Paleomagnetic analysis and radiocarbon dating of an expanded Holocene deep-sea sediment sequence recovered by Integrated Ocean Drilling Program (IODP) Expedition 303 from Labrador Sea Site U1305 (Lat.: 57°28.5 N, Long.: 48°31.8 W, water depth 3459 m) provides insights into mechanisms that drive both paleomagnetic secular variation (PSV) and magnetization acquisition in deep-sea sediments. Seventeen radiocarbon dates on planktonic foraminifera define postglacial (ca. 8 ka) sedimentation rates as ranging from 35 to > 90 cm/kyr. Alternating field (AF) demagnetization of u-channel samples show that these homogeneous sediments preserve a strong, stable, and consistently well-defined component magnetization. Normalized remanence records pass reliability criteria for relative paleointensity (RPI) estimates. Assuming that the age of magnetization is most accurately defined by well dated PSV records with the highest sedimentation rates, allows us to estimate and correct for temporal offsets at Site U1305 interpreted to result from postdepositional remanence acquisition at a depth of ∼20 cm. Comparisons indicate that the northern North Atlantic PSV and RPI records are more consistent with European than North American records, and the evolution of virtual geomagnetic poles (VGP) are temporally and longitudinally similar to global reconstructions, though with much larger latitudinal variation. The largest deviations from a geocentric axial dipole (GAD) are observed during times of the highest intensities, in contrast to the usual assumption. These observations are consistent with the idea that PSV in the North Atlantic and elsewhere during the Holocene results from temporal oscillations of high-latitude flux concentrations at a few recurrent locations.
 Paleomagnetic secular variation (PSV) data derived from geological and archeological materials have extended our knowledge of geomagnetic field variability over timescales of 103 years and longer [e.g., Merrill et al., 1996]. Continuous spherical harmonic models (CSHM) using PSV data provide a useful tool for extending historical assessments of the space/time structure of the geomagnetic field [Korte and Constable, 2003, 2005, 2011; Korte et al., 2005, 2009, 2011]. Although historical observations indicate persistent high-latitude regions of concentrated nonaxisymmetric flux [e.g., Bloxham and Gubbins, 1985; Jackson et al., 2000] and time-averaged field studies suggest that these features may be truly long lived [e.g., Gubbins and Kelly; 1993; Johnson and Constable, 1995], analyses of CSHM suggest that these features are not truly persistent, but drift to and from preferred locations [e.g., Bloxham, 2002; Dumberry and Finley, 2006; Korte and Holme, 2010; Amit et al., 2010, 2011]. Given enough well dated high-resolution PSV records, CSHM will eventually be able to accurately resolve the evolution of nonaxisymmetric features. At present, however, the spatial and temporal resolution of CSHM are limited by the distribution and quality, including chronology, of available PSV data. As a result, our understanding of these significant geodynamo structures and their impact on PSV, even in the Holocene, remains poorly constrained.
 Paleomagnetic records preserved in deep-sea sediments have revolutionized our understanding of the longer-term variability of the geomagnetic field [Opdyke and Channell, 1996]. Variations in intensity, polarity reversals, and magnetic excursions have all been recovered from these archives [e.g., Guyodo and Valet, 1996; Channell and Lehman, 1997; Lund et al., 2006; Laj and Channell, 2007]. The recent focus on relative paleointensity is transforming our understanding of paleogeomagnetic change [Tauxe, 1993; Valet, 2003; Tauxe and Yamazaki, 2007; Ziegler et al., 2011], while providing an important new technique for high-resolution global stratigraphy [e.g., Mazaud et al., 1994, 2002; Guyodo and Valet, 1996, 1999; Stoner et al., 1995a, 2000, 2002; Channell et al., 2000, 2009; Laj et al., 2000, 2004]. In contrast, directions preserved in deep-sea sediment paleomagnetic records [e.g., Lund and Keigwin, 1994; Lund et al., 2005] are rarely considered in detail outside of reversals [e.g., Channell and Lehman, 1997; Mazaud and Channell, 1999; Mazaud et al., 2009] and excursions [e.g., Lund et al., 2005; Channell, 2006; Laj et al., 2006]. As a result, and largely owing to the fact that deep-sea sedimentation rates are usually not sufficiently high, locations like the northern North Atlantic that are optimal for constraining the behavior of nonaxisymmetric geomagnetic features [e.g., Bloxham and Gubbins, 1985; Jackson et al., 2000] are rarely utilized as archives of Holocene PSV.
 During Integrated Ocean Drilling Program (IODP) Expedition 303, drilling at Site U1305 (Lat.: 57°28.5 N, Long.: 48°31.8 W, water depth 3459 m; Figure 1 and Table 1) recovered an expanded version of a well-studied sediment sequence off southern Greenland in the Labrador Sea region of the northern North Atlantic Ocean [Expedition 303 Scientists, 2006]. Shipboard paleomagnetic measurements indicated that sediments from Site U1305 preserve directional variability consistent with Holocene PSV. Facilitated by the recent development of ultra-high resolution paleogeomagnetic records from surrounding regions [e.g., St-Onge et al., 2003; Stoner et al., 2007; Genevey et al., 2008], the recovery of undisturbed Holocene sediments at Site U1305 provides an unexpected opportunity to explore the full-vector Holocene paleomagnetic record of the northern North Atlantic.
