2.a. Geological and stratigraphic setting

ODP Site 762 (19°53.24′S, 112°15.24′E) was drilled at a water depth of 1360 m in the western part of the central Exmouth Plateau, off NW Australia in the eastern Indian Ocean (Fig. 1) (Haq et al. Reference Haq, von Rad and O'Connell1990). Sediments were deposited in an upper bathyal setting and consist of Cenozoic and Upper Cretaceous nannofossil oozes and chalk, with varying amounts of foraminifers and clay. Biostratigraphic analysis of calcareous nannofossils and planktonic foraminifera suggest a mostly complete Coniacian–Paleocene sequence (Bralower & Siesser, Reference Bralower, Siesser, von Rad and Haq1992; Wonders, Reference Wonders, von Rad and Haq1992). Magnetostratigraphic analysis of the recovered succession from Hole 762C allowed recognition of all polarity chrons from upper Santonian Chron C34n to upper Eocene Chron C13r (Galbrun, Reference Galbrun, von Rad and Haq1992; Corbin et al. Reference Corbin, Galbrun, Renard and Emmanuel1995). These previous magnetostratigraphic studies were based on a sampling resolution of two samples for each 1.5 m core section. Therefore, polarity chron boundaries were not very precisely defined. In addition, the natural remanent magnetization (NRM) intensities of these chalks are very weak and a few samples (about 15%) could not be assigned to normal or reversed polarity. In order to better constrain the lower and upper limits of Chron C31r, additional magnetic analyses were performed on 7 samples across the C31r/C31n boundary and 13 samples across the C32n/C31r boundary with a resolution of 0.15 m. Magnetic measurements were carried out with a cryogenic CTF 3-axis magnetometer. All samples were demagnetized by alternating field (AF) up to 30 or 35 mT. This treatment was adequate to isolate the primary magnetization component of normal or reversed polarity (Fig. A1 in online Appendix at http://www.journals.cambridge.org/geo). These new data allow the location of the Chron C31r boundaries with an accuracy of a few centimetres (Fig. 2). At this site, polarity Chron C31r is therefore exactly 13.3 m thick.

Figure 1. Mollweide palaeogeographic reconstruction for the early Maastrichtian (70 Ma) modified from Global Tectonics Home Page (http://www.serg.unicam.it/Reconstructions.htm) showing location of ODP Site 762 (Indian Ocean) and the Contessa Highway section near Gubbio (Italy).

Figure 2. ODP Hole 762C, Exmouth Plateau, NW Australia, Indian Ocean. (a) Interval analysed and nannofossil events (this study): last occurrences (LO) of Broinsonia parca group (1), Tranolithus orionatus (2) and Reinhardtites levis (3). (b) Lithology, palaeoinclinations (this study and previous data from Galbrun, Reference Galbrun, von Rad and Haq1992) and identification of magnetic polarity Chron C31r. (c) Grey level signal after suppression of lighting effect artificial cycles and cracks. (d) Cyclostratigraphic analysis of grey level variations: spectral analysis of the entire signal using the multitaper method (top), significance test of the spectrum using the F-test method and amplitude spectrogram (bottom).

Polarity Chron C31r is characterized by key nannofossil events, which are well identified at Site 762: the last occurrence (LO) of Reinhardtites levis followed by the closely spaced LOs of Tranolithus orionatus and Broinsonia parca group (Fig. 2).

2.b. Cyclostratigraphic data: colour variations

On this rather old ODP Leg 122, none of the usual palaeoclimatic proxies were measured with a sample density high enough for a cyclostratigraphic purpose. Maastrichtian pelagic carbonates show regular, decimetre-scale alternations of dark and light coloured layers (Fig. A2 in online Appendix at http://www.journals.cambridge.org/geo). According to Huang, Boyd & O'Connell (Reference Huang, Boyd, O'Connell, von Rad and Haq1992), diagenesis is not responsible for the colour cycles observed, and light layers are characterized by a biogenic input higher than the terrigenous supply. A previous low resolution cyclostratigraphic analysis of these alternations highlighted an orbital control of the sedimentation (Huang, Boyd & O'Connell, Reference Huang, Boyd, O'Connell, von Rad and Haq1992). More recent informatic tools allow the measurement of colour variations on core photographs at a very high resolution. In this study, an encoding of the sediment colour in greyscale has been performed using a free image analysis software, Image-J (http://rsb.info.nih.gov/ij/). A grey level value was attributed to each pixel along a line traced across the length of each core section, ranging from 0 (black) to 250 (white). The grey level signal obtained along a 1.5 m core has a length of c. 2100 pixels, which corresponds to a sampling resolution of c. 0.7 mm.

