Holocene paleomagnetic secular variation records from the western Equatorial Pacific Ocean

Holocene Paleomagnetic Secular Variation Records from the Western Equatorial Pacific Ocean Steve P. Lund1 Lowell Stott1 Martha Schwartz1 Robert Thunell2 Amy Chen3 1) Department of Earth Sciences, University of Southern California, Los Angeles 2) Department of Geology, University of South Carolina, 3) Torrance High School, Torrance, California

Abstract We have carried out detailed rock magnetic and paleomagnetic studies on a suite of four deep-sea sediment cores from Indonesia and the Philippine Islands, which form a transect from 6.3°N to 10.6°S. The cores have average sediment accumulation rates of 35-65 cm/kyr that all yield high-resolution paleomagnetic secular variation (PSV) records. The PSV records contain 44 correlatable inclination and declination features that corroborate the individual PSV records and indicate that they are all continuous with no significant stratigraphic gaps. Thirty-four AMS radiocarbon dates provide a highresolution time stratigraphic framework for the PSV records, which extend from ~018,000 Years BP. These are the first high-resolution Holocene PSV records ever recovered from anywhere on Earth within 17° of the Equator (almost 1/3 of the Earth’s surface area). Statistical analysis of the Holocene PSV records indicates that they have ΔI anomalies of -5° to -9°, consistent with 5 MA averaged results. The strong similarity in ΔI anomaly pattern of Holocene and longer-term (5 Ma) averaged PSV data suggest that the dynamo process causing the anomaly is largely described by Holocene PSV. By contrast, the Holocene PSV records have VGP angular dispersions that are significantly lower than 5 MA averaged PSV data yet significantly higher than PSV data averaged over only the last few thousand years. This difference indicates that whatever dynamo process causes the overall directional variability of the Earth’s magnetic field is not totally described by recent (last few thousand year) PSV or even by late Quaternary (last ~20,000 year) PSV. Introduction Our knowledge of prehistoric geomagnetic field secular variation since the last magnetic polarity reversal (~780,000 yrs. BP) is very fragmentary due to the limited number of high-resolution paleomagnetic secular variation (PSV) records available. There are no continuous Holocene (0-12,000 yrs. BP) PSV records from anywhere on Earth within 10° of the Equator and only three PSV records from two localities (NE

Australia, 17.3°S [1]; Hawaii, 19.7°N [2]) within 20° of the Equator (more than one-third of the Earth’s surface area). This lack of spatial coverage limits our ability to determine the evolutionary space/time pattern of geomagnetic field secular variation and consider its implications for the core dynamo process that generates the main field. This paper presents new PSV results from a group of four deep-sea sediment piston cores collected from Indonesia and the Philippine Islands region (Figure 1) that form a transect from 6.3°N to 10.6°S latitude. Three of the cores contain complete Holocene sediment sequences with average accumulation rates of 40-65 cm/kyr, which preserve very high-resolution PSV records. The resulting PSV records contain a reproducible record of Holocene geomagnetic field variability that adds significantly to our understanding of low-latitude field variability. Core Descriptions and Sampling In 1998, the RV Marion Dufresne collected an extensive set of deep-sea sediment piston cores from the western Equatorial Pacific Ocean as part of the IMAGES paleoclimate initiative. Four of those cores (Table 1) were sampled for this paleomagnetic study: MD982162 (hereafter referred to as MD62), MD982170 (MD70), MD982177 (MD77), and MD982181 (MD81). MD70 was the only piston core to have a short gravity core (MD70G) collected with it as a trigger core. All five piston and gravity cores were cut into ~150 cm sections, split in half on shipboard, and visually described. All five cores are composed of fine-grained, brown to grey hemi-pelagic sediments that are uniform in appearance. The working halves of all cores are stored at the Oregon State University core repository. The uppermost 900 cm of MD62, 450 cm of MD70, 700 cm of MD77, and 1500 cm of MD81 were identified as containing Holocene and Holocene/Pleistocene transition sediments based on oxygen isotope stratigraphy and sampled for this study. Cores MD70G (0-120 cm) and MD62 (0-900 cm) and the uppermost sections of MD70 (0-150 cm) and MD81 (0-150 cm) were sampled discretely using a square-cross section mini corer [3] to recover one 2x2x2 cm sediment cube per horizon. The bottom half of one section from core MD62, which extended from 225-300 cm, was not sampled for it appeared that the sediments had been distorted during