Magnetic remanences

The natural remanent magnetization (NRM) was measured on u-channel samples using a 2-G Enterprises Model 760R cryogenic magnetometer at 1 cm intervals. However, due to the finite spatial resolution of the pick-up coils (half-width of the response function = 4.5 cm) sorne smoothing occurs (Weeks et al. 1993). The data from the upper and lower 4 cm of each u-channel were thus excluded because they are affected by an edge effect (i.e., the integration of ‘no sediment’ at the end and beginning of each u-channel). In order to isolate the characteristic remanent magnetization (ChRM), the NRM was measured and progressively demagnetized using stepwise peak altemating fields (AF) of 10-60 mT in 5 mT steps, 70 and 80 mT. Magnetic declination and inclination of the ChRM (labelled ChRM D and ChRM l, respectively) were calculated at 1 cm intervals using a least-square line-fitting procedure (Kirschvink 1980). Since the cores were not azimuthally oriented, the ChRM D profiles are relative. The ChRM D profiles were also corrected by rotating the mean declination of the entire core to North. The precision of the best-fit procedure was estimated by the maximum angular deviation (MAD; Kirschvink 1980).

MAD values below 5° were recently suggested for high quality marine u-channel data. Finally, the median destructive field (MDFNRM) of the NRM (the value of the peak AF necessary to reduce the NRM intensity to half of its initial value) was calculated using the software developed by Mazaud (2005). The MDFNRM is a coercivity and grain size dependent parameter useful to estimate magnetic mineralogy. An anhysteretic remanent magnetization (ARM) was imparted using a 100 mT AF fi eld with a 50 /lT direct CUITent (DC) biasing field. The ARM was measured and demagnetized at peak fields of 20, 25, 30, 35, 40, 45, 50, 55, 60 mT and 10,20, 25, 30, 35, 40, 50, 60, 80 rnT for cores 05 and 803, respectively. The ARM was also expressed as anhysteretic susceptibility (kARM) by norrnalizing the ARM with the biasing field. If rnagnetic rnineralogy is principally controlled by pseudo-single dornain magnetite, kARMIkLF varies inversely with magnetic grain size (e.g., King et al. 1982). Two isotherrnal remanent magnetizations (IRMs) were imparted to the z axis of the u-channels with a De pulse field of 0.3 T (IRMo.3T) and 0.95 T (SIRM) using a 2-G Enterprise pulse rnagnetizer. Each IRM was then demagnetized and measured at peak AF of 0, 10, 20, 25, 30, 35, 40, 50, 60, 70, 80 mT and 0, 10, 20, 25, 30, 35, 40, 50, 60, 80 mT for cores 05 and 803 , respectively. These IRMs were used to construct a pseudo S-ratio by dividing the IRMo.3T by the SIRM. Pseudo S-ratio values close to 1 are indicative of low coercivity mineraIs su ch as magnetite, whereas lower values are indicative of the presence of high coercivity mineraIs such as hematite. Finally, magnetite grain size was estimated by the kLF versus SIRM diagram (Thompson and Oldfield 1986). Even though the Iater diagram is based on the use of pure magnetite for calibration, it is often used in paieomagnetism for the estimation of grain size (e.g., Gogorza et al. 2004; Sager and Hall 1990).

Radiocarbon dating

The chronology ofboth cores were deterrnined using accelerator mass spectrometry (AMS) 14C measurements on six and four calcareous pelecypod shells for cores 05 and 803, respectively (Table 2). All radio carbon ages were calculated using Libby’s half-life (5568 yr) and corrected for natural and sputtering fractionation (8 13C= -25 %0 VPDB; Stuiver and Polach 1977). The conventional 14C ages were calibrated using the on-line CALIB 5.0.2 software (Stuiver et al. 2005) and the Hughen et al. (2004) marine datas et. A regional reservoir correction (L1R) of 400 years was applied to core 803 based on the average L1R value derived from 5 dates realized on pelecypod shells collected prior to nuclear testing from the Amundsen Gulf (McNeely et al. 2006). In contrast, no regional reservoir correction was applied to core 05, as that core was raised from the slope in interrnediate Atlantic-derived waters. aDepth corrected for the missing sediment (see text for details). bAIl ages were measured by the AMS method using Libby’s half-life (5568 yr) and corrected for natural and sputtering fractionation (Ô l3 C = -25%0 VPDB).