As noticed in Cramer (Reference Cramer2001), the edges of core photographs are darker than the centre, due to the flash of the camera. This may induces artificial cycles with a high power, and weakens the expression of natural cyclicities. Their periods are of 1.5 m (length of the section) and 8.5 to 9 m (length of the cores). This lighting effect was suppressed by filtering the corresponding frequencies. In addition, every core presents cracks, which are coded by very low grey level values and represent a major source of noise. A function created with MATLAB® allowed the suppression of data with grey values lower than a threshold chosen by the user.

3.a. Geological and stratigraphic setting

The Cretaceous–Palaeogene pelagic successions of the Umbria–Marche Apennines (Italy) played a major role in the development of the GTS. The upper part of the Cretaceous–Palaeogene Scaglia Rossa Formation in the Bottaccione section, near Gubbio (Fig. 1), has been the subject of multiple bio- and magnetostratigraphic studies allowing dating of the C-sequence marine magnetic anomalies (Alvarez et al. Reference Alvarez, Arthur, Fischer, Lowrie, Napoleone, Premoli Silva and Roggenthen1977; Lowrie & Alvarez, Reference Lowrie and Alvarez1977; Premoli Silva, Paggi & Monechi, Reference Premoli Silva, Paggi and Monechi1977; Monechi & Thierstein, Reference Monechi and Thierstein1985). The Contessa section (43°22′47″N, 13°33′49″E) crops out in a valley parallel to the Bottaccione Gorge, about 3 km NW from Gubbio. The magnetostratigraphy of this section from the Cretaceous–Palaeogene boundary to the top of Chron C31r was established by Chauris et al. (Reference Chauris, Lerousseau, Beaudoin, Propson and Montanari1998), who located the C31n/C31n boundary at 315.5 m, and recognized the extinction of inoceramids at the 312 m level. In this study, the magnetostratigraphic analysis of the section was extended downward, considering a potential thickness of at least 20 m for Chron C31r, as established in the Bottaccione section (Lowrie & Alvarez, Reference Lowrie and Alvarez1977).

Twenty-five samples between 314.9 and 292.6 m were measured with a LETI/CEA 3-axis RS-01 cryogenic magnetometer and subjected to progressive thermal demagnetization up to 580–600°C. The demagnetizing behaviour is quite similar to the well-known magnetic properties of the Scaglia Rossa limestones (Channell et al. Reference Channell, Freeman, Heller and Lowrie1982; Lowrie et al. Reference Lowrie, Alvarez, Napoleone, Perch-Nielsen, Premoli Silva and Toumarkine1982; Chauris et al. Reference Chauris, Lerousseau, Beaudoin, Propson and Montanari1998; Galeotti et al. Reference Galeotti, Angori, Coccioni, Ferrari, Galbrun, Monechi, Premoli Silva, Speijer and Turi2000): an initial viscous remanent magnetization was removed by 150°C, a secondary reversed polarity component was removed over the range 150–350°C and finally the characteristic remanent magnetization (ChRM) was clearly defined above 400°C (Fig. A1 in online Appendix at http://www.journals.cambridge.org/geo). Directions of the ChRM were determined with a routine PCA technique. Most of the samples are of reversed polarity and correspond to Chron C31r, except the last three samples at the base of the studied interval, which yield a normal polarity (top of Subchron C32n.1n) (Fig. 3).

Figure 3. The Contessa Highway section, Gubbio, Umbria–Marche Apennines (Italy). (a) Interval studied and nannofossil events (this study): last occurrences of Reinhardtites levis (1), Tranolithus orionatus (2) and Broinsonia parca group (3). (b) Lithology, palaeoinclinations (the white circle at the top of the studied interval is from Chauris et al. Reference Chauris, Lerousseau, Beaudoin, Propson and Montanari1998) and identification of magnetic polarity Chron C31r. (c) Magnetic susceptibility (MS) variations. (d) Cyclostratigraphic analysis of MS variations: spectral analysis using the multitaper method (top), significance test of the spectrum using the F-test method and amplitude spectrogram (bottom).

