady decrease from bottom totop, which is not the case. This indicates that atElles temperature is not the major control for theSr/Ca ratios in benthic and planktonic foraminif-era.The increasing Sr/Ca trend suggests lower sea

levels near the CF3/CF2 boundary, in the middleof CF1 and again near the top of CF1 (Fig. 5).This interpretation seems to be supported by thehigh quartz content at the CF3/CF2 transitionand a short hiatus or condensed interval, andthe high quartz content between 1 and 2 m belowthe K/T boundary (Fig. 7). But the correlation isnot precise and the low detrital in£ux in CF1corresponds to the low quartz in£ux. The rela-tively poor correlation may be due to the lag-time in the release of Sr to seawater (dependingon the sedimentation rate, the lag-time may cor-respond to 0.2^0.8 m at 10 000 years). Alterna-tively, the high Sr/Ca values in CF1 may be dueto increased chemical weathering, rather than thesea level fall, as suggested by warmer tempera-tures (Fig. 5) and clay mineralogy that re£ectsincreasingly humid conditions at Elles (Adatte etal., 2002).

Although Sr is presumed to be homogeneous inits distribution in the water column (e.g. de Vil-liers, 1999), the gradient of the Sr/Ca ratios (vSr/Ca) between benthic and planktonic foraminiferavaries with depth because of the dependency ofthe distribution coe⁄cient between seawater andforaminiferal calcite on sea level, salinity, pH, andtemperature changes. Consequently, the surface-to-deep gradient of the Sr/Ca ratio will mainlybe controlled by the surface-to-deep temperatureand/or salinity gradient.The low but nevertheless signi¢cant cross-cor-

relation coe⁄cient (r=0.254, n=230, P6 0.1%)of the surface-to-deep Sr/Ca ratio (vSr/Ca) andthe v

18O surface-to-deep temperature gradient(vT, Fig. 6) suggests that the surface-to-deepSr/Ca ratio is mainly controlled by variations inthe temperature di¡erence of surface and bottomwaters. Low surface-to-deep Sr/Ca gradients cor-respond to cool intervals and high gradients towarm climatic intervals. These £uctuations inthe Sr/Ca ratios of benthic and planktonic fora-minifera may re£ect variations between surfacewater and sediment surface temperatures. E¡ectsof pressure dependency (Rosenthal et al., 1997)related to short-term sea level £uctuations are ofminor importance in this shallow water section.Additional variability in the Sr/Ca ratios maybe introduced by vital e¡ects (resulting in di¡erentSr contents in planktonic and benthic foraminif-era), or by varying local gradients of the Srconcentration in the water column as a conse-quence of variations in the riverine £ux (Stoll etal., 1999).

4.3. Paleotemperature trends

Although diagenetic alteration of foraminiferaltests at Elles shifts the original temperature sig-nals to warmer values, relative changes in plank-tonic and benthic foraminifera and long-termtrends are preserved (e.g. Schrag et al., 1992,1995; Sakai and Kano, 2001; Pearson et al.,2001). As a result, calculated temperatures, partic-ularly of planktonic species, tend to be signi¢-cantly higher than the original temperature val-ues. Nevertheless, temperature calculations ofsomewhat recrystallized planktonic and benthic

PALAEO 3174 9-9-03

D. Stu«ben et al. / Palaeogeography, Palaeoclimatology, Palaeoecology 199 (2003) 107^127 119

foraminiferal tests are useful to evaluate surface-to-deep temperature gradients and watermassstrati¢cation. For this purpose, temperature andvT values were calculated for both planktonicand benthic species based on Erez and Luz(1992) (Fig. 6). A Sr/CaForam thermometer basedon the foraminiferal data of Lea et al. (1999)(vTSr) was applied to the variations in the sur-face-to-deep Sr/Ca gradient for comparison. Dif-ferences in salinity of bottom and surface watersare likely due to density di¡erences, with highersalinities expected in bottom waters. Reduced sa-linity of the surface waters by dilution with freshwater would increase the apparent temperaturedi¡erence obtained from the N

18O thermometer,but reduce the temperature di¡erence from theSr/CaForam thermometer. The similar temperaturetrends and range of variations indicate that thee¡ects of diagenesis and salinity variations ofthe surface waters on long-term temperaturetrends are relatively minor.As would be expected, planktonic foraminifera

show greater variations than benthic species be-cause surface waters are more sensitive to temper-ature £uctuations than bottom waters. Both long-term and short-term trends are apparent. Long-term trends in the lower and upper parts of theElles section (zones CF3 and CF1 intervals) showrelatively close benthic and planktonic values,whereas in zone CF2 they diverge. The sametrend is observed in the vTSr and vT N

18O values(Fig. 6). This can be explained by changes in sealevel, climate, and mixing of the water column,with all of them in£uencing the observed temper-ature trends.A lower sea level prevailed during deposition of

the CF3 interval with a sea level lowstand at theCF2/CF3 boundary (Li and Keller, 1998c; Li etal., 2000; Abramovich et al., 2002). The divergentbenthic and planktonic values in zone CF2 corre-spond to a rising sea level and global climatewarming (Stott and Kennett, 1990; Barrera,1994; Li and Keller, 1998b) and suggest increasedwatermass strati¢cation. The global warming con-tinued into CF1, accompanied by warmer bottomwater temperatures, and was followed by coolingin the uppermost Maastrichtian. At Elles, the cli-mate warming of CF1 appears to be re£ected by

increased mixing of the water column and some-what warmer bottom water temperatures. Thereare also three short-term episodes of strongly con-vergent N18O values that re£ect cooling of surfacewaters and strong mixing of the water column,possibly accompanied by lower sea levels. Theseepisodes occurred at the CF3/CF2 and CF2/CF1transitions and in the upper part of CF1 (Fig. 6).