Table 1. Sites Discussed in This Study
Water Depth (mbsl)
IODP Site U1305
CSS Hudson Piston Core
ODP Site 646
IMAGES Calypso Core
IMAGES Calypso Core
IMAGES Calypso Core
IMAGES Calypso Core
IMAGES Calypso Core
2. Regional Setting and Background
 Eirik Ridge, sometimes referred to as Eirik Drift, has been explored by the Ocean Drilling Program (ODP) Leg 105 [Site 646; Srivastava et al., 1987] and the Integrated Ocean Drilling Program (IODP) Exp. 303 [Sites U1305, U1306, and U1307; Channell et al., 2006] and several other coring cruises [e.g., Hillaire-Marcel et al., 1991; Turon et al., 1999]. The Pliocene-Quaternary depositional structure is fashioned by sediment redeposition as the Western Boundary Undercurrent (WBUC) sweeps around southern Greenland [Johnson and Schneider, 1969; Srivastava et al., 1987; Hunter et al., 2007]. IODP Site U1305 is located on the lee side of the southwestern extremity of the Eirik Ridge (Figure 1) where sediments accumulate rapidly during interglacials in response to a deepening of the WBUC [Lucotte and Hillaire-Marcel, 1994; Hillaire-Marcel et al., 1994a, Hunter et al., 2007]. Drilled 82 km south of previously lee side occupations (e.g., ODP Site 646; piston cores Hu90-013-013 and MD99-2227) (Figure 1 and Table 1), Site U1305 captures an expanded version of the previously observed lithologic pattern [e.g., Expedition 303 Scientists, 2006; Hillaire-Marcel et al., 2011; Mazaud et al., 2012]. Lithogenic sediments arrive at these sites from the erosive activity of the Greenland, Iceland, and Laurentide ice sheets and subsequent redistribution by the WBUC, with rapidly deposited detrital layers commonly observed during times of ice sheet instability [Hillaire-Marcel et al., 1994a; Stoner et al., 1994, 1995b, 1996; Evans et al., 2007; Hunter et al., 2007; Carlson et al., 2008]. Biological productivity is highest during interglacials and the biogenic carbonate flux increases through the Holocene [Hillaire-Marcel et al., 1994a, 1994b]. The lithology reflects these environmental variations and ranges from silty clay with sand and little biogenic material during glacial and deglacial times, to homogenous nannofossil ooze with silty clay and common bioturbation during interglaciations [see, Expedition 303 Scientists, 2006]. As a result, sediments from the lee side of the Eirik Ridge have been used to reconstruct paleoceanographic [e.g., Hillaire-Marcel et al., 1994a, 1994b, 2001, 2011; Hillaire-Marcel and Bilodeau, 2000; Fagel et al., 1997, 2002, 2004; Fagel and Hillaire-Marcel, 2006; Kleiven et al., 2008; Winsor et al., 2012], ice sheet [e.g., Hillaire-Marcel et al., 1994a; Stoner et al., 1994, 1995b, 1996; Evans et al., 2007; Carlson et al., 2008; de Vernal and Hillaire-Marcel, 2008; Kleiven et al., 2008; Colville et al., 2011], and paleomagnetic [e.g., Stoner et al., 1995a, 1998; Evans et al., 2007; Mazaud et al., 2012] evolution over a range of time scales. The primary objective for drilling Site U1305 was to use IODP's advance piston coring (APC) technology to extend these high-quality records through the recovery of a complete uppermost Pliocene to Quaternary sediment section [Expedition 303 Scientists, 2006].
 The undisturbed recovery of an expanded postglacial Holocene section at Site U1305 [Expedition 303 Scientists, 2006, Site U1305] presented a significant and unexpected opportunity. Shipboard data and prior studies [e.g., Hillaire-Marcel et al., 1994a, 1994b; Stoner et al., 1994, 1995b, 1996; Fagel et al., 2004; Evans et al., 2007] suggest that the rapid accumulation of homogenous postglacial sediments would be ideal for paleomagnetic reconstructions. As a result of piston coring induced soft sediment deformation [e.g., Turon et al., 1999], earlier studies did not explicitly interpret the Holocene part of the record [Stoner et al., 1995a; Evans et al., 2007]. At Site U1305, a well-recovered Holocene is implied by preservation of horizontal structures including the Fe redox boundary at ∼0.20 m composite depth (mcd) and the reproduction of shipboard physical and paleomagnetic properties from two overlapping Holes, Hole U1305B and Hole U1305C (see supporting information Figure fs011) [Expedition 303 Scientists, 2006, Site U1305]. In contrast, data collected as part of Stoner et al. [1995a] (Hu90-013-013) and Evans et al.,  (MD99-2227) studies show that neither piston core yielded correlative nor clearly interpretable Holocene PSV records (see supporting information Figure fs01).
