Sea-level change and remagnetization of continental shelf sediments off New Jersey (ODP Leg 174A): magnetite and greigite diagenesis

Transcription

1 February 17, 24 19:1 Geophysical Journal International gji2162 Geophys. J. Int. (24) 156, doi: /j X x Sea-level change and remagnetization of continental shelf sediments off New Jersey (ODP Leg 174A): magnetite and greigite diagenesis Hirokuni Oda 1,2 and Masayuki Torii 3 1 Institute for Marine Resources and Environment, Geological Survey of Japan, AIST, Tsukuba , Japan. hirokuni-oda@aist.go.jp 2 Palaeomagnetic Laboratory Fort Hoofddijk, Faculty of Earth Sciences, Utrecht University, Budapestlaan 17, 3584 CD Utrecht, the Netherlands. h.oda@geo.uu.nl 3 Department of Biosphere-Geosphere System Science, Okayama University of Science, Okayama 7-5, Japan Accepted 23 October 16. Received 23 October 13; in original form 23 February 5 SUMMARY Palaeomagnetic and rock magnetic studies were performed on a seaward transect of continental shelf sediments from sites 171 and 172 of Ocean Drilling Programme Leg 174A on the New Jersey margin. The sediments recorded a polarity change from reversed to normal, which was tentatively interpreted as the Matuyama Brunhes boundary. The polarity boundary is closely correlated with the sequence boundary pp3(s), which is considered to have formed as an erosional unconformity during regression and subsequent transgression related to global eustacy. Rock magnetic analyses indicate that the sediment contains comparable amounts of magnetite and greigite (Fe 3 S 4 ) with possible trace amounts of haematite. A marked change in rock magnetic properties is recorded at Site 172 with an increase in greigite concentration at the polarity boundary (62.4 metres below seafloor), where the grain size does not change. At Site 171, greigite is dominant from 2 m below pp3(s) to.8 m above it. These intervals are characterized by higher peak coercivities of remanence and subdued remanence in both the magnetite and greigite components. Palaeomagnetic analysis indicates that the intervals just below pp3(s) have dual-component magnetizations of reversed polarity and are considered to have been deposited during the Matuyama ron (C1r.1r) and were possibly remagnetized during the Brunhes ron through the formation of the pp3(s) surface. Detrital magnetite and early diagenetic greigite might have carried the reversed polarity magnetization. During the formation of pp3(s) in the Brunhes ron, greigite might have formed at the oxidation front with ongoing downward formation due to oxidation of pyrite by percolating fresh water, which might be a cause of the remagnetization. Our study indicates that careful rock magnetic investigation is necessary for magnetostratigraphic studies of continental shelf deposits in order to recognize remagnetizations induced by sea-level changes. Key words: diagenesis, glacial eustacy, greigite, isothermal remanent magnetization, magnetic hysteresis, magnetite, remagnetization, sea-level change. GJI Marine geoscience 1 INTRODUCTION Coring associated with Ocean Drilling Programme (ODP) Leg 174A was conducted on the New Jersey continental margin to unveil the sedimentary history controlled by eustatic sea-level changes. Stratigraphic interpretation was achieved by means of down-hole logging and detailed seismic stratigraphy (Austine et al. 1998; Metzger et al. 2; Delius et al. 21), because sediment core recovery rates were low due to drilling in shallow water and loss of unconsolidated sand from the core barrel. Sites 171 and 172 were drilled on the shelf at depths shallower than 1 m. These sites form a seaward transect together with Site 173 on the edge of the continental shelf (Fig. 1a). The recovered sediments consist of clay to sand of Miocene to Holocene age, which were deposited in a shallow marine environment. Several sequence stratigraphic boundaries, generated in response to sea-level fluctuations, were recognized (Austine et al. 1998). These sequence boundaries are mostly traceable from site to site on the seismic records (Fig. 1b). Determining the age of these boundaries was crucial for reconstructing the sedimentary history; however, this task was not easy due to the paucity and/or reworking of microfossils and to poor sediment recovery. Knowledge of the geomagnetic reversal stratigraphy, coupled with a thorough understanding of the origins of the carriers of the natural remanent magnetization, may help to constrain the age of deposition. At sites 171 and 172, a Plio-Pleistocene sequence boundary pp3(s) separates homogeneous clayey silts below from overlying thinly interbedded sandy silts and clayey sands (Fig. 2). The pp3(s) sequence boundary is interpreted as an erosional unconformity formed during regression that was further modified by erosion C 24 RAS 443

2 February 17, :1 Geophysical Journal International gji2162 H. Oda and M. Torii Figure 1. (a) Location map of the studied ODP drill sites. (b) Seismic line Oc27 Profile 885 along sites 171 and 172. Interpreted sequence boundaries are marked by thick broken lines. due to shoreface retreat during subsequent sea-level transgression (Metzger et al. 2). McCarthy & Gostlin (2) interpreted pp3(s) to have formed during a low-sea-level stand corresponding to marine oxygen isotope stage 12 (MIS ka). This is consistent with a large increase in sea level ( 16 m) estimated for the transition from glacial stage 12 to interglacial stage 11 (Thunell et al. 22). Benthic foraminifera indicate that the sediments below pp3(s) were deposited in inner neritic environments ( 5 m) at Site 171, and in upper middle neritic environments (5 65 m) at Site 172 (Austine et al. 1998). The sequence boundary pp3(s) is almost coincident with a reversed to normal polarity boundary and was tentatively interpreted as the Matuyama Brunhes boundary during ODP Leg 174A (Fig. 2). At Site 172, magnetic susceptibility abruptly decreases upward at the polarity boundary, which occurs 4.9 m below pp3(s). This susceptibility change may be related to a mineral magnetic change after deposition because grain size does not change at the boundary, and thus mineral magnetic study is important to constrain the palaeomagnetic polarity at the time of deposition. Some samples from ODP Leg 174A showed evidence for acquisition of gyroremanent magnetization (GRM) above 6 mt during alternating field demagnetization (AFD) (Austine et al. 1998), which might be an indication of the presence of single-domain grains such as greigite (Snowball 1997a). eigite has been reported in lacustrine sediments (Skinner et al. 1964; Dell 1972; C 24 RAS, GJI, 156,