transport. MD70G, MD70, and MD81 (0-150 cm) were sampled at a 2.5 cm interval; MD62 was sampled at a 5 cm interval. All remaining sediment sections were sampled using u-channels to collect ~150cm long, continuous sediment columns with a 2x2 cm square cross section. Paleomagnetic and Rock Magnetic Measurements The initial NRMs of all discrete samples were measured using a 2G cryogenic magnetometer. Selected samples were then demagnetized in alternating magnetic fields (af) up to 100 mT at 10 mT steps and remeasured after each demagnetization step. All samples lost a soft component of magnetization with somewhat random directions by 20 mT and thereafter demagnetized in a straight line toward the origin (Figure 2a). All other samples were then demagnetized at 20 mT and 40 mT and the 20 mT NRM results were

used for the PSV analysis. U-channel measurements followed a similar pattern except that the NRMs of all samples were initially measured and then step-wise af demagnetized and remeasured at 10 mT steps up to 100 mT using an automated 2G u-channel cryogenic magnetometer. The u-channel NRM demagnetization pattern (Figures 2b, c, d) was the same as that for discrete samples with simple af demagnetization in a straight line toward the origin from 20 mT to 100 mT. Here, too, 20 mT NRM results were used for the PSV analysis. Rock magnetic studies started with bulk magnetic susceptibility measurements of all discrete samples using a discrete-sample sensor attached to a Bartington MS-2 susceptibility meter. The magnetic susceptibility of all u-channels was measured at a 2cm interval using a 7-cm diameter long-core sensor attached to the Bartington meter. After NRM demagnetization, all discrete samples and u-channels were then given an anhysteretic remanence (ARM) by applying a 0.05 mT dc field superposed on a 100 mT af field. The ARMs of all discrete samples and uchannels were measured and then stepwise af demagnetized and remeasured at 20 mT and 40 mT. Finally, a saturation isothermal remanence (SIRM) was given to all samples using a 1 T dc field. The SIRMs were then af demagnetized at remeasured at 20 mT and 40 mT. The rock magnetic data for cores MD62, MD70 (NRM and Chi only), MD77, and MD81 are summarized in Figures 3-6. MD62, MD70, and MD77 all contain evidence for a significant change in rock magnetic properties one to nine meters below the sediment/water interface. The rock magnetic properties of all of the older sediments are typically an order of magnitude lower in intensity than those of the younger sediments. The boundaries between stronger and weaker intensities are very abrupt in MD62 (860890 cm, Figure 3) and MD77 (660-680 cm, Figure 5) and occur in Pleistocene-aged sediment (see below). By contrast, the transition in MD70 is much less abrupt (130-210 cm) and occurs in Holocene-aged sediments. We interpret the transition in MD70 to reflect a current chemical redox boundary with oxic conditions above and sulfate reduction and significant magnetic mineral dissolution below. The older boundaries in MD62 and MD77 probably reflect paleoredox boundaries, which terminated when significant magnetic mineral dissolution stopped in late Pleistocene time as organic flux decreased significantly or clastic (Fe) flux increased significantly. Our paleomagnetic records are limited to portions of the cores that do not show significant evidence for magnetic mineral dissolution. Paleomagnetic Secular Variation Records The inclination and declination records for cores MD62, MD70-MD70G, MD77, and MD81 are displayed in Figures 7-10. The data are NRM values after 20-mT af demagnetization. The declination records for each of the cores have been reconstructed by rotating the average declination of each ~150-cm core segment to 0° and then making minor shifts between adjacent core segments based on comparisons with the other independent cores. Cores MD70G and MD70 were correlated using both paleomagnetic and rock magnetic features. The two cores have moderately different sedimentation rates in their uppermost 1.5 meters, even though they were collected within meters of one another. It is possible that one core was compressed during coring relative to the other