4.4. Mineralogy

Relative changes in bulk rock compositions in-dicate variations in sediment sources and re£ectthe variable intensity of weathering and erosionunder arid and humid climates (Chamley, 1989;Weaver, 1989), as well as the variable in£ux ofterrigenous sediments into the oceans duringhigh and low sea levels. High detrital in£ux (phyl-losilicates and quartz) is generally associated withincreased erosion and transport during lower sealevels and seasonally cool climate. In contrast,high carbonate deposition is generally associatedwith humid warm climates and transgressive seas.At Elles, periods of increased phyllosilicates (Fig.7) and quartz contents correspond to lower sealevels and increased erosion near the CF3/CF2boundary and near the top of the section belowthe K/T boundary. In zone CF2, the decreaseddetrital in£ux (lower quartz and phyllosilicates)suggests a rising sea level, whereas a decreasingsea level is suggested in CF1 by decreased quartz,but high phyllosilicate content. The sea levelchanges inferred from mineralogical proxies aresupported by sea level changes from biostrati-graphic data (Li et al., 1999, 2000), which revealsea level lowstands at V21 m (65.5 Ma) and arapidly decreasing sea level in the uppermost 3 m(100 kyr) of the Maastrichtian. The time controlis calculated from the sedimentation rates derivedfrom variations of the cyclicity of the magneticmass susceptibility data discussed below.The general trend of increasing calcite/detritus

ratios between V21 m (65.5 Ma) and 3 m (65.1Ma) is overlain by high- and low-frequency sinoi-dal £uctuations, which show minima suggestinglower sea levels at V23 m (V65.53 Ma), V15^18 m (65.40^65.44 Ma), 13 m (65.38 Ma), andV8 m (65.27 Ma). Maxima in Ca/detritus ratios

PALAEO 3174 9-9-03

D. Stu«ben et al. / Palaeogeography, Palaeoclimatology, Palaeoecology 199 (2003) 107^127120

indicate higher sea levels at V19 m (65.46 Ma),V14 m (V65.39 Ma), 12 m (65.36 Ma) and 1 m(V65.03 Ma, Fig. 6). These local sea level varia-tions derived from the Ca/detritus ratio at Ellesgenerally coincide with the sea level curve of Li etal. (1999) from mid-Atlantic DSDP Site 525A.

4.5. Magnetic mass susceptibility

Magnetic susceptibility of sediments is relatedto the mineralogical composition of the sedimentsand depends on grain size and concentration ofmagnetic minerals. Only Fe-bearing rock-formingminerals show positive susceptibility, with ferri-magnetic minerals, like magnetite, orders of mag-nitudes higher than the susceptibility of all otherrock-forming minerals. Thus, even in low concen-trations these minerals dominate the bulk suscep-tibility of sediments (Dekkers, 1997). Sea levelchanges alter the detrital input, resulting in possi-ble shifts in mineral sources and types. In thesupposedly ice-free world of the late Cretaceous,sea level changes are unlikely to be driven by Mi-lankovich forcing although weathering conditionsstrongly depend on climate, which is in£uenced byvariations in insolation. Changing weatheringconditions mainly a¡ect riverine input and eolianin£ux, which e¡ectively act in the Milankovichfrequency range (Tauxe, 1993).At Elles, magnetic mass susceptibility varies

from 4 to 16 WSI/g and generally mimics the trendof phyllosilicates with a reasonable correlation be-tween intervals of high phyllosilicates and highmagnetic susceptibility (r=0.48, n=291, P6 0.00;Fig. 7). The long-term quartz trend does not cor-relate with either phyllosilicates or magnetic sus-ceptibility in zone CF1, possibly due to changes inclimate to more humid seasonal cool conditionsand hence decreased quartz transport (Adatte etal., 2002). Moreover, short-term cycles are evidentwith a good correlation between increased phyllo-silicate input and high magnetic susceptibility.The magnetic susceptibility record at Elles thus

exhibits both a long-term trend and short-termvariations in the amount and composition of ter-rigenous sediment £ux relative to biogenic car-bonate production, and these are governed by cli-mate and local sea level changes. Magnetic

susceptibility is tracking changes in weatheringintensity on adjacent continents and, unlike sealevel changes in an ice-free world, they likelyvary with Milankovich-driven climate variations.