3 February 17, 24 19:1 Geophysical Journal International gji2162 Magnetism of ODP site 171 and 172 sediments 445 Figure 2. Schematic lithological columns for sites 171 and 172. Cored intervals with sediment recovery are shown by black areas. Lithology by inference is based on cored material and down-hole logging. A Plio-Pleistocene sequence boundary pp3(s) is closely associated with the Matuyama Brunhes (M B) boundary, as discussed in this paper. All modified after Austine et al. (1998). Snowball & Thompson 199), in rapidly deposited marine sediments that were subsequently uplifted and exposed on land (e.g. Roberts & Turner 1993; Roberts 1995; Horng et al. 1998), and shallow marine Holocene sediments (Lee & Jin 1995). The bacterially mediated reduction of sulphate and H 2 S production in suboxic and anoxic sediments control the dissolution of iron oxides (Canfield & Berner 1987). This process is often accompanied by a subsequent conversion of magnetite by way of greigite into paramagnetic pyrite (Karlin & Levi 1983, 1985). eigite may be preserved if the pyritization process is arrested (Roberts & Turner 1993). In magnetostratigraphic studies it is important to understand the effects of diagenesis on both magnetic iron oxide and iron sulphide minerals in continental shelf sediments whose deposition is controlled by sea-level change. The present study was designed to identify the magnetic minerals at sites 171 and 172, to investigate the diagenetic effects of sea-level change on magnetic minerals, and to clarify the possibility of remagnetization below pp3(s). C 24 RAS, GJI, 156,

4 February 17, 24 19:1 Geophysical Journal International gji H. Oda and M. Torii 2 MATERIALS AND METHODS Palaeomagnetic samples were taken from clayey to silty sediments spanning the tentatively determined polarity boundary that occurs near sequence boundary pp3(s). Discrete samples (7 cm 3 cubes) were collected at stratigraphic intervals of approximately 1 cm from the working halves of cores from Holes 171C and 172A. Anisotropy of magnetic susceptibility (AMS) was measured with a susceptibility anisotropy meter (AGICO Kappabridge KLY-3S) at the Geological Survey of Japan (GSJ) to check for physical disturbance during or after sedimentation or coring. The degree of anisotropy (P J ) and the shape parameter (T) were calculated according to Jelinek (1981). Subsequently, the natural remanent magnetization (NRM) was measured with a 2G Enterprises superconducting rock magnetometer (SRM model 76) at the GSJ. Typical noise levels of the magnetometer are lower than Am 2. All the samples were subjected to progressive alternating field demagnetization (PAFD) at 8 steps up to 8 mt using a 2G Enterprises demagnetizer that is configured in line with the SRM. Demagnetization was conducted successively along the sample Z-, X - and Y-axes. After measurement of the NRM, an anhysteretic remanent magnetization (ARM) was imparted with a DC bias field of 1 µt applied parallel to the peak alternating field of 8 mt. Isothermal remanent magnetizations (IRMs) were imparted with a pulse magnetizer (2G Enterprises model 66) during stepwise acquisition up to 2.5 T. S.3T was also calculated according to Bloemendal et al. (1992): S.3T = [ (IRM.3T /SIRM) + 1]/2, (1) where SIRM is the saturation IRM, which was imparted with a pulsed magnetic field of 2.5 T, and IRM.3T is the magnetization after the application of a back-field of 3 mt. Kruiver & Passier (21) pointed out that the new definition of the S ratio by Bloemendal et al. (1992) is more logical than the classical definition of the S ratio by Thompson & Oldfield (1986) because the new definition directly expresses the ratio of saturation remanence carried by lower-coercivity minerals to the total magnetic mineral assemblage provided that no magnetic interactions occur. Magnetic hysteresis measurements were made on 12 selected samples using an alternating gradient magnetometer (MicroMag model 29, Princeton Measurements Corporation) at the University of Utrecht (the Netherlands) with a phenolic probe. Sediment samples were weighed, glued in to plastic sample cylinders while wet and measured immediately, to avoid problems caused by chemical change during drying. Hysteresis loops were obtained by cycling the applied magnetic field between +1 T and 1 T. Sample weights were 1 2 mg and the sensitivity of the measurements was 1nAm 2. Four samples were selected from each of holes 171C and 172A and were subjected to low-temperature magnetic measurements with a Quantum Design magnetic property measurement system (MPMS-XL5) at the GSJ. Samples were cooled in zero field to 6 K, at which temperature the 2.5 T DC field was applied for 6 s. The magnet was then reset to zero. The SIRM was then measured at 2 steps during warming up to 3 K. Two samples were selected from each of holes 171C and 172A and magnetic minerals were extracted with a magnetic finger (Kirschvink et al. 1992). X-ray diffraction (XRD) analyses were made on the magnetic extracts using a JDX-83W instrument (JEOL) at the GSJ. Diffractograms were collected between 5 and 55 (2θ) with Cu Kα radiation at.2 step scan, with 1 s measurement time per step. After XRD analysis, thermomagnetic analyses were made with a thermomagnetic balance at Kyoto University (noise level of Am 2 ). The measurements were made in a DC magnetic field of.85 T at a heating/cooling rate of 8 min 1 in air. 3 RESULTS 3.1 NRM demagnetization behaviour and magnetic mineralogy Due to coring with extended or rotary core barrels, the recovered sediments underwent relative rotation between adjacent sediment blocks, and palaeomagnetic declination values are not meaningful. Typical examples of PAFD results are shown for samples with stable normal polarity from Hole 171C in Fig. 3(a) and for samples with stable reversed polarity from Hole 172A in Fig. 3(b), respectively. Soft magnetic components were generally removed by demagnetization with a peak field of 2 mt. Some samples (Fig. 3c) have a higher-coercivity component with reversed polarity and an overlapping normal polarity component. For some other samples, unstable behaviour is observed (Fig. 3d), where the primary magnetization might have had reversed polarity; however, a stable palaeomagnetic direction cannot be determined. Fig. 4(a) is an example of a sample for which a deflection away from the origin is observed above 6 mt in the direction perpendicular to the Y-axis (the last axis along which a field was applied during demagnetization treatment at GSJ). Shipboard PAFD treatment revealed that some samples from holes 171C and 172A unambiguously acquired a GRM in the direction perpendicular to the Z-axis (the last axis along which a field was applied during static AFD on the ship) above 6 mt and up to 2 mt (e.g. Fig. 4b). GRM has been shown to exist in anisotropic magnetic material composed of single-domain magnetic grains (Stephenson 198), where the spurious magnetization is produced in the direction perpendicular to the applied alternating field and anisotropy axes. The general characteristics observed for these samples are similar to the strong GRM observed during static three-axis PAFD on greigite (Snowball 1997a; Hu et al. 1998; Sagnotti & Winkler 1999; Stephenson & Snowball 21). Zero-field warming curves of an SIRM acquired at 5 K for sites 171 and 172 show the Verwey transition between 17 and 119 K (Fig. 5). The transition temperature is mostly above 11 K, therefore the composition is close to magnetite without a significant amount of low-temperature oxidation or titanium/aluminium substitution (Kozlowski et al. 1996). Proterozoic low-ti iron oxide deposits are widely exposed in New York and New Jersey, including the Adirondack Mountains (e.g. Foose & McLelland 1995), from where the Hudson River originates. The nearly pure magnetite in the studied material might have been transported from that region. The presence of detrital magnetite has also been reported for the Pleistocene continental slope sediments off New Jersey at ODP sites 93 and 94 (Urbat 1996). The absence of a characteristic magnetic transition at 3 34 K in the low-temperature SIRM (Fig. 5) suggests that pyrrhotite does not occur in detectable amounts in these samples (Rochette et al. 199). XRD results for four selected magnetic extracts from representative intervals of high SIRM/K ratios for holes 171C and 172A contain peaks corresponding to magnetite and greigite (Fig. 6). Thermomagnetic results indicate the Curie temperature of magnetite to be at C (Fig. 7). A large decrease in magnetization is also evident at around 25 C (Fig. 7a). Figs 7(b) and (c) C 24 RAS, GJI, 156,