core, but there is no evidence for sediment disruption in the correlatable magnetic features. The paleomagnetic results for MD70 and MD70G are plotted versus the MD70 depth scale in Figure 8. As will be discussed below, both cores appear to be missing the last ~2000 years of sediment. The PSV records of all five cores were initially correlated using their known oxygen isotope stratigraphies and 34 AMS radiocarbon dates (Table 2, see also Figures 710). On the basis of these initial correlations, we were able to identify 21 inclination features and 23 declination features, which are distinctive and correlatable among the five independent PSV records. Those features are numbered in Figures 7-10. These 44 correlatable features are consistent stratigraphically with the independent isotope and radiocarbon dating; the features are also consistent among themselves with proper inclination/declination phase relationships preserved and with no significant features missing (or extra) in any one record. The correlatable paleomagnetic features have been used to develop stratigraphic transfer functions between cores MD62, MD77, and MD81. These depth-depth transfer functions are shown in Figure 11; MD81 is used as the ‘standard’ reference core because it has the highest average sediment accumulation rate of all studied cores. The transfer functions generally indicate smoothly proportional and nearly constant sediment accumulation among the three long records. The smoothly continuous transfer functions between cores and presence of all significant directional features in all cores both suggest that no single core has marked discontinuities in sedimentation (>10-20 cm) relative to the other cores. Chronostratigraphy 34 AMS radiocarbon dates, listed in Table 2, have been recovered from the Holocene and Latest Pleistocene parts of cores MD62, MD70, MD70G, MD77, and MD81. These dates have been used to develop individual chronologies for all of the cores. All of the dates have been calibrated using the Calib 5.0 program of Stuiver et al. [4] and reservoir corrections of Broecker et al. [5]. The median calibrated ages and the lower and upper limits of permissable ages are listed in Table 2; the calibrated ages are used below to develop individual core chronologies. 28 radiocarbon dates for cores MD62, MD77, and MD81 are plotted in Figure 12 using the MD81 stratigraphy and transfer functions from Figure 11. Four radiocarbon dates from cores MD70 and MD70G were not used for this analysis because the PSV correlations from those cores to MD81 were less constrained than we deemed appropriate. Similarly, the radiocarbon dates at 992 cm in MD62 and 12 cm in MD77 are not plotted in Figure 12 because they are located beyond the interval of correlatable PSV data and are therefore not well constrained in their possible stratigraphic correlations to MD81. The 28 radiocarbon dates in Figure 12 generally follow a smooth trend from 0 to ~23,000 Cal Yrs. BP. But, four of the dates, indicated by arrows in Figure 12, have age ranges that are too old relative to the remaining dates. Three of these dates come from MD81 and one from MD62. We presume these dates are contaminated by anomalously old carbon and have removed them from further consideration. The pair of anomalous old

dates near 10 meters in MD81 raise some possibility that this is a contaminated interval (slump?) in MD81. However, the PSV within this interval is distinctive and correlatable with cores MD62 and MD77, and a date of 13,545 Cal. Yrs. BP from MD77 falls right on the expected trend of all other dates near 10 meters in Figure 12, based on the PSV correlations and transfer function in Figure 11. Therefore, we think that the MD81 dates near 10 meters are anomalous but the sediment interval itself and PSV record are reasonable. The solid curve in Figure 12 is our preferred chronology for MD81. The curve was developed using only the MD81 radiocarbon dates and smoothly goes through the allowable age ranges of all MD81 dates. The curve also goes through the allowable age ranges of all other dates from MD62 and MD77 that have been transferred to the MD81 stratigraphy. This corroborates both the MD81 radiocarbon dates and the PSV correlations noted above. We can also use the radiocarbon dates in Table 2 and depth-depth correlations of Figure 11 to develop chronologies for MD62 and MD77. Figure 13 plots all seven MD62 radiocarbon dates and 14 dates from MD77 and MD81 that could be correlated using the PSV. The dates follow a smooth progression and we have drawn a simple smooth curve composed of five linear segments though all the data. This curve goes through all of the MD62 permissable age ranges and most of the MD77 and MD81 age ranges as well. Figure 13 is our preferred chronology for MD62. Figure 14 plots all five of the MD77 radiocarbon dates together with 16 dates from MD77 and MD81 that could be correlated using the PSV. These dates also follow a smooth progression and we have drawn a simple smooth curve composed of five linear segments though all these data as well. This curve goes through three of the MD77 permissable age ranges and most of the MD62 and MD81 age ranges. But, two of the MD77 dates (3586, 10050) appear to be biased to old age relative to other nearby dates from MD62 and MD81; these dates are also biased to old age relative to the derived chronologies for MD81 and MD62. We have tried to draw interval line fits