4.6. Time series analysis

To determine periodicities in the sediment col-umn at Elles, time series analysis was performedon magnetic mass susceptibility data and the Sr/Ca and C and O isotope values of benthic (Cibi-cidoides pseudoacuta) and planktonic (Rugoglobi-gerina rugosa) foraminifera. Before applying timeseries analysis, variations in sedimentation rateswere evaluated based on susceptibility and bio-stratigraphic data. Based on planktonic foraminif-eral biostratigraphic data sediment accumulationrates for the combined CF1^CF2 interval aver-ages of 4.88 cm/kyr, or 3.2 cm/kyr for CF1 and8 cm/kyr for CF2 were obtained.In the susceptibility data two main short cycles

are apparent with high-frequency variationssuperimposed at about 0.8^1.0 m and 1.6 m inter-vals (Fig. 7). This suggests the presence of 20 kyrand 40 kyr cycles based on biostratigraphic agecorrelations and an average sediment accumula-tion rate of 4.88 cm for the CF1^CF2 interval,which spans the 450 kyr of chron 29R below theK/T boundary (Li and Keller, 1998a,b). The dom-inating very low frequent sinoidal periodicity ofthe susceptibility data is assumed to represent theV400 kyr Milankovich cycle. The wavelength ofthis periodicity increases with depth. The clearlyvisible elongated periodicity in the lower half ofthe pro¢le re£ects increased sedimentation rates inCF2/CF3 as compared to CF1.To account for the e¡ects of varying sedimen-

tation rates, a sinus function was ¢tted to the longperiodicity, and the sedimentation rate was step-wise adjusted in order to get the best possible ¢tfor the long periodicity. After this stepwise adjust-ment, estimated sedimentation rates average 3.2cm/kyr for the interval between 0 and 6.4 m(V65^65.2 Ma), 2.5 cm/kyr for the interval be-tween 6.6 and 9.8 m (V65.2^65.3 Ma), 8 cm/kyrbetween 9.8 and 17.6 m (V65. 3^65.4 Ma), and5 cm/kyr for the older succession (17.8^31.45 m).The average adjusted sedimentation rate is 5.04

PALAEO 3174 9-9-03

D. Stu«ben et al. / Palaeogeography, Palaeoclimatology, Palaeoecology 199 (2003) 107^127 121

cm/kyr for CF1 and CF2, which is close to theaverage sedimentation rate of 4.88 cm/kyr esti-mated from biostratigraphy. Similar results wereobtained by adjusting the sedimentation ratesbased on ¢tting the 20 kyr cycle to the suscepti-bility record (Hambach and Adatte, 2000).Based on these stepwise adjusted sedimentation

rates, time series analysis was carried out usingthe REDFIT program (Schulz and Mudelsee,2002), in order to isolate the shorter Milankovichcycles (20, 40 and 100 kyr). REDFIT is based onthe SPECTRUM algorithm (Schulz and Statteg-ger, 1997), which allows the analysis of unevenlyspaced time series. REDFIT estimates the rednoise spectra from unevenly spaced time seriesby ¢tting a ¢rst order autoregressive, AR1, func-tion. To reduce the e¡ects of white noise the rawdata were smoothed by a ¢ve-point moving aver-age, using weights derived from a parabolicsmoothing function (Savitzky and Golay, 1964).The resulting periodograms for N18O, N13C, and

Sr/Ca ratios for benthic and planktonic foraminif-era as well as for the magnetic susceptibility dis-play clear low and short periodicities of 20Y 3,

40Y 5 and 100Y 15 kyr, which correspond to Mi-lankovich cycles. However, the peaks are belowthe 80% false alarm level for red noise, exceptfor susceptibility data. For the surface-to-deepgradients, v18O (vT), v13C, and (vSr/Ca), thesecyclicities are more pronounced and the 20 kyrprecision cycle, 40 kyr obliquity cycle, and eccen-tricity cycle of V100 kyr can be identi¢ed at the80% and for vT at the 95% signi¢cance level (Fig.8). A 60 kyr resonance (possibly the half of the127 kyr eccentricity cycle) is present in the perio-dograms of N

18O, N13C, and Sr/Ca ratios for

benthic and planktonic foraminifera, but is nottraceable in the surface-to-deep gradients.

5. Discussion

5.1. Climate and productivity: the last 700 kyr ofthe Maastrichtian

During the last 700 kyr of the Maastrichtianthe eastern Tethys (Tunisia) experienced majorvariations in climate and productivity that re£ect

Fig. 8. Red noise normalized energy spectrum of the periodicities from time series analysis of magnetic mass susceptibility,vN

18O, and vSr/Ca for the Elles Maastrichtian section showing 20, 40, and 100 kyr Milankovitch cycles. The lines 95% s.l. and80% s.l. mark the Monte Carlo simulations of the 95% and 80% signi¢cance levels. The time series analysis is based on sedimen-tation rates of 3.2 cm/kyr for CF 1, 2.5 cm/kyr for CF2, and 5^8 cm/kyr for CF2 to CF3.

PALAEO 3174 9-9-03

D. Stu«ben et al. / Palaeogeography, Palaeoclimatology, Palaeoecology 199 (2003) 107^127122

global paleoceanographic changes. Based on ahigh-resolution database that includes various pa-leoceanographic proxies, including N

13C, N18O, Sr/Ca, mineralogical and magnetic susceptibility dataof the Elles section in Tunisia, the environmentalhistory can be reconstructed.Between 65.7 and 65.5 Ma (CF3) relatively high

but gradually decreasing N18O values, a £uctuat-

ing and decreasing surface-to-deep temperaturegradient, and an increasing N

13C gradient indicateclimate cooling and relatively high productivity. Amajor positive excursion in N

18O between 65.50and 65.55 Ma (CF3/CF2 boundary, 21.5^23.5m) marks maximum upper Maastrichtian cooling(Figs. 5 and 6), whereas the £uctuating N