 U-channel samples from an optimal off-splice section were collected from the working halves of cores U1305A-1H, U1305C-1H, and U1305C-2H, measured and returned prior to the Expedition 303 Sampling Party, whereupon they were subsampled for other analyses. The focus of this study is the paleomagnetic record from 1305C-1H that captures an almost complete postglacial Holocene (ca. 8.2 ka) record from 0.05 to 5.1 mcd [Expedition 303 Scientists, 2006].
3. Magnetic Methods
 Magnetic remanence measurements before and after demagnetizations were performed using a 2G Enterprises superconducting rock magnetometer with inline alternating field (AF) coils designed to measure u-channel samples at the University of Florida. The natural remanent magnetization (NRM) of the u-channel samples were measured at 1 cm spacing before demagnetization, and after each of 12 AF demagnetization steps with peak fields from 10 to 100 mT (Figure 2). Though measured at a 1 cm interval, the response function of the magnetometer integrates ∼4.5 cm (width at half-height) of stratigraphic depth [Weeks et al., 1993; Guyodo et al., 2002]. For each u-channel, anhysteretic remanent magnetization (ARM) was acquired in a peak AF of 100 mT and a 50 µT DC biasing field. ARM was also measured at a 1 cm interval before and after AF demagnetization at the same steps used for the NRM, up to peak AF of 60 mT (Figure 2). Isothermal remanent magnetization (IRM) was then acquired in a DC field of 0.3T and was measured and demagnetized using the same procedure as for the ARM (Figure 2).
 Volumetric magnetic susceptibility (κ) (Figure 2) was measured on u-channel samples at 1 cm spacing using a track built at the University of Florida with a Saffire Instruments SI2B magnetic susceptibility loop system that averages over 3 cm [Thomas et al., 2003]. Normalizing ARM by the strength of the DC biasing field allows it to be expressed as an anhysteretic susceptibility (κARM) and is used in concert with κ to produce a magnetic grain-size sensitive ratio κARM/κ [Banerjee et al., 1981; King et al., 1982; Stoner et al., 1996] (Figures 2 and 3). Hysteresis parameters (Mrs, Ms, Hc, and Hcr) were acquired in a saturating field of 1000 mT on a Princeton Measurements Vibrating Sample Magnetometer at the Pacific Northwest Paleomagnetic Laboratory at Western Washington University.
4.1. Natural and Laboratory Remanent Magnetizations
 Figure 2 illustrates the transition at ∼4.5 mcd from the late deglacial, where large fluxes of terrigenous sediment are episodically delivered to the sea floor, to the postglacial Holocene where minimum ice conditions, increased productivity, and the general stability of the WBUC result in the consistent delivery of homogenous nannofossil ooze with silty clay [Expedition 303 Scientists, 2006]. The characteristic remanent magnetization (ChRM) directions (Figure 2) from u-channel NRM measurements were calculated by principal component analysis [Kirschvink, 1980] using 10 consecutive AF demagnetization steps with peak fields between 20 and 80 mT. Maximum angular deviation (MAD) associated with the ChRM calculation are less than 1.5° for the postglacial Holocene (Figure 2), indicating a consistently well-defined magnetization. ChRM inclinations vary around the geocentric axial dipole (GAD) inclination (∼72°) for the site latitude (Figure 2). Because independent azimuthal control was not available for the uppermost cores, relative ChRM declinations were rotated to an entire core mean of zero (Figure 2). This assumption is reasonable and necessary, although only ∼8000 years of geomagnetic behavior is averaged [e.g., Merrill and McFadden, 2003]. ChRM directions from the u-channel samples are consistent with shipboard derived paleomagnetic directions replicated from two holes at Site U1305 [Expedition 303 Scientists, 2006] (see supporting information Figure fs01).
 Hysteresis data derived from Site U1305 sediments are consistent with prior Eirik Ridge studies [Stoner et al., 1995a; Evans et al., 2007; Kleiven et al., 2008] and show that postglacial Holocene sediments fall within the pseudo-single domain (PSD) field on a Day et al.  plot (Figure 3). Except for the oxidized upper ∼20 cm, all samples fall on a single-domain to multidomain magnetite mixing line [DunlopandCarter-Stiglitz, 2006]. On a Banerjee et al.  plot (κARM versus κ, Figure 3), postglacial Holocene sediments at Site U1305 display a linear trend indicating magnetic uniformity. Both hysteresis and κARM/κ data indicate that postglacial Holocene sediments at Site U1305 are similar to, but magnetically coarser (Figure 3) than those observed from prior lee side occupations 82 km further to the north (e.g., ODP Site 646; cores Hu90-013-013 and MD99-2227) [Stoner et al., 1995a; Evans et al., 2007] (Figure 1). The strong, low-coercivity magnetization at Site U1305 is consistent with magnetite being the primary remanence carrier, as found in other Erik Ridge studies [e.g., Stoner et al., 1995a, 1995b; Evans et al., 2007; Kawamura et al., 2012; Mazaud et al., 2012]. While the higher accumulation rates and slightly coarser magnetic grain sizes at the more proximal Site U1305 location further implicate the WBUC in the deposition of these sediments.