that best reflect all of the plotted dates in each interval. Figure 14 is our preferred chronology for MD77. Finally, we have developed a tentative chronology for MD70 and MD70G. We previously used the PSV, rock magnetic, and limited oxygen isotope data from cores MD70 and MD70G to develop a depth-depth correlation between MD70 and MD70G, which permitted us to plot the PSV data in Figure 8. We have used that same correlation to plot the four MD70 and MD70G radiocarbon dates versus depth in MD70 in Figure 15; we have added three age estimates for deeper parts of MD70 based on microfossil datums. We have also added three radiocarbon dates from cores MD62, MD77, and MD81 that we have tentatively correlated using the PSV records in Figure 8. All seven of the radiocarbon dates can be fit by a single straight line in Figure 15 with an intercept of about 1700 Cal. Yrs. BP. The intercept suggests that cores MD70 and MD70G are both missing the last ~1500-2000 years of sediment. We have no clear explanation for the apparent sediment loss, but it could be due to either loss of soft surface sediment from both cores during coring or a local erosional discontinuity. The straight-line curve in Figure 15 is our preferred chronology for the Holocene part of MD70 and MD70G.

PSV Analysis The final inclination and declination records for MD62, MD70, and MD81 are plotted versus time in Figures 15 and 16. The numbered inclination and declination features from Figures 7-10 are shown as well. It is clear that the three records share the same general PSV pattern with all numbered features being present in all three cores and having ages that are not significantly different. One complication is that MD77 has inclination and declination amplitudes that are only half that of the two other cores, which bracket it spatially. We think that the diminished PSV amplitudes in MD77 are due to increased smoothing of the PSV record due to relatively low sediment accumulation rate (~40 cm/kyr versus ~50 cm/kyr and ~65 cm/kyr for MD62 and MD81 respectively) and shallower water depth (~900 m versus ~1900 m and ~2100 m for MD62 and MD81 respectively), which probably produce a thicker interval of sediment bioturbation and smoothing at the site of MD77. Our new PSV records can be used to assess the general characteristics of lowlatitude geomagnetic field variability by statistical comparison with PSV from other parts of the World. Two statistical parameters most often measured in PSV studies are the ΔI anomaly, which is the site mean inclination minus the expected axial dipole field inclination, and the angular dispersion associated with a site's vector (or equivalent VGP) variability. The global pattern of the ΔI anomaly estimates how well the axial-dipole hypothesis, the cornerstone of plate tectonic reconstructions, fits the actual spatial pattern of geomagnetic field behavior. The global pattern of angular dispersion should provide some measure of the spatial pattern of intrinsic ‘energy’ or ‘dynamics’ in the core dynamo process. This variability may be due to differing proportions of dipole versus non-dipole field variability [6] or it could be interpreted as the relative importance of primary versus secondary family spherical harmonic components [7], two orthogonal components of the geomagnetic field that may have fundamental relationships to dynamo theory [8]. Table 3 summarizes the ΔI anomaly and VGP angular dispersion in our four new Equatorial PSV records. The ΔI anomaly was perhaps first quantified by Wilson [9] who noticed that the average paleomagnetic pole positions associated with individual geographic regions (e.g., Australia, Europe, North America) were always farther from the sampled region than the known geographic pole. This offset, termed the far-sided effect, is due to paleomagnetic inclinations that are systematically more negative than their axial dipole expectation (negative ΔI anomaly), on average. McElhinney et al. [10] have determined the global ΔI anomaly for the last 5 Ma (Figure 18, solid circles), on average, and noted that the ΔI anomaly is mostly negative with a maximum anomaly near the Equator and a latitudinal variation that is symmetric about the Equator and zonal (any site along a line of latitude will have the same magnitude of ΔI anomaly). Their analysis also indicates that the ΔI anomaly has persisted with the same general pattern and magnitude for the last 5 million years during both normal and reversed polarity. But, Johnson and Constable [11] have argued that non-zonal components of the ΔI anomaly are significant and have persisted for the last several million years, with some Equatorial regions having more significant negative ΔI anomalies than other regions and some high-latitude regions actually having positive ΔI anomalies (see also [12]).