18O val-ues, and very low surface-to-deep Sr/Ca ratiosand temperature gradients indicate strong mixingof the water column during cooling (Figs. 5 and6). The major drop in benthic N

13C values, butonly minor decrease in surface waters, probablyre£ects changes in carbon isotopes of the oceanictotal dissolved carbon, and increased carbon £uxdue to erosion and weathering related to the cli-matic cooling and accompanying sea level fall.This stratigraphic interval corresponds to a globalcooling maximum in the upper Maastrichtian(Barrera, 1994; Barrera et al., 1997; Li and Kel-ler, 1998a,b), associated with a sea level fall andwidespread hiatus (Haq et al., 1987; Keller andStinnesbeck, 1996; Li et al., 1999, 2000). A shorthiatus or condensed interval may be present atthis interval at Elles as suggested by abrupt fau-nal, isotopic, and mineralogical changes (Figs. 5and 7; Adatte et al., 2002; Abramovich and Kel-ler, 2002).After the cooling maximum, surface and bot-

tom water temperatures rapidly increased, reach-ing the ¢rst of two warm events in CF2 (12^16 m,65.33^65.38 Ma), whereas the N13C gradient grad-ually decreased due to lower surface N

13C values.This warm event marks the onset of a globalshort-term warming which is best documented atDSDP Site 525 (Walvis Ridge, mid-latitude SouthAtlantic), where temperatures are estimated tohave increased by 3^4‡C in both surface and in-termediate waters (Li and Keller, 1998b), but thesame warm event has also been recognized in nu-merous other deep-sea sections from high to low

latitudes (Shackleton et al., 1984; Stott and Ken-nett, 1990; Barrera et al., 1997; Olsson et al.,2001). A warming-up period indicates oceanicturnover and probably decreasing thermohalinecirculation during a sea level rise and conse-quently decreasing productivity. The likely causefor this short-term global warming is increasedgreenhouse gases probably from major DeccanTrap volcanism estimated between 65.2 and 65.4Ma (Courtillot et al., 1988, 1996; Baski, 1994;Ho¡mann et al., 2000).Surface waters cooled markedly across the

CF2^CF1 transition (11^8 m, 65.33^65.26 Ma)and productivity continued to decrease as indi-cated by decreasing surface N

13C, and v13C gra-

dients. Another short-term warming in CF1 (8^4 m, 65.26^65.12 Ma) and strongly diverging Sr/Casuggest an increased surface-to-deep temperaturegradient during the upper Maastrichtian. At thistime, primary productivity reached minimum val-ues for the upper Maastrichtian, as indicated bymajor decreases in surface and bottom water N13Cand a decreasing surface-to-deep gradient (Figs. 5and 6).Major climate cooling occurred in surface

waters during the upper part of CF1 (2^4 m,65.12^65.04 Ma) and was accompanied by highbut variable primary productivity £uxes andstrong mixing evident in Sr/Ca ratios. Thisshort-term cooling phase was also observed inthe South Atlantic Site 525A (Li and Keller,1998a) and re£ects a change in watermasses.The increase in surface productivity probably trig-gered a change in the isotope composition of thetotal dissolved inorganic carbon reservoir, whichmay have been enhanced by increased carbonburial and decreased carbon £ux from weatheringdue to a sea level transgression.The last 40 kyr below the K/T boundary are

marked by very unstable paleo-oceanographicconditions, rapid warming, and strong mixing ofsurface and bottom waters. The surface-to-deeptemperature gradient decreased to almost zero,which corresponds to a decrease in the Sr/Ca ratioof planktonic foraminifera. Surface productivitydecreased dramatically while the productivity ofbenthic foraminifera increased slightly. Similarobservations were made based on carbon isotope

PALAEO 3174 9-9-03

D. Stu«ben et al. / Palaeogeography, Palaeoclimatology, Palaeoecology 199 (2003) 107^127 123

studies from various localities (South Atlantic, ElKef, Stevns Klint, Nye KlZv) but were interpretedas a consequence of increasing benthic values,rather than decrease in surface productivity (Kel-ler and Lindinger, 1989; Schmitz et al., 1992;Keller et al., 1993; Li and Keller, 1998a). Thissuggests that the benthic carbon isotope increaseduring the last 100 kyr of the Maastrichtian pri-marily re£ects changes in the rate and source ofdeep water productivity, rather than reduced or-ganic carbon oxidation as a result of lower sur-face productivity (Zachos and Arthur, 1986).

6. Summary and conclusions

High-resolution N13C, N18O, Sr/Ca, mineralogi-

cal and magnetic susceptibility measurements ofhemipelagic sediments spanning the last 700 kyrat Elles, Tunisia, indicate major changes in cli-mate, productivity and sea level. Although diage-netically overprinted, the general trends are pre-served in stable isotopes, Sr/Ca, the surface-to-deep gradient, and mineralogical and susceptibil-ity records. The following conclusions can bereached:(1) Correlations of vT (derived from v

18O andvSr), primary productivity (v13C), and magneticsusceptibility with Milankovich cycles indicatethat climate, weathering, and nutrient conditionsin the Kasserine Island area are mainly driven byvariations in insolation, whereas sea level changesduring the presumably ice-free Cretaceous worldare assumed to be tectonically driven.(2) Climate changes observed in N