 The postglacial Holocene sediments above 4.5 mcd at Site U1305 fall within the criteria of King et al.  and Tauxe  for sediments likely to provide reliable relative paleointensity (RPI) estimates (Figures 2 and 3). NRM, ARM, IRM, and κ all vary by less than a factor of two (Figure 2), well within the order of magnitude advocated by Tauxe  for variations in magnetic mineral concentrations. Magnetic grain size, indicated by κARM/κ and ARM/IRM at Site U1305 (Figures 2 and 3), and hysteresis ratios from this and previous regional studies [Stoner et al., 1995b, 1996; Evans et al., 2007; Kleiven et al., 2008], show only minor down-core variations. NRM normalized by ARM and IRM for all demagnetization steps between 10 and 60 mT with mean ratios and slope values of the best-fit lines calculated using the UPmag software [Xuan and Channell, 2009] are shown in Figure 4. Similar to many North Atlantic normalized remanence records [e.g., Channell et al., 1997; Stoner et al., 2000; Evans et al., 2007], NRM/ARM increases with increasing peak AF, while NRM/IRM decreases (Figure 4), indicating that neither is a perfect coercivity-match to the NRM, although the means and slopes of the two RPI estimates are similar (Figure 4). To check if the normalized remanence records are independent from and therefore not contaminated by their normalizers, we calculate the squared wavelet coherence between the mean NRM/ARM and ARM, and between the mean NRM/IRM and IRM. The calculation is performed both before and after the records are placed on an age model. Squared wavelet coherence measures the similarity of the cross-wavelet transforms of two time series in the time-frequency domain. Compared with traditional coherence analysis methods [e.g., Tauxe and Wu, 1990; Tauxe, 1993], which identify frequency bands within which the two time series are covarying, the squared wavelet coherence is used to identify both the frequency bands and time intervals within which the two time series are covarying significantly [e.g., Torrence and Webster, 1999; Xuan and Channell, 2008]. Little significant coherence in depth or time was observed between the normalized remanence and either normalizer (see supporting information, Figure fs02) suggesting that these sediments provide reliable RPI estimates. The RPI proxies are similar (Figure 4), although NRM/IRM implies higher intensities for the upper 1.5 mcd and a rapid intensity drop over the upper 0.25 mcd, not observed in the NRM/ARM records. Because the variance within the normalized remanence values are similar and coercivity of NRM falls between that of ARM and IRM, we average the means and the slopes of NRM/ARM and NRM/IRM to estimate postglacial relative paleointensity and its uncertainty at Site U1305 (Figure 4).
4.2. Radiocarbon Chronology for Site U1305
 The most recent rapidly deposited detrital layer, observed between 5.11 and 5.22 mcd (U1305C-1H-4, 56–67 cm) [Expedition 303 Scientists, 2006] marks the base of the studied section. This visually identifiable layer is unique when compared with other rapidly deposited detrital layers [e.g., Stoner et al., 1996; Evans et al., 2007; Kleiven et al., 2008] as it is associated with fine magnetic grain sizes and low concentrations intercalated within a magnetically coarser, higher concentration interval (Figure 2). Kleiven et al.  dated this layer in nearby core MD03–2665 (Figure 1 and Table 1) at ∼ 8.3 Cal kyrs B.P. and attributed it to the outburst flood of glacial Lake Agassiz [Barber et al., 1999; Lajeunesse and St-Onge, 2008]. The discharge of fresh water associated with this event is thought to have been responsible for the “8.2” climate event [Alley et al., 1997; Kleiven et al., 2008] dated at 8236 ± 47 cal yrs BP in the Greenland NGRIP Ice Core [Rasmussen et al., 2006]. Sedimentation rates and the age model above this layer are constrained by 17 calibrated radiocarbon dates derived from mixed planktonic foraminifera picked from the >125 μm fraction (Figure 5 and Table 2). Radiocarbon analyses were performed at the UC Irvine Keck AMS facility. Radiocarbon dates were converted to calendar ages using CALIB 6.02 INTCAL09 [Reimer et al., 2009] with a standard ∼400 yr marine reservoir correction (Table 2). These dates describe sedimentation rates that vary from ∼35 to 90 cm/kyr for the upper 4.5 mcd with higher sedimentation rates below. A few dates fall off this general trend (Figure 5 and Table 2). An age reversal is observed for the oldest two dates, when considered along with the observation of the “8.2” event, implies exceptionally rapid sedimentation (>2 m/kyr) below ∼4.5 mcd. A 2-σ age overlap is observed between dates at 2.85 and 3.10 mcd implying rapid sediment accumulation in this interval, with much lower rates above (Figure 5). At present, however, there is little lithologic evidence to support this and therefore our preferred age model is constrained by 8.2 ka event, prescribed with an age of 8250 cal yr BP at 5.15 mcd, with a linear interpolated fit to all but two of the dates above (Figure 5 and Table 2).