Holocene PSV records provide an important shorter-term view of the ΔI anomaly. PSV records for the last several thousand years ([13], PSVMOD1.0 time series downloadable at http://eart.usc.edu/~slund), shown by open circles in Figure 18, display the same general ΔI anomaly pattern as the 5-Ma averages (although they are missing data from Equatorial latitudes), but they do suggest the presence of positive ΔI anomalies near 50° latitude in both hemispheres. Previously published PSV data for the last ~30,000 years [2, 12] and our new Holocene results from Equatorial latitudes, shown by open squares in Figure 18, also replicate the general ΔI anomaly pattern of the 5-Ma averages. Moreover, they also corroborate the positive ΔI anomaly results near 50°N noted above and they fill in the short-term statistical data gap for Equatorial latitudes. The strong similarity between the short-term (103 and 104 yr ave) and long-term (106 yr ave) ΔI anomaly results suggests that whatever causes the ΔI anomaly pattern is a dynamo process that is active today and totally represented in the PSV behavior of the last few thousand years and overall Holocene. Given the apparent common appearance of positive ΔI anomalies near 50°N and S, we would hypothesize that high-latitude flux lobes [14], which occur historically [14, 15] and in the paleomagnetic record of the last few thousand years [13], are the primary source for the high latitude positive ΔI anomalies. It is also likely that their global pattern [14, 16] may well be the source for the significant negative ΔI anomalies at low latitude as well. In this view, the overall variability of high-latitude flux lobes (and their overall outer core structure) is largely described by PSV of the past few thousand years and has persisted with both normal and reversed polarity for millions of years. It is also important to note that the ΔI anomaly in our new PSV records cannot be caused by inclination errors (anomalous shallowing of paleomagnetic field directions) in sediment sequences. MD81 and MD77 are sediment sequences in the Northern Hemisphere that have shallower than expected inclinations, which produce negative ΔI anomalies; while, MD70 and MD77 are sediment sequences in the Southern Hemisphere that have steeper than expected inclinations, which produce negative ΔI anomalies. If sediment shallowing were significant in all four of these sediment sequences, we would see negative ΔI anomalies in the Northern Hemisphere and positive ΔI anomalies in the Southern Hemisphere. Analysis of PSV angular dispersions (either directions or their equivalent virtual geomagnetic poles, VGPs) has established that angular dispersion also displays a distinctive zonal pattern of amplitude variation with latitude. The average latitudinal pattern of VGP angular dispersion for the last 5 Ma [17] is shown in Figure 19 (closed circles). (We use and prefer the form of VGP angular dispersion that is calculated relative to the mean VGP of the data distribution. An alternative form of the VGP angular dispersion [6] is calculated relative to the geographic North Pole. This adds a bias to the latitudinal pattern of VGP angular dispersion due to the latitudinal dependence of the ΔI anomaly.) The smallest variability in VGP angular dispersion occurs within 30° of the Equator and then increases smoothly with increasing latitude up to at least 70°. PSV records for the last several thousand years [13], shown by open circles in Figure 19, display a similar overall pattern of VGP angular dispersion variability but with significantly lower values when compared with the 5-Ma averages. (The exceptionally low variability of three European sites in Figure 19 is notable, but not easily explained.)