18O of plank-tonic foraminifera are more pronounced in thesurface-to-deep gradients of v18O and vSr.(3) Cooling between 65.7 and 65.5 Ma coincides

with increased N13C values, very low Sr/Ca ratios,

and a major sea level fall that indicates strongmixing of waters and increased carbon £ux dueto erosion.(4) Rapid warming occurred between 65.38 and

65.33 Ma and was accompanied by a decrease insurface productivity that continued through thesubsequent rapid cooling of surface waters be-tween 65.33 and 65.26 Ma (CF2^CF1 transition).(5) The long-term trend in decreasing primary

productivity parallels a sea level rise between 65.4and 65.2 Ma (zones CF2 and CF1) and is prob-ably due to decreased thermohaline circulationassociated with climate warming.(6) During the last 200 kyr of the Maastricht-

ian, rapidly increasing Sr/Ca ratios associatedwith warm and cool temperatures may be relatedto increased chemical weathering and changes inbioproductivity, rather than a sea level fall, assuggested by clay mineralogy.The results from the Elles succession demon-

strate that it is possible to separate paleoclimaticfrom diagenetic signals by using a multiproxy ap-proach, assuming that the response to diagenesisis systematically di¡erent for each proxy. Corre-lation of the climate, productivity and sea leveltrends at Elles with similar trends at DSDP Site525A in the mid-latitude South Atlantic supportsthis approach and also indicates that these areglobal, rather than local trends.

Acknowledgements

This study was funded by the Deutsche For-schungsgemeinschaft (STU 169/20-1, Sti 128/4-1), NSF INT 95-04309 (G.K.), the Swiss NationalFund No. 8220-028367, and US-Israel BinationalFoundation (BSF) Grant 9800425. Markus Leos-son is greatly acknowledged for the measurementof the stable isotope. We thank the reviewer H.Stoll and B. Malmgren for their helpful commentsand suggestions for improvement of this manu-script.

References

Abramovich, S., Almogi-Labin, A., Benjamini, C., 1998. De-cline of the Maastrichtian pelagic ecosystem based onplanktic foraminiferal assemblage changes: Implicationsfor the terminal Cretaceous faunal crisis. Geology 26, 63^66.

Abramovich, S., Keller, G., 2002. High stress late Maastricht-ian paleoenvironment: inference from planktic foraminiferain Tunisia. Palaeoceanogr. Palaeoclimatol. Palaeoecol. 178,145^164.

Abramovich, S., Keller, G., Adatte, T., Stinnesbeck, W., Hot-tinger, L., Stu«ben, D., Berner, Z., Ramanivosoa, B., Ran-driamanantenasoa, A., 2002. Age and Paleoenvironment ofthe Maastrichtian-Paleocene of the Mahajanga Basin, Ma-

PALAEO 3174 9-9-03

D. Stu«ben et al. / Palaeogeography, Palaeoclimatology, Palaeoecology 199 (2003) 107^127124

dagascar: a multidisciplinary approach. Mar. Miccropaleon-tol. 47, 17^70.

Adatte, T., Stinnesbeck, W., Keller, G., 1996. Lithostrati-graphic and mineralogic correlation of near K/T boundaryclastic sediments in northeastern Mexico; implications fororigin and nature of deposition. In: Ryder G., Fastovsky,D., Gastner, S. (Eds.), the Cretaceous-Tertiary Event andother Catastrophes in Earth History. Geol. Soc. Am. Spec.Pap. 307, 211^226.

Adatte, T., Keller, G., Stinnesbeck, W., 2002. Late Cretaceousto early Paleocene climate and sea level £uctuations; theTunisian record. Palaeogeogr. Palaeoclimatol. Palaeoecol.178, 165^196.

Baker, P.A., Gieskes, J.M., Elder¢eld, H., 1982. Diagenesis ofcarbonates in deep-sea sediments ^ evidence from Sr/Ca ra-tios and interstitial dissolved Sr2þ data. J. Sediment. Petrol.52, 71^82.

Barrera, E., 1994. Global environmental changes preceding theCretaceous-Tertiary boundary: early-late Maastrichtiantransition. Geology 22, 877^880.

Barrera, E., Huber, B.T., 1990. Evolution of Antarctic watersduring the Maastrichtian: Foraminifera oxygen and carbonisotope ratio, Leg 113. Proc. ODP Sci. Results 113, 813^827.

Barrera, E., Savin, S.M., Thomas, E., Jones, C.E., 1997. Evi-dence for thermohaline-circulation reversals controlled bysea level change in the latest Cretaceous. Geology 25, 715^718.

Baski, A.K., 1994. Geochronological studies on whole-rockbasalts, Deccan Traps, India; evaluation of the timing ofvolcanism relative to the K-T boundary. Earth Planet. Sci.Lett. 121, 43^56.

Be¤dard, P., Hillaire-Marcel, C., Page¤, P., 1981. 18O modellingof freshwater inputs in Ba⁄n Bay and Canadian Arcticcoastal waters. Nature 293, 287^289.

Burollet, P.F., 1967. General Geology of Tunisia. Pet. Explor.Soc. Libya, 9th Annu. Field Conf., Tripoli, pp. 51^58.