The total thickness of C31r at Contessa is 21.4 m, thus similar to the thickness of this magnetozone in the Bottaccione section. Polarity Chron C31r in the Contessa Highway section is also characterized by LOs of nannofossil taxa: Broinsonia parca group, Tranolithus orionatus and Reinhardtites levis (Fig. 3). The reversed order of nannofossil LOs that exists between the two sites is primarily owing to different biogeographic settings (Site 762 lies in a ‘boundary zone’ between subtropical and austral provinces while the Contessa section is a Tethyan site), and to preservational control (poor preservation in the indurated limestones of the Contessa section).

3.b. Cyclostratigraphic data: magnetic susceptibility variations

Magnetic susceptibility (MS) measures the capacity of a substance to acquire magnetization when submitted to an external magnetic field. Usually, in sedimentary marine records, the high-frequency MS variations are directly related to the terrigenous supply in oceanic basins, and are therefore a good palaeoclimatic proxy (Ellwood et al. Reference Ellwood, Crick, Hassani, Benoist and Young2000). MS is easily and quickly measurable on samples or directly on sedimentary cores, and is a non-destructive tool. MS is thus commonly used in cyclostratigraphic studies (Weedon et al. Reference Weedon, Jenkyns, Coe and Hesselbo1999; Boulila et al. Reference Boulila, Galbrun, Hinnov and Collin2008a, Reference Boulila, Hinnov, Huret, Collin, Galbrun, Fortwengler, Marchand and Thierryb).

MS was measured on samples from the Contessa section with a sample step of 5 to 10 cm from 293.8 to 314.9 m using a Kappabridge KLY-2. The amplitude of the cycles expressed in the MS signal increases upward, reaching 1 × 10⁻⁸ m³/kg (Fig. 3). This could indicate an increase in the terrigenous supply.

4.a. Spectral methods

Two methods have been used to detect and identify the cycles present in the signal and characterize their evolution: spectral analysis by multitaper method (MTM) (Thomson, Reference Thomson1982) and amplitude spectrograms (Maurer, Hinnov & Schlager, Reference Maurer, Hinnov, Schlager, D'Argenio, Fischer, Premoli Silva, Weissert and Ferreri2004).

The MTM analysis has been applied using the freeware Analyserie (Paillard, Labeyrie & Yiou, Reference Paillard, Labeyrie and Yiou1996), along with a harmonic F-test allowing evaluation of the significance of spectral peaks detected and the levels of confidence of the spectral analysis. Prior to the analysis, a linear de-trending was performed. Remaining long-term cycles were removed by a polynomial de-trending of the data. The method of frequency ratios (Mayer & Appel, Reference Mayer and Appel1999) was then applied to test the link between the cycles detected in the signals and variations of the orbital parameters, using the orbital frequencies estimated for the Cretaceous Period (Laskar et al. Reference Laskar, Robutel, Joutel, Gastineau, Correia and Levrard2004). The amplitude spectrogram method is based on a succession of spectral analysis, performed along the signal within a sliding window. The power of the detected frequencies is highlighted in colour (red for high values, blue for low values) (Figs 2, 3). This technique illustrates timewise shifts in frequencies that result from variations of the sedimentation rate (Maurer, Hinnov & Schlager Reference Maurer, Hinnov, Schlager, D'Argenio, Fischer, Premoli Silva, Weissert and Ferreri2004; Meyers, Sageman & Hinnov, Reference Meyers, Sageman and Hinnov2001). A transfer from the spatial to the temporal scale has been performed by orbital tuning. This tuning allows the correction of the variations in sedimentation rate and an easier comparison between the two studied sites. The tuning method consists of an attempt to tie a recorded sedimentary cycle to the corresponding orbital parameter extracted from the theoretical astronomical solution of the same epoch (Hinnov, Reference Hinnov2000). For stages prior to the Eocene, such as the Maastrichtian, the chaotic character of the motions in the solar system increases the uncertainties of the astronomical solution. The long eccentricity cycle is the only parameter that can be calculated with accuracy throughout the Cretaceous Period (Laskar et al. Reference Laskar, Robutel, Joutel, Gastineau, Correia and Levrard2004). Therefore, the orbital tuning of each record is based on the identification of 405 ka eccentricity cycles.