5 February 17, 24 19:1 Geophysical Journal International gji2162 Magnetism of ODP site 171 and 172 sediments 447 (a) 171C-2X1, 44 cm (58.84 mbsf) (b) 172A-9R2, 6 cm (62.56 mbsf) W Up 1. Mo= 9.1x1-3 A m -1 W Up S Div.= 2 x 1-2 A m -1 N 5 Demagnetizing field (mt) S 6 8 N 3 1. Mo= 1.4x1-1 A m Div.= 1.x1-3 A m -1 E Dn 5 Demagnetizing field (mt) E Dn (c) S 171C-2X2, 18 cm (6.98 mbsf) W Up Div.= 2.5 x 1-3 A m -1 N Mo= 5.1x1 A m -1 5 Demagnetizing field (mt) S (d) 172A-9R1, 118 cm (62.18 mbsf) 8 W Up Div.= 5. x 1-4 A m -1 N 1. Mo= 2.x1-2 A m -1 E Dn E Dn 5 Demagnetizing field (mt) Figure 3. Typical examples of vector endpoint diagrams of progressive alternating field demagnetization experiments. Solid (open) circles denote the projection of the vectors on to the horizontal (vertical) plane. (a) Normally magnetized sample from Hole 171C and (b) reversely magnetized sample from Hole 172A with a stable magnetization above 1 mt. Samples with unstable features for (c) Hole 171C and (d) Hole 172A. are thermomagnetic curves measured through several minor cycles to reveal the temperature at which the magnetic minerals change or decompose. Both curves indicate that below 22 C the thermomagnetic curves are reversible, but between about 22 C and 34 C the curves become irreversible. This indicates that a magnetic phase decomposed between 22 C and 34 C. It has been reported that greigite decomposes irreversibly above 2 C (Krs et al. 1992; Reynolds et al. 1994; Roberts 1995; Torii et al. 1996). Thermomagnetic curves for lake sediments containing greigite show a decrease of magnetization during heating above 3 C, which is similar to that observed in this study (Roberts et al. 1996). Dekkers et al. (2) also reported that synthetic greigite starts to alter at 2 C and that it loses half of its magnetization between 243 C and 276 C during heating in air. This thermomagnetic behaviour combined with low temperature magnetic properties and XRD analysis supports the interpretation that greigite is common in the studied sediments in addition to magnetite. eigitebearing samples usually have high SIRM/K ratios between 5 and 8 ka m 1 (Snowball 1991, 1997b; Roberts 1995), but greigitebearing sediments have been found with low SIRM/K values of 5kAm 1. SIRM/K for our samples (see Figs 8 and 9) is above 15 ka m 1 for the intervals at metres below seafloor (mbsf) and 61.8 mbsf in Hole 171C, and at 62. mbsf in Hole 172A. The contribution of paramagnetic minerals to the susceptibility might have decreased the ratio (see high-field slope values in Table 1). Nevertheless, these intervals with elevated SIRM/K values are considered to represent horizons with higher greigite content. 3.2 Hole 171C Palaeomagnetic and rock-magnetic results for Hole 171C are plotted versus depth in Fig. 8. The NRM inclinations indicate that the polarity boundary occurs at 61.3 mbsf. This polarity boundary lies 3 cm below the erosion surface, which is recognized as a sharp lithological boundary between clay (below) and silt/sand C 24 RAS, GJI, 156,