 Three Marion Dufresne II (MD) cores obtained from ultra-high resolution (>100 cm/kyr) continental margin sediments (Figure 1 and Table 1) help define the Holocene paleomagnetic record of the Labrador Sea region of the northern North Atlantic Ocean. Core MD99-2269 collected from Húnaflói on the north Iceland shelf and core MD99-2322 collected from the deepest part of the Kangerlussuaq Trough on the southeast Greenland shelf [Stoner et al., 2007] are located northeast of Site U1305 (Figure 1). The declination and inclination records from both cores are combined into a single composite PSV record based on correlation between the two sites [Stoner et al., 2007] (Figure 6). The Greenland/Iceland PSV composite averages the PSV records from these two MD cores during periods of overlap, while using the best available record to fill gaps caused by sediment deformation and section breaks. An initial age model for these cores was developed using 25 radiocarbon dates from core MD99-2269 and 19 from core MD99-2322 combined into a single age to depth relationship [Stoner et al., 2007]. To account for radiocarbon reservoir age uncertainties in the Nordic Seas [e.g., Eiriksson et al., 2004], terrestrial ages of seven crypto-tephras and 1 discrete tephra layer (Saksunarvatn) in core MD99-2269 [Kristjansdottir et al., 2007] were used to further adjust the Stoner et al.  age model. This results in a chronology up to 250 yrs younger (mean ∼ 130 yrs, median ∼ 90 yrs) than previously reported [Stoner et al., 2007] (Figure 6b). The Greenland/Iceland PSV composite has a nominal 5 yr resolution, although additional smoothing due to stacking, sedimentation rate changes, and the response function of the u-channel magnetometer are not taken into account. High mean sedimentation rates of ∼200 cm/kyr in both cores and minimal mixing depth of 5–15 cm (based on cryptotephra distributions, G. Kristjansdottir, personal communication, 2008 and A. Jennings, personal communication, 2010) implies that the age of these sediments should provide a reasonable estimate for the age of the magnetization. This is supported by agreement between the Greenland/Iceland PSV composite and CSHM (ARCH3k_cst.1) predictions [Korte et al., 2009] (Figure 6). It should be noted that there is less agreement between Greenland/Iceland PSV composite and model predictions made with less constrained data sets [Korte and Constable, 2011; Korte et al., 2011]. ARCH3k_cst.1 is derived exclusively from archeomagnetic data that passes quality criteria [Donadini et al., 2009], suggesting that data quality (including chronology) may be more important than data completeness for the accuracy of CSHM predictions in reasonably well-constrained locations.
 Another existing ultra-high resolution PSV record to the southwest of Site U1305 is core MD99-2220 from the St Lawrence estuary (Figure 1) [St-Onge et al., 2003]. The fidelity of this paleomagnetic record is supported by recent PSV and RPI stacks from the same region [Barletta et al., 2010]. The age model for core MD99-2220 is based on six calibrated AMS radiocarbon dates augmented by an additional eight dates transferred through lithologic correlation from nearby core MD99-2221 [St-Ongeetal., 2003]. The high mean sedimentation rate of ∼150 cm/kyr and the minimal mixed layer thickness, estimated at 5–10 cm based on 210Pb analysis [St-Onge et al., 2003], implies that the age of the sediment is a close estimate for the age of the magnetization. Assuming that the control of the geomagnetic field on the production of cosmogenic nuclides [e.g., Elsasser et al., 1956] holds even at centennial timescales, St-Onge et al.  took the additional step of correlating the RPI from core MD99-2220 to the well-dated record of radiocarbon production [Stuiver et al., 1998]. Age adjustments of no more than 200 yrs, well within chronological uncertainty, were required to maximize this correlation. This supports the contention that the geomagnetic shielding model for cosmogenic nuclide production extends to submillennial time scales [St-Onge et al., 2003] and that the age of magnetization in core MD99-2220 is close to the age of the sediment.
 Using the radiocarbon-based age model in Figure 5, inclination and declination patterns for Site U1305 are shown in Figure 7. Comparisons, with ultra-high resolution PSV records from the Greenland/Iceland composite and the St. Lawrence Estuary, and the ARCH3k_cst.1 CSHM predictions, show that Site U1305 preserves similar features, but at consistently older ages. Using a limited number of tie points (6) and linear interpolation, the ages of PSV features can be adjusted using the Analyseries program [Paillard et al., 1996] to obtain a higher degree of correlation (Figure 8). The tie points used to adjust Site U1305 to the Greenland/Iceland composite are shown in Figures 7 and 8. Correlation coefficients (r) between Site U1305 and the Greenland/Iceland composite record obtained after tuning (Figure 8) are high (r = 0.75 for declination and r = 0.343 for inclination), while those to core MD99-2220 are lower (r = 0.505 for declination and r =0.33 for inclination). Higher correlation coefficients could be obtained using more tie points. Lower correlation coefficients are observed to either PSV record with Site U1305 on its radiocarbon age model (to the Greenland/Iceland composite r =0.616 for declination and r = −0.092 for inclination; to core MD99-2220 r = −0.12 for declination and r = 0.134 for inclination) as in Figure 7.