Chamley, H., 1989. Clay Sedimentology. Springer, 623 pp.Courtillot, V., Feraud, G., Maluski, H., Vandamme, D., Mor-eau, M.G., Besse, J., 1988. Deccan £ood basalts and theCretaceous/Tertiary boundary. Nature 333, 843^846.

Courtillot, V., Jaeger, J.J., Yang, Z., Feraud, G., Hofmann,C., 1996. The in£uence of continental £ood basalts on massextinctions; where do we stand? In: Ryder, G., Fastovosky,D., Gartner, S. (Eds.), The Cretaceous-Tertiary Event andOther Catastrophes in Earth History. Geol. Soc. Am. Spec.Pap. 307, 513^525.

Dekkers, M.J., 1997. Environmental magnetism: an introduc-tion. Geol. Mijnbouw 76, 163^182.

de Villiers, S., 1999. Seawater strontium and Sr/Ca variabilityin the Atlantic and Paci¢c oceans. Earth Planet. Sci. Lett.171, 623^634.

D’Hondt, S., Lindinger, M., 1994. A stable isotopic record ofthe Maastrichtian ocean-climate system: south AtlanticDSDP Site 528. Palaeogeogr. Palaeoclimatol. Palaeoecol.112, 363^378.

Elder¢eld, H., Bertram, C.J., Erez, J., 1996. A biomineraliza-tion model for the incorporation of trace elements into fo-

raminiferal calcium carbonate. Earth Planet. Sci. Lett. 142,409^423.

Erez, J., Luz, B., 1992. Temperature control of oxygen-isotopefractionation of cultured planktonic foraminifera. Nature297, 220^222.

Faure, G. 1986. Principles of Isotope Geology, 2nd edn. Wiley,589 pp.

Graham, D.W., Bender, M.L., Williams, D.F., Keigwin, L.D.,Jr., 1982. Strontium calcium ratios in Cenozoic planktonicforaminifera. Geochim. Cosmochim. Acta 46, 1281^1292.

Hambach, U., Adatte, T., 2000. High-resolution magnetic sus-ceptibility records from the Maastrichtian and Lower Paleo-cene of NW Tunesia. Sixth Int. Cretaceous Symp., Vienna.

Hambach, U., Rose, Th., Krumsiek, K., 2002. Magnetic sus-ceptibility measurements on reseach cores kirchrode I andII: indication for sea level variations and Milankovitchforced cyclicity in the Albian. Cretac. Res. (submitted).

Haq, B.U., Hardenbol, J., Vail, P.R., 1987. Chronology of£uctuating sea levels since the Triassic. Science 235, 1156^1167.

Harland, W.B., Armstrong, R.L., Cox, A.V., Craig, L.E.,Smith, A.G., Smith, D.G., 1990. A geologic time scale,1989. Cambridge Earth Science Series. Cambridge Univer-sity Press, Cambridge, 131 pp.

Ho¡mann, C., Feraud, G., Courtillot, V., 2000. 40Ar/39Ar dat-ing of mineral separates and whole rocks from the WesternGhats lava pile: further constraints on duration and age ofDeccan traps. Earth Planet. Sci. Lett. 180, 13^27.

Jenkyns, H.C., Gale, A.S., Cor¢eld, R.M., 1994. Carbon- andoxygen-isotope stratigraphy of the English Chalk and ItalianScaglia and its paleoclimatic signi¢cance. Geol. Mag. 131, 1^34.

Keller, G., Lindinger, M., 1989. Stable isotope, TOC andCaCO3 records across the Cretaceous/Tertiary boundary atEl Kef, Tunisia. Palaeogeogr. Palaeoclimatol. Palaeoecol.73, 243^265.

Keller, G., Barrera, E., Schmitz, B., Matsson, E., 1993. Grad-ual mass extinction, species survivorship, and long termenvironmental changes across the Creataceous-Tertiaryboundary in high latitudes. Geol. Soc. Am. Bull. 105, 979^997.

Keller, G., Stinnesbeck, W., 1996. Sea level changes, clasticdeposits and megatsunamis across the Cretaceous-Tertiaryboundary. In: MacLeod, N., Keller, G. (Eds.), Cretaceous/Tertiary Boundary Mass Extinction: Biotic and Environ-mental Changes. W.W. Norton, New York, pp. 415^450.

Keller, G., Adatte, T., Stinnesbeck, W., Luciani, V., Karoui,N., Zaghbib-Turki, D., 2002. Paleobiogeography of the Cre-taceous-Tertiary mass extinction in planktic foraminifera.Palaeogeogr. Palaeoclimatol. Palaeoecol. 178, 257^298.

Khim, B.-K., Park, B.K., 2000. Recent change of hydro-graphic features in the Cheju Strait (southeastern YellowSea), Korea: contrast between planktonic and benthic fora-minifera isotope composition. Mar. Geol. 169, 411^419.

Kucera, M., Malmgren, B.A., 1998. Terminal Cretaceouswarming event in the mid-latitude South Atlantic Ocean;evidence from poleward migration of Contusotruncana con-

PALAEO 3174 9-9-03

D. Stu«ben et al. / Palaeogeography, Palaeoclimatology, Palaeoecology 199 (2003) 107^127 125

tusa (planktonic Foraminifera) morphotypes. Palaeogeogr.Palaeoclimatol. Palaeoecol. 38, 1^15.