6 February 17, 24 19:1 Geophysical Journal International gji H. Oda and M. Torii (a) 171C-2X2, 125 cm (61.15 mbsf) (b) 171C-13X1, 59 cm ( mbsf) W Up W Up 1. Mo= 1.3x1-1 A m -1 S -3-1 Div.= 1x1 A m Normalized intensity 8 1. N Mo= 8.2x1-3 A m -1 S Normalized intensity 1 2 Demagnetizing field (mt) N E Dn 5 Demagnetizing field (mt) E Dn Div.= 2x1-2 A m -1 Figure 4. Vector endpoint diagrams of progressive alternating field demagnetization suggesting the presence of a gyroremanent magnetization (GRM) above 6 mt. (b) Reproduced from Austine et al. (1998) for sample 171C-13X1, 59 cm ( mbsf), which shows strong acquisition of a GRM from 6 to 2 mt. (a) 171C-2X1, 93 cm (59.33 mbsf) Temperature (K) (b) 171C-2X3, 3 cm (61.43 mbsf) Temperature (K) Magnetization (x1-3 Am 2 kg -1 ) Magnetization (x1-3 Am 2 kg -1 ) dm/dt (x1-4 Am 2 kg -1 K -1 ) dm/dt (x1-5 Am 2 kg -1 K -1 ) (c) 172A-9R1, 123 cm (62.23 mbsf) Magnetization (x1-3 Am 2 kg -1 ) (d) 172A-9R2, 15 cm (62.65 mbsf) Magnetization (x1-3 Am 2 kg -1 ) Temperature (K) Temperature (K) dm/dt (x1-5 Am 2 kg -1 K -1 ) dm/dt (x1-4 Am 2 kg -1 K -1 ) Figure 5. Results of low-temperature magnetic measurements for holes 171C, and 172A. Solid curves (left axis) are the results of zero-field warming of an IRM imparted in a DC field of 2.5 T from 6 K after zero-field cooling. Dotted curves are the derivative of the curves (right axis). layers (above) at 61. mbsf. The NRM intensity increases up-core above 61. mbsf from about.2 A m 1 to around.2 A m 1 and increases again to around.7 A m 1 at 6. mbsf. In contrast to the results from Hole 172A (see Section 3.3 below), the magnetic susceptibility does not change across the polarity boundary in Hole 171C, although higher susceptibilities are observed above 6. mbsf. Magnetic anisotropy parameters (not shown here) indicate a primary sedimentary fabric (T >, K min inc >67 ) throughout the studied interval, which suggests that the sediments have not been disturbed since deposition. Parameters indicative of magnetic mineral concentration (ARM and SIRM) increase above 61. mbsf. SIRM values have a peak at 61.8 mbsf, where there was only a small local maximum for ARM. S.3T is fairly constant at around.98 below 61. mbsf, and gradually decreases to.95 at 58.4 mbsf. This indicates that the contribution of higher coercivity minerals, such as haematite, is relatively small, but that it increases slightly upward. The remarkably low value of S.3T at 6.98 mbsf is 8 cm below the pp3(s) erosional surface. SIRM/K values gradually increase up-core, with a peak value at 61.8 mbsf, which increases suddenly at 61. mbsf to a value of around Am 1. C 24 RAS, GJI, 156,

7 February 17, 24 19:1 Geophysical Journal International gji2162 Magnetism of ODP site 171 and 172 sediments C-2X1, 93 cm (59.33 mbsf) 15 / Counts 1 5 Mh Qz Qz/ Mt/ Mh/ Mh Mt/Mh/ Mt/ Qz Mt/ Mh Qz Qz θ (deg) A-9R2, 15 cm (62.65 mbsf) 15 Counts 1 / 5 Mh Qz/ Mt/ Mh Mh/ Mt/Mh/ Mt/ / Mt/Mh θ (deg) Figure 6. X-ray diffractograms for (a) Sample 171C-2X1, 93 cm and (b) Sample 172A-9R2, 15 cm. Identified peaks are labelled for magnetite (Mt), maghemite (Mh), greigite (), chlorite (), and quartz (Qz). 3.3 Hole 172A Palaeomagnetic and rock-magnetic results for Hole 172A are plotted versus depth in Fig. 9. The NRM intensity suddenly diminishes up-core by one order of magnitude at 62.3 mbsf. The palaeomagnetic inclination is negative below 62.3 mbsf and positive above 62.3 mbsf for shipboard measurements on half-core samples. The inclinations for the discrete samples below 62.4 mbsf are negative; however, the inclinations above 62.3 mbsf are not always positive, but show a general trend toward negative values. The positive inclination above 62.3 mbsf in the half-core data might result from incomplete removal of a positive drilling-induced overprint at the low maximum demagnetization field of 2 mt and from integration of the signal from undisturbed and disturbed outer parts of the half core. Discrete samples do not provide clearly defined primary magnetizations due to the unstable nature of these sediments. Magnetic susceptibility abruptly decreases up-core above 62.3 mbsf. Magnetic anisotropy parameters (not shown here) indicate a primary sedimentary fabric (T >, K min inc >67 ) throughout the studied interval, as was also observed for Hole 171C. Other concentration-dependent magnetic parameters (ARM and SIRM) also consistently decrease above 62.3 mbsf. S.3T is constant at values of around.98 below 62.3 mbsf, but it gradually decreases to.96 at 61. mbsf. SIRM/K values are constant at about 5 ka m 1 below 62.3 mbsf, but increase up-core at 62.3 mbsf forming a broad maximum with a peak value of 15 ka m 1 at 62. mbsf before decreasing to around 6 ka m 1 at 61.1 mbsf. 4 DISCUSSION 4.1 Hysteresis loop analysis A typical magnetic hysteresis loop for the studied sediments is shown in Fig. 1. For the analysis of hysteresis loops, Jackson et al. (199) proposed taking the derivative after subtracting the lower part C 24 RAS, GJI, 156,