 Comparisons between the Site U1305 RPI record on the tuned chronology with the European archeomagnetic intensity compilation [Genevey et al., 2008], intensity predictions for the northern North Atlantic from ARCH3k_cst.1 CSHM [Korte et al., 2009], and St. Lawrence Estuary core MD99-2220 RPI [St-Onge et al., 2003] are shown in Figure 9. When compared with the European archeomagnetic intensity record, again correlation coefficients are lower (r = 0.64) with Site U1305 RPI on a radiocarbon-based age model and higher (r = 0.77) after PSV tuning as in Figures 8 and 9. Similarities with the ARCH3k_cst.1 intensity predictions for Site U1305 suggest that differences between European and North Atlantic records reflect regional geomagnetic differences. Comparisons to St. Lawrence Estuary RPI record show that it is less similar (Figure 9) with lower correlation coefficients (r = 0.61). Regional studies suggest that these differences grow as one moves west over North American [e.g., King et al., 1983; Lund and Banerjee, 1985; Lund and Schwartz, 1999; Brachfeld and Banerjee, 2000; St-Onge et al., 2003; Genevey et al., 2008; Barletta et al., 2010].
4.4. Age of the Magnetization
 It is generally assumed that sediments acquire their remanent magnetization not at deposition, but at depth through a mechanism termed post-depositional remanent magnetization (pDRM) [Irving and Major, 1964; deMenocal et al., 1990; Channell and Guyodo, 2004; Suganuma et al., 2010]. This results in the age of the magnetization being younger than the age of the sediment as determined by the radiocarbon dated planktonic foraminifera. Assuming that the age of magnetization in this region is best defined by robustly dated (as described above) PSV records with the highest sedimentation rates (>200 cm/kyr), where the time lag between magnetization acquisition and sediment deposition is minimized. Allows us to estimate and correct for temporal offsets (from pDRM or any other reason) to derive a reasonable age for the magnetization at Site U1305. The age depth relationship that results from the six tie points (Table 3) is shown in Figure 5 (red dots). The differences between the paleomagnetically tuned age model and the calibrated radiocarbon age model at the tie points is on average −289 ± 77 yrs at the same depth and 20 ± 5 cm at the same age (Figure 5 and Table 3). This ∼20 cm depth offset is consistent with other pDRM estimates [deMenocal et al., 1990; Channell and Guyodo, 2004; Suganuma et al., 2010]. Any unrecognized radiocarbon reservoir ages (ΔR) greater than the standard marine reservoir correction [e.g., Tisnerat-Laborde et al., 2010] would affect these estimates. However, on the timescale of interest here, the subarctic North Atlantic during the postglacial Holocene is thought to reflect global surface ocean [Broecker et al., 1985; Bard, 1988] or slightly negative ΔR [Reimer et al., 2002] values, justifying the use of a standard marine reservoir correction [e.g., Hillaire-Marcel et al., 2001, 2007; Fagel et al., 2004; Carlson et al., 2008; Kleiven et al., 2008].
Table 3. Tie Points
Depth mcd (m)
Radiocarbon Age model (cal yrs BP)
(Green/Ice) (cal yrs BP)
5.1. Paleogeomagnetic Record of the Northern North Atlantic
 Transforming the PSV records (Figure 8) into virtual geomagnetic poles (VGP) by mapping a geocentric dipole that would cause the observed PSV record provides a way to visualize PSV while taking site location into account (Figures 10 and 11). Even though the records are derived from cores collected from locations with different deposition rates and sediment environments, VGP latitudes and longitudes derived from Site U1305 and the Greenland/Iceland composite are similar, supporting the interpretation of a regional geomagnetic origin of the common signal. Even the large amplitude change centered at ∼2500 Cal years BP, where VGP latitudes deviate more than 40° from a geocentric axial dipole (GAD) is reproduced (Figure 10). This large deviation in VGP latitude is a result of the large eastward deviation in declination at this time. It should be noted that VGP latitudes are sensitive to declination and therefore, to the zero-mean declination assumption. Additional replication from archives with azimuthal control should be sought to confirm these observations. VGP latitudes from core MD99-2220 do not show such a large departure from GAD at ∼2500 cal yrs BP, but rather a series of lower latitude VGP intervals every few thousand years (Figure 10). The differences between this North American and the North Atlantic records are even more apparent when comparing VGP longitudes (Figure 10). Core MD99-2220 is separated by ∼1800 km from Site U1305, not much farther than Greenland/Iceland composite sites (∼1600 km), yet differences in the records imply that North America and the northern North Atlantic have distinct geomagnetic behaviors, with the northern North Atlantic appearing more European in character [e.g., Thompson and Barraclough, 1982; Lund, 1996]. Comparisons between paleointensity records (Figure 9) support this contention. The RPI record from Site U1305 shows substantial similarity with European archeomagnetic intensity record (r = 0.77) and less with St. Lawrence Estuary (r = 0.61) or any other North American RPI record [e.g., King et al., 1983; Lund and Banerjee, 1985; Brachfeld and Banerjee, 2000; Barletta et al., 2010].