Lea, D., Mashiotta, T., Spero, H., 1999. Controls on magne-sium and strontium uptake in planktonic foraminifera deter-mined by live culturing. Geochim. Cosmochim. Acta 63,2369^2379.

Li, L., Keller, G., 1998a. Maastrichtian climate, productivityand faunal turnovers in planktic foraminifera in south At-lantic DSDP Sites 525 A and 21. Mar. Micropaleontol. 33,5^86.

Li, L., Keller, G., 1998b. Diversi¢cation and extinction inCampanian-Maastrichtian planktic Foraminifera of north-western Tunisia. Eclogae Geol. Helv. 91, 75^102.

Li, L., Keller, G., 1998c. Abrupt deep-sea warming at the endof the Cretaceous. Geology 26, 995^998.

Li, L., Keller, G., 1999. Variability in Late Cretaceous climateand deep waters: evidence from stable isotope. Mar. Geol.161, 171^190.

Li, L., Keller, G., Stinnesbeck, W., 1999. The Late Campanianand Maastrichtian in northwestern Tunisia: paleoenviron-mental inferences from lithology, macrofauna and benthicforaminifera. Cretac. Res. 20, 231^252.

Li, L., Keller, G., Adatte, T., Stinnesbeck, W., 2000. LateCretaceous sea level changes in Tunisia: a multi-disciplinaryapproach. J. Geol. Soc. Lond. 157, 447^458.

Malone, M., Baker, P., 1999. Temperature dependence of thestrontium distribution coe⁄cient in calcite; an experimentalstudy from 408 degrees to 2008 degrees C and application tonatural diagenetic calcites. J. Sediment. Res. 69, 216^223.

Martin, J.M., Letolle, R., 1979. Oxygen 18 in estuaries. Nature282, 292^294.

McLeod, K.G., Huber, B.T., Pletsch, T., Ro«hl, U., Kucera,M., 2001. Maastrichtian foraminiferal and paeoceano-graphic changes on Milankovitch timescales. Paleoceanogra-phy 16, 133^154.

Mitchell, S.F., Ball, J.D., Crowley, S.F., Marshall, J.D., Paul,C.R.C., Veltkamp, C.J., Samir, A., 1997. Isotope data fromcretaceous chalks and foraminifera: Environmental or dia-genetic signals? Geology 25, 691^694.

Olsson, R.K., Wright, J.D., Miller, K.G., 2001. Paleobiogeog-raphy of Pseudotextularia elegans during the latest Maas-trichtian global warming event. J. Foraminif. Res. 31, 275^282.

O’Neil, J.R., 1987. Preservation of H, C, O isotopic ratios inthe low temperature environment. In: Kyser, T.K. (Ed.),Short Course in Stable Isotope Geochemistry of Low Tem-perature Fluids. Miner. Assoc. Can., pp. 85^128.

Pearson, P.N., Ditch¢eld, P.W., Singano, J., Harcourt-Brown,K.G., Nicholas, C.J., Olsson, R.K., Shackleton, N.J., Hall,M.A., 2001. Warm tropical sea surface temperatures in theLate Cretaceous and Eocene epochs. Nature 413, 481^487.

Pierre, Ch., 1999. The carbon and oxygen isotope distributionin the Mediterranean water mases. Mar. Geol. 153, 41^55.

Rathburn, A.E., De Deckker, P., 1997. Magnesium and stron-tium compositions of recent benthic foraminifera from theCoral Sea, Australia and Prydz Bay, Antarctica. Mar. Mi-cropaleontol. 32, 231^248.

Richter, F.M., Liang, Y., 1993. The rate and consequences ofSr diagenesis in deep-sea carbonates. Earth Planet. Sci. Lett.177, 553^565.

Rohling, E.J., de Rijk, S., 1999. Holocene Climate Optimumand Last Glacial Maximum in the Mediterranean: the ma-rine oxygen isotope record. Mar. Geol. 153, 57^75.

Rosenthal, Y., Boyle, E.D., Slowley, N., 1997. Temperaturecontrol on the incorporation of magnesium, strontium, £u-orine and cadmium into benthic foraminiferal shells fromLittle Bahama Bank: Prospects for thermocline paleocean-ography. Geochim. Cosmochim. Acta 61, 3633^3643.

Sakai, S., Kano, A., 2001. Original oxygen isotopic composi-tion of planktic foraminifera preserved in diagenetically al-tered Pleistocene shallow-marine carbonates. Mar. Geol.172, 197^204.

Savitzky, A., Golay, M.J.E., 1964. Smoothing and di¡erentia-tion of data by simpli¢ed least squares procedures. Anal.Chem. 36, 1627^1639.

Schmidt, G.A., 1999. Forward modelling of carbonate proxydata from planktonic foraminifera using oxygen isotopetracers in a global ocean model. Paleoceanography 14,482^497.

Schmitz, B., Keller, G., Stenvall, O., 1992. Stable isotope andforaminiferal changes across the Cretaceous/Tertiary bound-ary at Stevns Klint, Denmark: Arguments for long-termoceanic instability before and after bolide impact. Palaeo-geogr. Palaeoclimatol. Palaeoecol. 96, 233^260.

Schrag, D.P., 1999. E¡ects of diagenesis on the isotopic recordof late paleogene tropical sea surface temperatures. Chem.Geol. 161, 215^224.