8 February 17, 24 19:1 Geophysical Journal International gji H. Oda and M. Torii of the loop from the upper part with the use of the Fourier transform. Instead of using derivatives, analysis using hyperbolic basis functions was proposed by von Dobeneck (1996). This approach uses not only the remanent hysteretic magnetizations (upper loop minus lower loop divided by two; dotted curve in Fig. 1), but also the induced hysteretic magnetizations (mean of upper and lower loops divided by two; dashed curve in Fig. 1). Hyperbolic basis functions are more effective than taking the simple derivative when decomposing the hysteresis curves, especially when overlapping coercivity components are involved. With the use of the analysis program produced by von Dobeneck (1996), basic parameters for the hysteresis loops were also obtained and are listed in Table 1. The program is designed to incorporate drift correction, gridding with the use of second-degree polynomials, and centralization (rotate 18 along (a) Normalized Strong-field Magnetization (b) Normalized Strong-field Magnetization Temperature ( C) 1 171C-2X3, 3 cm 171C-2X3, 3 cm in air.75 T 8 /min in air.75 T 8 /min Temperature ( C) Figure 7. Thermomagnetic curves for magnetic extracts. The vertical axes are normalized to the initial value. (a) Sample 171C-2X3, 3 cm with heating and cooling cycles, (b) subsample from the same stratigraphic level with minor heating/cooling cycles and (c) Sample 172A-9R2, 15 cm with minor cycles. The heating curves are reversible up to 22 C, but are irreversible between 22 C and 34 C. Above 34 C, the heating curve shows the Curie temperature of magnetite ( 58 C). During cooling no features were observed except the Curie temperature of magnetite. (c) 1 Normalized Strong-field Magnetization 172A-9R2, 15 cm in air.75 T 8 /min Temperature ( C) Figure 7. (Continued.) the origin to average) to reduce noise. The most important part of the program is the inversion utilizing the hyperbolic basis functions with a set of characteristic coercive forces. The inversion is designed to find the best combinations of the basis functions with the constraint of simplest solutions to minimize the number of functions. The spectra of the hyperbolic basis functions for samples from holes 171C and 172A are shown in Fig. 11. Three fractions are recognized with low (phase I; 3 55 mt), medium (phase II; mt) and high (phase III; 15 1 mt) coercivities. 4.2 Decomposition of IRM acquisition curves and greigite content Investigations of the coercivity spectrum were also made by the analysis of IRM acquisition curves. Robertson & France (1994) showed that the IRM acquired by natural magnetic assemblages can be expressed by a cumulative log-gaussian function of the magnetizing field. McIntosh et al. (1996) used a subjective fit byeye on the gradient of the IRM acquisition curve versus the log of the applied field. On the other hand, Stockhausen (1998) introduced goodness-of-fit parameters for the analyses. These are the residual sum of squares between the model and the data for: (1) the gradient of the IRM acquisition curve, (2) the gradient of the log of the IRM acquisition curve, or (3) the IRM acquisition curve. Kruiver et al. (21) introduced the standardized acquisition plot as a target residual to be minimized in order to facilitate a better fit at the lowerand higher-coercivity ends in addition to the gradient of log IRM and the IRM acquisition curve. Although both of these methods are robust, the criteria for which residuals should be prioritized depend on the operator. Heslop et al. (22) proposed a fully automated method to fit the gradient of the logarithm of the IRM according to the maximum likelihood principle (IRMunmix software). Their method is fairly robust and powerful thanks to the EM algorithm (Dempster et al. 1977). An example of the analysis for representative samples from Leg 174A is shown in Fig. 12. The logarithm of the IRM acquisition curve (Fig. 12a) and the logarithm of the IRM gradient (Fig. 12b) are plotted along with the residuals of the fits. The optimized solutions were checked by eye for the goodness-offit between data and model, both on the plot of log IRM gradient C 24 RAS, GJI, 156,

9 February 17, 24 19:1 Geophysical Journal International gji2162 Magnetism of ODP site 171 and 172 sediments 451 Figure 8. Palaeomagnetic and rock magnetic results for Hole 171C. (a) Palaeomagnetic inclinations for the discrete samples determined by fitting linear regression lines (solid circles) are plotted, together with those of the pass-through measurements (at 2 mt) conducted on the ship (dotted curves). (b) NRM intensity is after demagnetization at 2 mt. Rock magnetic parameters are plotted down-hole for: (c) ARM, (d) SIRM, (e) S.3T, (f) volume magnetic susceptibility (K), and (g) SIRM/K, respectively. The right-hand column is a simplified lithological log with the upper legend indicating grain size variations. IW marks the position of an interstitial water sample. PP3(s) is a Plio-Pleistocene sequence stratigraphic boundary (Austine et al. 1998). The horizontal dashed line is the preliminary polarity boundary estimated from the shipboard results. Figure 9. Palaeomagnetic and rock magnetic results for Hole 172A. Details are the same as in Fig. 8. The horizontal dashed line is the preliminary polarity boundary determined from the shipboard results. C 24 RAS, GJI, 156,