 Comparisons between northern North Atlantic VGPs and north geomagnetic pole reconstructions, either from VGP [Nilsson et al., 2011] or CSHM [Korte et al., 2005, 2011], show substantial similarity (for clarity only the Nilsson et al.  reconstruction is shown in Figures 11 and 12). Considering that North Atlantic data were not included in the Nilsson et al. [2010, 2011] reconstructions, the similarities are intriguing and suggest that the northern North Atlantic is particularly sensitive to the global signal though with much greater amplitude (Figure 11).
 Observations of a geomagnetic field that deviates substantially from a GAD at around 2500 cal yrs BP (Figures 10 and 12) have been recognized elsewhere. Results from lavas from western North America [Hagstrum and Champion, 2002] and Hawaii [Mankinen and Champion, 1993; Herrero-Bervera and Valet, 2007] display shallow inclinations and far sided VGP locations, while European inclinations are steep with VGP latitudes almost 30° near sided [Thompson and Turner, 1979, Gallet et al., 2002; Snowball et al., 2007, Haltia-Hovi et al., 2010]. Even global reconstructions [Nilsson et al., 2010, 2011] show substantial deviations from a GAD field at that time (Figure 11). Excursional directions from around this time have even been reported [e.g., Raspopov et al., 2003, Dergachev et al., 2004], with key records coming from Georgian archeomagnetic data, and Baltic and Barents Seas sediments. Though evidence for a magnetic excursion at this time is tenuous, the VGP positions estimated using both Site U1305 data and the Greenland/Iceland composite deviate by more than 40° from GAD (Figures 10-12), and therefore, could be classified as excursional [Barbetti and McElhinny, 1972]. Observations of high-amplitude PSV in the North Atlantic region are reinforced by excursional directions from Icelandic lakes at around 2500 cal yrs BP [Ólafsdóttir et al., 2013] and exceptionally high VGP dispersion associated with Icelandic lavas [Kristjansson and Jonsson, 2007]. Archeomagnetic intensity compilations and CSHM [e.g., Genevey et al., 2008; Knudsen et al., 2008; Donadini et al., 2009; Korte et al., 2009, 2011] along with our results in Figures 11 and 12 show that North Atlantic, European, and apparently global intensities were high at this time, indicating that significant deviations from a GAD field are not limited to weak field conditions nor is a dominant GAD a prerequisite for high intensities [e.g., Gallet et al., 2009].
5.2. Influence of the High-Latitude Flux Lobes on the Northern North Atlantic PSV Record
 Historically persistent regions of concentrated geomagnetic flux at the core-mantle boundary (CMB), commonly known as flux lobes, are clearly observed at high-latitude locations in the northern hemisphere below Canada and Siberia [e.g., Bloxham and Gubbins, 1985; Gubbins and Bloxham, 1987; Jackson et al., 2000]. The persistence of these features is one of several lines of evidence suggesting that the geodynamo is influenced by the lower mantle [e.g., Hide et al., 1967; Bloxham and Gubbins, 1987; Laj et al., 1991; Gubbins and Kelly, 1993; Johnson and Constable, 1995; Amit et al., 2011]. The degree of this influence and how it manifests in PSV records on timescales slightly longer than historical is poorly understood and critically important to how we view the geodynamo [Bloxham, 2002; Gubbins et al., 2007; Amit et al., 2010]. By mapping the northern North Atlantic and global VGPs from Figure 11 on to the vertical component of magnetic field (Z) at the core-mantle boundary from 1590 to 1840 [Gubbins et al., 2006], we find that the VGPs are bounded by both flux lobes and a third Euro/Mediterranean feature only hinted at from the historical record (Figure 12), but apparent in CSHMs over the last 3000 yrs [Amit et al., 2011]. The VGP time series in Figure 11 show how these patterns evolve with a shift from primarily North American to European longitudes at ∼3500 cal yrs BP, and a shift back toward North American longitudes prior to the historical period. This shift toward European longitudes is accompanied by increased intensity, both in the North Atlantic and globally (Figure 11). Suggesting that the existence of an ephemeral Euro/Mediterranean VGP attractor (flux lobe?) could be important, not only to the PSV record in the North Atlantic (and elsewhere) but also to the high global field intensities that characterize the late Holocene [e.g., McElhinny and Senake, 1982; Yang et al., 2000; Knudsen et al., 2008; Korte et al., 2011].
 The exceptionally high amplitude of PSV in the northern North Atlantic, most clearly observed over the last 5 kyr at both Site U1305 and from the Greenland/Iceland composite, largely results from the high amplitude of declination (Figure 8). Johnson and Constable  showed that declination at the Earth's surface reflects the radial magnetic field (Br) at the CMB longitudinally peripheral to a site's location. In contrast, inclination and intensity at high northern latitudes reflect Br at the CMB to the south of and centered around a site's location, respectively [e.g., Johnson and Constable, 1997; Johnson and McFadden, 2007]. Large shifts in northern North Atlantic declination would, therefore, result from Br increases at the CMB to its east and west. As a stand in for such a direct comparison, in Figure 13, declination from the northern North Atlantic is compared with archeomagnetic intensities from North America and Europe. High intensities in North America are associated with westward declinations in the North Atlantic and low intensities in Europe, whereas this alternates with low intensities in North American, eastward declinations in the North Atlantic, and high intensities in Europe (Figure 13). Therefore, we argue that the high amplitude of PSV in the northern North Atlantic reflects its location relative to the generally out of phase centennial to millennial scale intensity oscillations in North America and Europe [e.g., Lund and Schwartz, 1999; Gallet et al., 2009].