The statistical uncertainty of the age determination is given as one standard deviation (Stuiver and Polach 1977). cCalibrated using the on-line CALlB 5.0.2 software (Stuiver et al. 2005) using the Hughen et al. (2004) marine datas et. Regional reservoir correction (~R) for cores 05 and 803 is 0 and 400 yr, respectively. The first and last ages, in parentheses, represent the 2-() cal age range. The two age models were constructed using a linear fit between the available calibrated ages on a composite depth scale corrected for missing sediments due to the piston coring pro cess (Fig. 1.2). For core 05, the comparison of the attenuated gamma ray count measurements from the piston and trigger weight cores reveals that about 75 cm of sediment was lost during piston coring (Fig. 1.3a). An analogous estimate was made for core 803 using both the magnetic susceptibility and wet bulk density profiles, revealing that about 58 cm of sediments was missing in the piston core (Fig. 1.3b). For core 05, the age model was only constructed for Unit 1 (see below), as the sedimentation rates in the sediments of unit 2 are currently unknown and could be much higher as this unit was deposited in a glacial/deglacial environmental setting (see below).

Magnetic properties kLF values of core 05 range from 200 to 250 (x 10-6 SI) within lithologic unit 1 (Fig. l.4a). In lithological unit 2, kLF values reveal more variations and range from 80 to 300 (x 10-6 SI), consistent with mineralogical and grain size variations observed during visual description of the core. Aside from the first 800 cm, where NRM is relatively constant (about 0.03 Am-l, after the 30 mT demagnetization step), a long-term downcore decreasing trend is observed. The same long-term downcore decreasing trend is observed in the ARM and IRM profiles, consistent with a diminution of the ferrimagnetic concentration downcore (Fig. l.4a). Vector end-point orthogonal projection diagrams (Zijderveld 1967) reveal that the measured NRM is characterized by two magnetic components (Fig. 1.5a). A soft magnetic component (viscous magnetization) with a coercivity spectrum in the 0-20 mT AF range and a stable, well-defined, magnetic component in the 25-80 mT AF range. The characteristic component magnetization (ChRM) was isolated using ten demagnetization steps between 25 mT and 80 mT. The component inclinations (ChRM 1) fluctuate around the expected inclination (IGAD) calculated at the sampling site using a geocentric axial dipole (GAD) model (Fig. 1.6a). Aside from lithologic unit 2, where the ChRM is poorly defined, downcore MAD values are lower than 2°, indicative of very well-defined ChRM.

MDFNRM values fluctuate between 13.7 and 56 mT, with an average of29.40 ± 8.03 mT (Fig. 1.7a). Such an average and range are typical ofmagnetite. The kLF versus SIRM diagram is compatible with the presence of magnetite in the 1-16 f.l111 grain size range (Fig. 1.8a). Downcore variations of the grain size dependent ratio kARM/kLF depict a decreasing trend compatible with an increasing magnetite grain size (Fig. 1.7a). A submicron magnetite grain size from 319 cm to 1215 cm, as welI as multi domain (MD) magnetite grain size (> 16 f.l111) below 1605 cm is observed (Fig. 1.8a). Coarse MD magnetite grains are inefficient carriers of a stable detrital remanent magnetization (DRM) (e.g., Dunlop and Ozdemir 1997). AlI the studied samples faU in the pseudo-single domain reglOn (PSD) for magnetite (Fig. 1.9a; Day et al. 1977). Aside from lithologic unit 2, pseudo S-ratio values are close to l, indicative of low coercivity minerais su ch as magnetite (Fig. l. 7a). In summary, the magnetic mineralogy and grain size of lithologic unit 1 is compatible with the presence of PSD magnetite, whereas the magnetic mineralogy and grain size of lithologic unit 2 is compatible with coarse MD magnetite. The hysteresis loops indicate saturation fields weU below 200 mT and coercivities typical of magnetite (Fig. 1.9b). In addition, allloops are characterized by a paramagnetic component. However, an almost linear relationship between the measured magnetic moment M and the applied magnetic field H with no hysteresis loop is observed in sorne intervals of lithological unit 2 indicating that the magnetic behaviour of such samples is dominated by the paramagnetic fraction. For this reason, we will now only focus on the upper unit (postglacial sediments) of core 05 (unit 1).