10 February 17, 24 19:1 Geophysical Journal International gji H. Oda and M. Torii where N, K and S are the number of data points, number of parameters and sum of the squared residuals, respectively. When the ratio N/K < 4, the bias adjustment for small sample size is required for the calculation of AIC (Burnham & Anderson 1998), so that: AIC c = N ln(s/n) + 2K + [2K (K + 1)]/(N K 1). (3) Figure 1. A typical example of a magnetic hysteresis loop for sample 171C-2X3, 3 cm (solid curves) after paramagnetic slope correction. The dotted and dashed curves are the remanent hysteretic magnetization (difference between upper and lower loops) and the induced hysteretic magnetization (mean of upper and lower loops), respectively, as defined by von Dobeneck (1996). and IRM acquisition combined with the residuals. In general, two components are enough to produce a reasonable model with small residuals. However, their method fails when three-component models were selected for the case of a non-saturated solution, as predicted by Heslop et al. (22). Thus, we use the saturated option for the analysis described below. In order to determine the optimum number of parameters, we introduce Akaike s information criterion (AIC) instead of the t-test or F-test, as used by Kruiver et al. (21) and Heslop et al. (22), for the optimum number of components. AIC is a measure of distance between the model and observation for any kind of thoughtful model (Akaike 1974). The most likely model is selected as giving minimum AIC. On condition that the standard deviation of the observation error is unknown, AIC is calculated by the approximation with a large number of data points compared with the number of parameters (Burnham & Anderson 1998), so that: AIC = N ln(s/n) + 2K, (2) The model is optimal when AIC c is minimum among the considered models. AIC c was calculated for models with a different number of components, and the results were compared with each other (Table 2). AIC c was calculated on each sample using both models, and the models with two peaks are optimal (minimum AIC c ) in most cases. Although there are some samples for which the best model is a three-component model, we consistently used two-component models (Table 3). Another reason for choosing two-component models is that three-component models are unstable with several local minima of similar likelihood. The peak coercivity of remanence values for the component with lower coercivity of remanence is between 49 and 76 mt, which is considered to be typical of pseudo-single-domain magnetite. The peak coercivity of remanence values for the component with higher coercivity of remanence is between 66 and 98 mt. These values are similar to the measured coercivity of remanence of single domain greigite, which was reported as 6 95 mt for a wide range of greigite-bearing sediments (Roberts 1995; Snowball 1997a). From the investigation of magnetic minerals, as presented above in Section 3, the ratio of high-coercivity magnetizations to the total magnetization in the analysis of IRM acquisition curves is considered to represent the relative amount (in terms of magnetization) of greigite in the sample. The results are listed in Table 3 and are plotted versus depth in Fig. 13. Phases I and II in the hysteresis analyses are also considered to represent magnetite and greigite, respectively. Phase III is assumed to represent haematite, because detrital haematite is more likely to occur than goethite in continental shelf sediments. This is consistent with the fact that IRM acquisition curves also show a small portion of unsaturation. The fraction of phase III is minor, therefore we do not discuss this phase further. 4.3 S.7T Considering the observed coercivity of around mt for the higher-coercivity component, we use S.7T as a rough estimate for the ratio of contribution of the greigite to the total magnetization in the samples with the following equation: S.7T = [ (IRM.7T /SIRM) + 1]/2, (4) Table 1. Hysteresis parameters for representative samples. Sample Depth M s M r H c H cr Slope M r /M s H cr /H c (mbsf) (A m 2 kg 1 ) (A m 2 kg 1 ) (mt) (mt) (A m 2 kg 1 T 1 ) 171C-2X1, cm E 2 1.2E E C 2X2, cm E E E C 2X2, cm E E E C 2X3, 3 5 cm E E E C 2X3, cm E E E C 2X4, cm E E E A 9R1, cm E 2 1.7E E A 9R1, cm E E A 9R1, cm E 2 2.8E E A 9R1, cm E E E A 9R2, cm E E E A 9R2, cm E E E C 24 RAS, GJI, 156,

11 February 17, 24 19:1 Geophysical Journal International gji2162 Magnetism of ODP site 171 and 172 sediments 453 Figure 11. Plot of results of decomposition into hyperbolic basis functions (von Dobeneck 1996) for (a) Hole 171C and (b) Hole 172A. IH and RH represent the induced hysteretic magnetization and remanent hysteretic magnetization, respectively. The histograms denote the percentage contribution of IH or RH to the total magnetization of the samples. In general, three magnetic mineral components are recognized. Components I, II and III correspond to magnetite, greigite and haematite, respectively. See text for further explanation. where SIRM is the IRM measured after applying a pulsed field of 2.5 T along the +Z axis of the sample; IRM.7T is measured after imparting a reversed pulsed field of 7 mt along the Z axis. S.7T values are plotted in Fig. 13. The magnetization of the lowercoercivity (magnetite) and the higher coercivity (greigite) components estimated from IRM decomposition and from S.7T are almost identical. Values of about.65 for S.7T seem to be a good threshold for discriminating between the intervals with higher and lower greigite contents for this study. If S.7T is lower than.35, greigite is abundant in the sample. For holes 171C and 172A, S.7T is lower than.65 between 6.1 and 63.1 mbsf, and above 62.4 mbsf, respectively. The intervals with higher greigite content are marked with grey shading in Fig. 13. This interval is mainly characterized by subdued magnetizations of magnetite and greigite, and increased coercivity for both components. 4.4 Magnetic mineral diagenesis and time of magnetization acquisition At sites 171 and 172, the calcareous nanofossils that mark subzone CN14a (.46.9 Ma) were found at 64.4 mbsf and 65. mbsf, respectively (Fig. 13). At site 172, calcareous nanofossils that belong to zone CN14 (.25.9 Ma) were found at mbsf. The sediments above these horizons with reversed polarity are considered to have been deposited during the Matuyama ron (C1r.1r). Although it is not clear whether the sediments with doubtful polarity just below pp3(s) (Fig. 13; grey horizons) were originally deposited during the Matuyama ron, considering the documentation of negative inclinations after PAFD at 4 mt (Figs 8 and 9) the sediments might have been deposited during the Matuyama ron (Fig. 14a). During early diagenesis, bacterial sulphate reduction might have facilitated diagenetic growth of greigite with reversed polarity (Fig. 14b). Also, reductive diagenesis might have resulted in dissolution of magnetite, as was reported by Urbat (1996) for New Jersey continental slope deposits. From palynological analysis of sediments from ODP sites 172 and 173, McCarthy & Gostlin (2) suggested that unconformity pp3(s) can be correlated from the outer shelf to the upper slope and that it was generated during a glacio-eustatic sea-level lowstand, probably during marine oxygen isotope Stage 12 ( ka). They also suggested that a substantial hiatus is associated with the pp3(s) boundary at Site 172. McHugh & Olson (22) provided a chronological model for Site 173 based on oxygen isotope records C 24 RAS, GJI, 156,