Previously published PSV data for the last ~30,000 years [2, 12] and our new Holocene results from Equatorial latitudes, shown by open squares in Figure 19, also display a similar overall pattern of VGP angular dispersion but with values that are significantly lower than the 5-Ma average VGP angular dispersion yet significantly higher than the data for the last few thousand years. A variety of parametric models, summarized in [6], have been developed to attempt to explain the observed spatial pattern of angular dispersion (biased by the ΔI anomaly) in an ad hoc manner. More recently, McFadden et al. [7] were able to relate the long term average pattern of VGP angular dispersion to historical field observations and developed model G, based on the relative importance of primary versus secondary family magnetic field components [8], to explain the latitude dependence. Whatever the source of the observed latitude-dependent VGP angular dispersion, it is clear that the source variability is not adequately described by recent (last few thousand year) PSV or even by known late Quaternary (last ~30,000 year) PSV. If the global pattern of VGP angular dispersion provides some measure of the spatial pattern of intrinsic ‘energy’ or ‘dynamics’ in the core dynamo process, then the total range of energy or dynamics is only seen on timescales >>20,000 years. Conclusions We have carried out detailed rock magnetic and paleomagnetic studies on a suite of four deep-sea sediment cores (MD62, MD70, MD77, MD81) from Indonesia and the Philippine Islands, which form a transect from 6.3°N to 10.6°S. The cores have average sediment accumulation rates of 35-65 cm/kyr that all yield high-resolution PSV records. The four PSV records contain 21 correlatable inclination features and 23 correlatable declination features that corroborate the individual PSV records and indicate that they are all continuous with no significant stratigraphic gaps. Thirty-four AMS radiocarbon dates provide a high-resolution time stratigraphic framework for the PSV records. The final PSV records for MD62, MD77, and MD81 extend from ~0-18,000 Years BP, while the PSV record for MD70 only extends from 2-6,000 Years. BP. These are the first highresolution paleomagnetic records of Holocene PSV ever recovered from anywhere on Earth within 17° of the Equator (almost 1/3 of the Earth’s surface area). Statistical analysis of the four PSV records indicates that they have ΔI anomalies of -5° to -9°, consistent with previously published results. These results cannot be due to sediment inclination shallowing because the two Southern Hemisphere PSV records (MD62, MD70) have negative ΔI anomalies due to average inclinations steeper than axial dipole expectation. A strong similarity in ΔI anomaly pattern of Holocene and longer term (5 Ma) averaged PSV data suggest that the dynamo process causing the anomaly is largely described by Holocene PSV. By contrast, the four PSV records have VGP angular dispersions that are significantly lower than 5 MA averaged PSV data yet significantly higher than PSV data only averaged over the last few thousand years. This difference indicates that whatever dynamo process causes the directional variability of the Earth’s magnetic field is not totally described by recent (last few thousand year) PSV or even by late Quaternary (last ~30,000 year) PSV.

References [1] Constable, C. and M. McElhinny, Holocene geomagnetic field secular variation records from northeastern Australian lake sediments, Geophys. J. Roy. Astron. Soc., 81, 103-120, 1985. [2] Peng, L. and J. King, A late Quaternary geomagnetic secular variation record from Lake Waiau, Hawaii, and the question of the Pacific nondipole low, J. Geophys. Res., 97, 4407-4424, 1992. [3] Lund, S., Late Quaternary Secular Variation of the Earth's Magnetic Field as Recorded in the Wet Sediments of Three North American Lakes, unpublished PhD dissertation, University of Minnesota, Minneapolis, 254 p., 1981. [4] Stuiver et al. (2005) [5] Broecker et al. (2004) [6] Merrill, R., M. McElhinny, and P. McFadden, The magnetic field of the Earth, paleomagnetism, the core, and the deep mantle, Academic Press, 531 p., 1998. [7] McFadden, P., Merrill, R., and M. McElhinny, Dipole/quadrapole family modeling of paleosecular variation, J. Geophys. Res., 93, 11583-11588, 1988. [8] Roberts, P. and M. Stix, alpha-effect dynamos by the Bullard-Gellman formalism, Astron Astrophys., 18, 453-466, 1972. [9] Wilson, R., Permanent aspects of the Earth’s nondipole magnetic field over Upper Tertiary times, Geophys. J. Roy. Astron. Soc., 19, 417-437, 1970 [10] McElhinny, M., P. McFadden, and R. Merrill, The time-averaged paleomagnetic field 0-5 Ma, J. Geophys. Res., 101, 25007-25027, 1996. [11] Johnson, C., and C. Constable, The time-averaged geomagnetic field as recorded by lava flows over the last 5 myr, Geophys. J., 122, 489-519, 1995. [12] Lund, S., A comparison of the statistical secular variation recorded in some late Quaternary lava flows and sediments, and its implications, Geophys. Res. Lett., 12, 251-254, 1985. [13] Constable C., C. L .Johnson, and S. P. Lund, Global geomagnetic field models for 03 ka: transient or permanent flux lobes, Phil. Trans. Royal Society, A, 358, 991-1008, 2000. [14] Gubbins, D. and J. Bloxham, Morphology of the geomagnetic field and implications for the geodynamo, Nature, 325, 509-511, 1987. [15] Bloxham, J. and D. Gubbins, The secular variation of the Earth’s magnetic field, Nature, 317, 777-781, 1985. [16] Bloxham, J. and D. Gubbins, The evolution of the Earth’s magnetic field, Scientific American, 261, 68-75, 1989. [17] Johnson, C. and C. Constable, Paleosecular variation recorded by lava flows over the past 5 million years, Phil. Trans: Math. Phys. Eng., 354, 89-141, 1996.