 The past behavior of high-latitude flux lobes can in principle be at least partially observed through paleointensity reconstructions [Gallet et al., 2009]. Relative intensity highs, either in North America or Europe, and their associated northern North Atlantic declination anomalies, either eastward or westward, correspond to VGP latitude extrema and the timing of “archeomagnetic jerks” [Gallet et al., 2003] (Figures 11 and 13). Defined as sharp changes or cusps in directional drift and intensity maxima from European archeomagnetic data [Gallet et al., 2003], archeomagnetic jerks have been proposed as reflecting the development of a most eccentric (asymmetric) geomagnetic field resulting from the intensification of the high-latitude flux lobes [Gallet et al., 2009]. Therefore, we hypothesize that Holocene PSV largely results from time varying flux concentrations oscillating between a few recurrent high-latitude locations. A “North American mode,” characterized by high North American intensities, western North Atlantic declination, low European intensities, and GAD like VGPs (Figure 11) is more consistent with the time-averaged historical [e.g., Jackson et al., 2000], 10,000 year Holocene [e.g., Korte et al., 2011] and the 5 million year [Gubbins and Kelly, 1993] field models. Whereas a “European mode” characterized by relative lows in North American intensities, eastward North Atlantic declination, high European intensities, and lower latitude VGPs (Figures 11 and 13) are more consistent with time-averaged mid-to-late Holocene field [Korte and Constable, 2005] and much of the last 3000 yrs [Amit et al., 2011]. In this view, the present field appears to be in an interim state, heading from a North American and toward a European mode of variability.
 Site U1305 sediments preserve a high-quality paleomagnetic record of both directions and intensity that provide insights into the northern North Atlantic's paleogeomagnetic variability over the last 8000 yrs. Similarities with ultra-high resolution sediment records from very different environments, north Iceland and east Greenland continental margins [Stoner et al., 2007] and Icelandic lake [Ólafsdóttir et al., 2013], imply that the exceptionally high-amplitude PSV, as evidenced by apparently excursional VGP latitudes as recently as ∼2500 yrs ago, are a robust determination of the region's paleogeomagnetic field, supporting the concept of persistent high amplitude PSV in the Atlantic region [e.g., Gubbins and Gibbons 2004].
 Detailed temporal comparisons show that the age of the magnetization in Site U1305 deep-sea sediments is younger than the age of the sediment, consistent with a pDRM process. This temporal offset and its depth equivalent can be estimated (−289 ± 77 yrs, 20 ± 5 cm) and corrected for under an assumption that ultra high-resolution sediments (e.g., Greenland/Iceland PSV composite) preserve a magnetization with negligible time-lag relative to its chronology. That these records accurately reflect the timing of the paleogeomagnetic field is at least partially tested through comparisons between Greenland/Iceland PSV composite and archeomagnetic predictions for the last 3000 yrs.
 Taken together, these high-resolution and well-dated paleomagnetic records allow us to place northern North Atlantic PSV and RPI into a regional context showing that the northern North Atlantic PSV and RPI are more consistent with European than North American records, and the evolution of VGPs are similar to global reconstructions, though with much larger latitudinal variations. The largest deviation from a geocentric axial dipole are observed during times of highest field intensities in the North Atlantic and globally. This suggests that directional PSV in the northern North Atlantic are a sensitive indicator of temporal oscillations of high-latitude flux concentrations (lobes) at a few recurrent locations as evidenced in global field models and regional paleointensity compilations. More work will be required to fully define these modes, their evolution, their geodynamo significance, and whether they reflect long-term persistence consistent with the idea that mantle heterogeneity has significant control on the geodynamo and the resulting behavior of the geomagnetic field.
 We would like to thank the Captain, crew of the RV JOIDES Resolution, the technical and support staff of the Integrated Ocean Drilling Program (IODP), the Expedition 303 Scientific Party, the staff of the Bremen Core Repository for allowing early sampling, W. M. Keck Carbon Cycle Accelerator Mass Spectrometry Laboratory at UC Irvine for AMS 14C measurements, the Pacific NW Paleomagnetism Laboratory for hysteresis measurements, and Monika Korte for model predictions. We thank the United States Science Support Program and the US National Science Foundation grants OCE 0929066 (JSS) and OCE/EAR 088413/1014506 (JETC) for supporting this research. We acknowledge M. Davies and L. Ziegler for comments and careful reviews by four anonymous reviewers and T. Yamazaki. All data are available from the World Data Center for Paleoclimatology (http://www.ncdc.noaa.gov/paleo/paleo.html).