Schrag, D.P., DePaolo, D.J., Richter, F., 1992. Oxygen isotopeexchange in a two-layer model of oceanic crust. Earth Plan-et. Sci. Lett. 111, 305^317.

Schrag, D.P., DePaolo, D.J., Richter, F.M., 1995. Recon-structing past sea surface temperatures: correcting for dia-genesis of bulk marine carbon. Geochim. Cosmochim. Acta59, 2265^2278.

Schulz, M., Mudelsee, M., 2002. Spectrum: Spectral analysisof unevenly spaced paleoclimatic time series. Comput. Geo-sci. 23, 929^945.

Schulz, M., Stattegger, K., 1997. REDFIT: estimating red-noise spectra directly from unevenly spaced paleoclimatictime series. Comput. Geosci. 28, 421^426.

Scotese, C.R., Sager, W.W., 1988. Plate tectonic reconstruc-tions of the Cretacious and Cenozoic ocean basins. Tecto-nophysics 155, 27^48.

Shackleton, N.J., Hall, M.A., Boersma, A., 1984. Oxygen andcarbon isotope data from Leg 74 foraminifera. DSDP Init.Rep. 74, 599^612.

Spero, H.J., Bijma, J., Lea, D.W., Bemis, B.E., 1997. E¡ect ofseawater carbonate concentration on foraminiferal carbonand oxygen isotopes. Nature 390, 497^500.

Spicer, R.A., Cor¢eld, R.M., 1992. A review of terrestrial andmarine climates in the Cretaceous with implications formodeling the ‘Greenhouse’. Earth Geol. Mag. 129, 2, 169^180.

Stinnesbeck, W., Schulte, P., Lindenmaier, F., Adatte, T.,

PALAEO 3174 9-9-03

D. Stu«ben et al. / Palaeogeography, Palaeoclimatology, Palaeoecology 199 (2003) 107^127126

A¡olter, M., Schilli, L., Keller, G., Stueben, D., Berner, Z.,Kramar, U., Burns, S.J., Lopez-Oliva, J.G., 2001. LateMaastrichtian age of spherule deposits in northeastern Mex-ico; implication for Chicxulub scenario. Can. J. Earth Sci.38, 229^238.

Stoll, H.M., Schrag, D.P., 1996. Evidence for glacial control ofrapid sea level changes in the early Cretaceous. Science 272,1771^1774.

Stoll, H.M., Schrag, D.P., 1998. E¡ects of Quaternary sea levelcycles on strontium in seawater. Geochim. Cosmochim.Acta 62, 1107^1118.

Stoll, H.M., Schrag, D.P., Clemens, S.C., 1999. Are seawaterSr/Ca variations preseved in Quaternary foraminfera? Geo-chim. Cosmochim. Acta 63, 3535^3547.

Stoll, H.M., Schrag, D.P., 2001. Sr/Ca variations in Creta-ceous carbonates: relation to productivity and sea levelchanges. Palaeoceanogr. Palaeoclimatol. Palaeoecol. 168,311^336.

Stott, L.D., Kennett, J.P., 1990. The paleoceanographic andpaleoclimatic signature of the Cretaceous/Paleogene bound-ary in the Antarctic: Stable isotopic results from ODP Leg113. Proc. ODP Sci. Results 113, 829^848.

Stu«ben, D., Kramar, U., Berner, Z., Stinnesbeck, W., Keller,G., Adatte, Th., 2002. Trace elements and stable isotopes inforaminifera of the Elles II K/T pro¢le: Indications for sea

level £uctuations and primary productivity. Palaeogeogr.Palaeoclimatol. Palaeoecol. 178, 321^345.

Tauxe, L., 1993. Sedimentary records of relative paleointensityof the geomagnetic ¢eld: theory and practice. Rev. Geophys.31, 319^354.

Weaver, C.E., 1989. Clays, Muds and Shales. Developments inSedimentology. Elsevier, 44 pp.

Wol¡, T., Grieger, B., Hale, W., Du«rkoop, A., Mulitza, S.,Pa«tzold, J., Wefer, G., 1999. On the reconstruction of Paleo-salinities. In: Fischer, G., Wefer, G. (Eds.), Use of Proxiesin Paleoceanography: Examples from the South Atlantic.Springer, Berlin, pp. 207^228.

Zaier, A., Beji-Sassi, A., Sassi, S., Moody, R.T.J., 1998. Basinevolution and deposition during the early Paleogene in Tu-nisia. In: Macgregor, D.S., Moody, R.T.J., Clark-Lowes,D.D. (Eds.), Petroleum Geology of North Africa. Geol.Soc. Lond. Spec. Publ. 132, 375^393.

Zachos, J.C., Arthur, M.A., Thumell, R.C., Williams, D.F.,Tappa, E.J., 1985. Stable isotope and trace element geo-chemistry of carbonate sediments across the Cretaceous/Ter-tiary boundary at Deep Sea Drilling Project Hole 577, Leg86. Init. Rep. Deep Sea Drill. Proj. 86, 513^532.

Zachos, J.C., Arthur, M.A., 1986. Paleoceanography of theCretaceous/tertiary boundary event, inferences from stableisotopic and other data. Paleoceanography 1, 5^26.

PALAEO 3174 9-9-03

D. Stu«ben et al. / Palaeogeography, Palaeoclimatology, Palaeoecology 199 (2003) 107^127 127