Figure Captions Figure 1: Map of the western Equatorial Pacific Ocean region showing the locations of deep-sea sediment piston cores studied in this paper. Figure 2: Zijdereldt diagrams illustrating the NRM demagnetization pattern of four selected sample horizons from the studied cores. Solid (open) symbols track the horizontal (vertical) component of the initial NRM and NRM after step-wise af demagnetization in 10 mT steps up to 100 mT. Figure 3: Rock magnetic characteristics of core MD62. The grey area indicates the transition from relatively strong magnetic intensities above to very weak magnetic intensities below, which we interpret to be a paleoredox boundary. Figure 4: Selected rock magnetic characteristics of core MD70. The grey area indicates the transition from relatively strong magnetic intensities above to very weak magnetic intensities below, which we interpret to be the current redox boundary. Figure 5: Rock magnetic characteristics of core MD77. The grey area indicates the transition from relatively strong magnetic intensities above to very weak magnetic intensities below, which we interpret to be a paleoredox boundary. Figure 6: Rock magnetic characteristics of core MD81. This core does not show any significant evidence for magnetic mineral dissolution associated with current or paleo redox boundaries. Figure 7: Paleomagnetic results (20 mT NRMs) from core MD62. Selected inclination and declination features are numbered, which can be correlated with the other independent PSV records from the study area. Calibrated AMS radiocarbon dates from this core are shown at top. Figure 8: Paleomagnetic results (20 mT NRMs) from cores MD70 (open symbols) and MD70G (closed symbols). Selected inclination and declination features are numbered, which can be correlated with the other independent PSV records from the study area. Calibrated AMS radiocarbon dates from this core are shown at top. Figure 9: Paleomagnetic results (20 mT NRMs) from core MD77. Selected inclination and declination features are numbered, which can be correlated with the other independent PSV records from the study area. Calibrated AMS radiocarbon dates from this core are shown at top.

Figure 10: Paleomagnetic results (20 mT NRMs) from core MD81. Selected inclination and declination features are numbered, which can be correlated with the other independent PSV records from the study area. Calibrated AMS radiocarbon dates from this core are shown at top. Figure 11: Depth-depth transfer functions for cores MD62 and MD77 relative to MD81 (the master core for all stratigraphic correlations). The correlations are based on the numbered inclination and declination features noted in Figures 7-10. Figure 12: Time-depth plot for core MD81. Arrows indicate individual radiocarbon dates that are anomalously old and presumed to be contaminated with ancient carbon. See text for more detailed discussion. Figure 13: Time-depth plot for core MD62. See text for more detailed discussion. Figure 14: Time-depth plot for core MD77. See text for more detailed discussion. Figure 15: Time-depth plot for cores MD70 and MD70G. See text for more detailed discussion. Figure 16: Inclination time series for the three most complete PSV records recovered in this study. Numbered correlatable inclination features match those noted in Figures 7-10. Figure 17: Declination time series for the three most complete PSV records recovered in this study. Numbered correlatable declination features match those noted in Figures 7-10. Figure 18: ΔI anomaly as a function of latitude for selected time intervals: solid symbols – last 5 Ma average, open symbols last few thousand years average, open squares, last ~10-20,000 years average. New results from this study are labeled for clarity. Figure 19: VGP angular dispersion as a function of latitude for selected time intervals: solid symbols – last 5 Ma average, open symbols last few thousand years average, open squares, last ~10-20,000 years average. New results from this study are labeled for clarity.