12 February 17, 24 19:1 Geophysical Journal International gji H. Oda and M. Torii (a) (b) residual (x1-3 A m -1 ) residual IRM (A m -1 ) dirm/dlog(h) C-2X3, 3 cm (61.43 mbsf).5.5 Data Model Log applied field (mt) 2. Log applied field (mt) Figure 12. An example of fitting with a mixture of two normal distributions on the gradient of the log of the IRM acquisition curve. (a) IRM acquisition curve plotted with an abscissa of the log of the applied field. Solid circles and solid curves represent measured data and the model, respectively. (b) adient of the log of the IRM acquisition curve for the measured data (solid circles) and the model (solid curves). The resolved components are shown as dotted curves with peak coercivities of 76 and 93 mt, respectively and interpreted pp3(s) as a hiatus separating oxygen isotope stages 12 and 11. The MIS transition represents one of the most severe climate changes of the past half-million years (Howard 1997), and sea level rose by about 16 m at the termination of glacial stage 12 (Thunell et al. 22). Assuming that the sediment below pp3(s) was deposited during the Matuyama ron (C1r.1r), a hiatus of Myr (between.9.78 Ma and Ma) is expected at sites 171 and 172. Peak coercivities of remanence, based on decomposition of IRM curves (Figs 13b and e), for greigite and magnetite are both higher for horizons with higher greigite content. The coercivities of remanence are highest near the base of the doubtful polarity intervals (grey horizons). These horizons also have higher contributions from greigite in terms of the magnetization. Jiang et al. (21) suggested several stages for the formation of greigite by progressive diagenesis after deposition. They inferred that oxidation of pyrite released ferric iron, which was used for the late diagenetic formation of greigite in the vicinity of pyrite. Weaver et al. (22) provided evidence that magnetic iron sulphide minerals, in their case pyrrhotite, can grow during late sedimentary diagenetic reactions as a result of fluid flow events. On the New Jersey continental shelf, horizons with higher coercivities for greigite and magnetite could represent a trapped oxidation front (Fig. 14d) produced by downward percolation of oxic fresh water from the overlying porous sediments. This chemical front may have facilitated maximum production of secondary greigite by oxidation of pyrite during formation of pp3(s). Sediment grain size is coarser at Site 172 (silt) than at Site 171 (clay). The greater porosity of the coarser-grained sediments at Site 172 would have enabled more rapid and deeper penetration of oxic fresh water during formation of pp3(s). In the sediments above the oxidation front and below pp3(s), especially at Site 172, the secondary greigite that formed during progression of the oxidation front might have been partially dissolved, together with early diagenetic greigite, as a result of more oxic conditions which are not favourable for greigite preservation. At Site 172, the lower amount of magnetization carried by magnetite for the upper horizon may indicate that progressive oxidation of magnetite occurred, followed by suboxic conditions around the oxidation front that may have caused the effective removal of ferric iron by oxic fresh water during formation of pp3(s). It is also possible that highly permeable sand layers allow present-day fresh water to percolate via ground water connections to the continental shelf Table 2. AIC c for the IRMunmix analysis with the different number of components. Sample Depth N SSR AIC c (mbsf) C = 1 C = 2 C = 3 C = 1 C = 2 C = 3 K = 3 K = 6 K = 9 171C-2X1, cm E E C-2X2, cm E E E C-2X2, cm E+ 1.84E E C-2X3, 3 5 cm E+ 1.61E E C-2X3, cm E+ 1.71E E C-2X4, cm E E E A-9R1, cm E+ 1.99E E A-9R1, cm E+ 2.13E 1 1.1E A-9R1, cm E E A-9R1, cm E+ 2.38E E A-9R2, cm E+ 1.74E E A-9R2, cm E+ 2.6E E SSR: sum of squared residuals. AIC c : Akaike s information criterion (AIC) with small-number correction (see text for details). N: number of data in IRM gradient curve. C: number of components. K: number of parameters. Bold values for AIC c are the minimum AIC c values. C 24 RAS, GJI, 156,

13 February 17, 24 19:1 Geophysical Journal International gji2162 Magnetism of ODP site 171 and 172 sediments 455 Table 3. Parameters of peaks for the log-normal distributions calculated using IRMunmix (Heslop et al. 22) Two-peak model 1st peak 2nd peak Depth SIRM High-coercivity Sample (mbsf) (A m 1 ) IRM (A m 1 ) B 1/2 DP B 1/2 (mt) IRM (A m 1 ) B 1/2 DP B 1/2 (mt) ratio in M 171C-2X1, cm C-2X2, cm C-2X2, cm C-2X3, 3 5 cm C-2X3, cm C-2X4, cm A-9R1, cm A-9R1, cm A-9R1, cm A-9R1, cm A-9R2, cm A-9R2, cm B 1/2 : peak coercivity. DP: dispersion parameter (both are in log1 scale). Figure 13. Down-hole plots of the results of IRM decomposition and S.7T for holes 171C and 172A. Water depths of the sites are shown in brackets. (a, d) High-coercivity ratio and S.7T versus depth. The high-coercivity ratio (top axis; open triangles) is calculated by fitting model functions on the IRM acquisition curve (Table 2), and S.7T (bottom axis; solid circles) is given by eq. (4). Note the inverted axis for S.7T, because S.7T is unity when the higher-coercivity content is zero. The threshold of.65 is shown as vertical dashed lines for S.7T. (b, e) Solid (open) triangles represent coercivities of remanence calculated by model fitting on IRM acquisition curves for the high (low) coercivity component. (c, f) IRM intensity for the high (low) coercivity component calculated by model fitting on IRM acquisition curves is plotted as solid (open) triangles. The intensity of the IRM for the high (low) coercivity component calculated from S.7T are also plotted as solid (open) circles. Samples from the calcareous nanofossil subzones CN14 (.25.9 Ma) and CN14a (.46.9 Ma) are also shown with arrows in the left-most columns. Palaeomagnetic polarities are shown in vertical columns (black = normal polarity, white = reversed polarity). ey shading indicates the horizons identified as having normal polarity during Leg 174A, but which are suspected to have been deposited during the Matuyama ron. Diagonally striped horizons have no data. Shaded intervals correspond to zones with higher greigite contents. The horizontal thick dashed lines represent the unconformity pp3(s). C 24 RAS, GJI, 156,