4.b. ODP Hole 762C

The grey level signal shows clear cyclic variations of various frequencies (Fig. 2). The amplitude of the low-frequency cycles is higher in the lower half of the record. A first spectral analysis covering the entire C31r allowed detection of cycles with periods ranging from 0.26 to 3.70 m, with low-frequency cycles displaying the highest power (Fig. 2). Using the method of frequency ratios, cycles with a wavelength of 0.26 m are attributed to an obliquity forcing, 0.48 to 0.59 m cycles are attributed to the 100 ka eccentricity cycle, and 2.17 to 3.70 m cycles are attributed to the 405 ka eccentricity cycle. Precession and obliquity cycles are detected with difficulty, owing to the high amplitude of low-frequency cycles and to the low sedimentation rate. Variability observed in the high-frequency peaks and the presence of a broad peak covering the low frequencies (Fig. 2) suggests possible variations in the sedimentation rate. Indeed, when the sedimentation rate varies, the wavelengths of the cycles recording an orbital forcing vary accordingly. Thus, a single parameter can be recorded by a cycle with a small wavelength when the sedimentation rate is low and a cycle with a higher wavelength when the sedimentation rate increases (Meyers, Sageman & Hinnov, Reference Meyers, Sageman and Hinnov2001).

In conjunction with spectral analyses, an amplitude spectrogram was performed on the whole signal, with a 2.5 m window and 10 cm steps. Three frequencies are well expressed in the spectrogram (Fig. 2). Comparison of the spectrogram and spectral analyses allows assignment of these frequencies to the obliquity and 100 and 405 ka eccentricity cycles. The 405 ka eccentricity variations are well expressed throughout the record, whereas 100 ka eccentricity and obliquity variations present discontinuities. In addition, a slight shift toward high frequencies of the 405 and 100 ka cycles from 613 to 607 m below sea floor (mbsf) indicates a decrease in the sedimentation rate (Fig. 2). The short appearance of several powerful peaks, observed on the spectrogram at 599.10 mbsf, is an artefact due to a gap in core recovery around 600 mbsf perturbating the spectral analyses. The disappearance of the 100 ka cycles and the attenuation of the expression of the 405 ka cycles characterize the initiation of a perturbation in the sedimentation rate. A cycle with a 0.15 m period appears in the high-frequency domain from 599 to 596.3 mbsf, which could correspond to a record of the precession evolution.

4.c. Contessa Highway section

High- and low-frequency cycles are clearly recognized throughout the signal. The periodogram performed on the entire section shows cycles with wavelengths ranging from 0.25 to 8.3 m (Fig. 3). Cycles with periods of 1.06 to 1.28 and 2.83 m are well expressed, whereas high-frequency cycles are scattered and have low power. This is probably owing to the small fluctuation of a relatively low sedimentation rate on this section, which hampers an accurate detection of high-frequency variations.

The method of frequency ratios has been applied to identify cycles resulting from an orbital forcing. Cycles with wavelengths of 0.25 to 0.27 m could be attributed to precession, and cycles with wavelengths of 0.44 m attributed to obliquity, though their power is low. The 100 and 405 ka eccentricity cycles are well expressed (Fig 3). The 8.3 m cycles probably characterize long-term palaeoenvironmental changes affecting the section, and may be related to the increase of the amplitude of the cycles in the upper part of the section. Sea-level variations and/or climate changes might be responsible for these long-term cycles, but an analysis of a much larger interval would be required to ascertain their nature.

The amplitude spectrogram shows defined 405 ka eccentricity cycles in the upper part of the section, whereas only the 100 ka cycles can be identified in the lower third of the record (Fig. 3). A progressive shift of the 100 ka eccentricity cycles toward higher frequencies between the lower and upper half of Chron C31r indicates a decrease in the sedimentation rate throughout the studied interval. Precession and obliquity cycles are better expressed at the base of the record, where the sedimentation rate is higher.