14 February 17, 24 19:1 Geophysical Journal International gji H. Oda and M. Torii (a) Matuyama (C1r.1r) (b) Matuyama (C1r.1r) (c) Brunhes (d) Present day low stand deposition Mt Mt early diagenesis Mt Py Mt Py fresh water Mt Mt Fe 3+ Py Py erosion oxidation front Mt Mt dissolution Fe 3+ Py pp3(s) oxidation front (post-) depositional remanence (R) dissolution authigenic growth (R) low-t oxidation later stage greigite (N) Figure 14. Cartoon showing the possible mechanism of diagenesis and remagnetization for the studied sediments. Mt, and Py are magnetite, greigite and pyrite, respectively. (a) Acquisition of (post-) depositional remanence carried by magnetite during the Matuyama ron (C1r.1r). (b) Early diagenesis during the Matuyama ron (C1r.1r) causing authigenic growth of greigite with reversed polarity. Authigenic formation of pyrite and dissolution of magnetite are also depicted. (c) Downward progression of an oxidation front by percolation of oxic fresh water during a major sea-level low stand in the Brunhes ron. At the oxidation front, later stage authigenic greigite is inferred to have formed in the vicinity of pyrite with normal polarity magnetization. A maghemite skin formed on magnetite grains as a result of the low-t oxidation. (d) As the oxidation front progressed downward, early and later diagenetic greigite was partially dissolved. Surface maghemite skins on magnetite grains were also dissolved. (e.g. oen et al. 2). The higher coercivity for the magnetite in and around the zone of uncertain polarity can be interpreted as resulting from enhanced stress at the surface of magnetite grains due to low-temperature oxidation (e.g. Cui et al. 1994; van Velzen & Zijderveld 1995). Snowball & Thompson (199) and Sandgren & Snowball (21) found high concentrations of greigite associated with marine incursions into lacustrine settings. In both cases, the greigite probably formed due to the presence of sulphide-rich porewaters in brackish conditions. While these settings are different from those of the present study, the mechanism for greigite formation could be similar in that restricted sulphidic porewaters were present for limited amounts of time, which could have contributed to arrested pyritization and preservation of greigite. In British Early Pleistocene estuarine clays (Hallam & Maher 1994) the magnetization carried by greigite has reversed polarity in the middle of the clay unit, whereas the coarser-grained upper and lower margins of the unit have normal polarity. Hallam & Maher (1994) interpreted the reversed polarity carried by greigite to be an early diagenetic (syndepositional) signal, which was later altered and overprinted due to oxidation during the Brunhes ron. The remagnetized normal-polarity sediments are characterized by weak magnetizations and low susceptibilities, which is similar to the zone of higher greigite content with weaker magnetization at Site 172. Although the present-day situations are different, the British estuarine clays can be considered as a similar setting to the New Jersey continental shelf sediments, especially at Site 172, which was nearly at the shoreface during formation of pp3(s). 5 CONCLUSIONS Palaeomagnetic and rock magnetic studies were performed on continental shelf sediments from ODP sites 171 and 172. These sediments record a reversed to normal polarity transition, which was tentatively interpreted as the Matuyama Brunhes boundary. The polarity boundary is closely correlated with the sequence boundary pp3(s), which is considered to have formed as an erosional unconformity during glacio-eustatic regression and subsequent transgression of sea level. Concerning the origin of magnetic minerals and remagnetization mechanisms, the following issues have been clarified by our study. (1) Primary magnetite and secondary greigite are the dominant remanence carrying minerals. (2) The sediment in the remagnetized zone with uncertain polarity below pp3(s) might have been deposited during the Matuyama ron (C1r.1r). Any detrital magnetite would have acquired a (post-) depositional remanent magnetization with reversed polarity. Early diagenetic greigite might have been created as a result of sulphate reduction and a chemical remanent magnetization with reversed polarity could have been acquired. Subsequently, during formation of pp3(s), magnetite and greigite might have been dissolved below pp3(s) due to oxidation and effective removal of ferric iron by fresh water that penetrated downward from pp3(s). Additional greigite might have formed below pp3(s), especially at the base of the zone with uncertain polarity, which might have resulted from greigite formation via oxidation of pyrite. At the base of the zone with uncertain polarity, the coercivity of magnetite probably increased as a result of increased stress via cracking of the surface of the magnetite grains. (3) The parameter S.7T (see eq. 4) is a useful indicator of greigite concentration. S.7T values lower than.65, or magnetizations higher than.35 determined by the IRMunmix software, may be used as a threshold for a higher content of greigite. (4) Magnetostratigraphic studies of continental shelf deposits need to be conducted with special care because glacio-eustatically controlled remagnetization might affect the palaeomagnetic record of such sediments. ACKNOWLEDGMENTS Samples were provided through the Ocean Drilling Programme. The authors are indebted to T. Yamazaki and M. Dekkers for comments, and A. Roberts and I. Snowball for critical reviews of the manuscript. The authors are grateful to N. Ishikawa for the use of his thermomagnetic balance and to T. von Dobeneck for providing the C 24 RAS, GJI, 156,