IODP Proceedings    Volume contents     Search

doi:10.2204/iodp.proc.318.107.2011

Paleomagnetism

Paleomagnetic investigations at Site U1359 involved analysis of discrete samples from Holes U1359A, U1359B, and U1359D and measurement of archive halves from all four holes. In general, we took 2–7 samples per core. A subset of these (45 samples) were subjected to step-wise alternating-field (AF) demagnetization. After measuring the natural remanence of the archive halves, the halves were demagnetized at either 15 or 20 mT, based on the analysis of the discrete samples. The anisotropy of magnetic susceptibility (AMS) was measured on all discrete samples and the rock magnetic properties of a selected subset were also studied. We constructed a composite polarity log from the overlapping sedimentary sections relying on the composite depth determinations in “Stratigraphic correlation and composite section,” which we correlate to the GPTS of Gradstein et al. (2004), documenting a complete Pliocene section from the top of Chron C2An to the bottom of Chron C3An. There is a gap including Chron C2n as well as a period of extremely slow (and possibly discontinuous) sediment accumulation from Chron C3Ar to the top of C5n.

Analysis of discrete samples

Selected discrete samples were measured and demagnetized in fields of 5, 10, 15, 20, 30, and 40 mT and, when strong enough, 50 and 60 mT using the three-step protocol described in “Paleomagnetism” in the “Methods” chapter. Examples of the range of behavior during demagnetization are shown in Figure F29. We encountered two kinds of experimental challenges with these samples. The first challenge stemmed from the fact that the sediments are very weakly magnetized and instrumental noise cannot be neglected. The second challenge results from a bias in directions caused by acquisition of anhysteretic remanence (ARM) during AF demagnetization. After removing the effects of these two types of problems as described in the following sections, the data can in most cases be interpreted in terms of magnetic polarity.

Instrumental noise

The shipboard cryogenic magnetometer suffered from a very sensitive y-axis that undergoes “flux jumps,” which are noise in the form of quantized offsets in a particular detector. All cryogenic magnetometers suffer from this problem, but the shipboard instrument seems to suffer more than shore-based instruments. At the end of the expedition, we discovered that flux jumps could be virtually eliminated by treating the sample track with antistatic spray. However, our results from Site U1359 preceded this discovery. The shipboard sample measurement protocol exacerbated the problem because we measured the discrete samples in batches using the discrete sample tray on the cryogenic magnetometer and not one at a time. By measuring the background before and after measurement the offsets can be detected and the samples can be reinsured. However, if a jump occurs during measurement of a whole sequence of samples, appearing after the first baseline but disappearing again prior to the second baseline measurement, it cannot be easily detected. Incorporation of instantaneous visualization of the data would greatly facilitate the detection of contaminated data, but we have not yet had the time to do this. As a result, flux jumps only became evident during data processing. Because the samples at Site U1359 are so weak, flux jumps profoundly affect the results.

In Figure F29A, F29C, and F29D, we show an example of a demagnetization experiment done on Sample 318-U1359A-12H-5, 70 cm. These data exhibit 4 jumps out of the 26 different measurements. The jumps are readily identifiable as they manifest themselves as spikes in directions along a particular axis (most frequently the y-axis, as in the red symbols in Figure F29A and F29B, but also infrequently on the x- and z-axes) and in the intensity decay curve (jumps in Fig. F29C). Moreover, we measured batches of 14–16 samples at a time and the jump is observed in several sequential samples at the same step. We show the data after removal of the steps suffering from flux jumps and after averaging the data from a given demagnetization step (as described in “Paleomagnetism” in the “Methods” chapter) in Figure F29D. We calculate best-fit lines through these averaged data using principal component analysis (see “Paleomagnetism” in the “Methods” chapter); the best-fit line is shown in green and represents the direction of the characteristic remanence in this sample.

Acquisition of anhysteric remanent magnetization during alternating-field demagnetization

The data shown in Figure F29E–F29H suffer from the second challenge encountered in analysis of Site U1359 samples: ARM acquisition during AF demagnetization. The AF demagnetizer inline with the cryogenic magnetometer has three coils aligned along the instrument’s x-, y-, and z-axes. In turn, each coil ramps up to the specified peak field, the samples (or core) travel through the coil and then the field ramps down again. The procedure is done first on the x-coil, then the y-coil, and finally the z-coil. ARM is acquired parallel to the axis of demagnetization and so is evident in sample coordinates along the last axis exposed to AF demagnetization, the instrument’s z-axis.

During Expedition 318, we developed a protocol whereby each sample gets demagnetized along all three instrumental axes in first the “top-toward” orientation, in which the sample’s z-axis is demagnetized by the AF coil’s z-axis last and could acquire an ARM parallel to the z-axis before measurement. The sample is then demagnetized again at the same peak field, but in the “top-to-right” orientation. In this orientation, the sample’s y-axis is demagnetized by the AF coil’s z-axis last and would acquire an ARM parallel to that axis. Finally, the sample is rotated to the “away-up” orientation, and the sample’s x-axis is exposed to the final demagnetization. This procedure is repeated for each demagnetization step. If a particular sample is susceptible to ARM, it will acquire a magnetization along the z-, y-, and x-axes sequentially, making the distinctive triangular pattern in Figure F29E and the three groups of directions in Figure F29F (labeled x, y, and z). Our demagnetization protocol was designed to compensate for this tendency and, although the example in Figure F29E is a very extreme case, the data averaged for each demagnetization step (Fig. F29H) trend to the origin and can be interpreted in terms of polarity.

Most samples exhibited the behavior shown in Figure F29I–F29L and Figure F29M–F29P. These show only moderate scatter (from ARM) with occasional flux jumps (see a typical reverse sample in Fig. F29L and a typical normal sample in Fig. F29M and F29P). The downward-directed drill string remanence is removed by demagnetization to 10 or 15 mT.

Analysis of archive halves

We measured the natural remanent magnetization (NRM) of archive halves from all holes. Based on the step-wise demagnetization experiments described in the last section, archive halves from Holes U1359A–U1359C were demagnetized to 20 mT and archive halves from Hole U1359D were demagnetized to 15 mT. All NRM data are shown as gray circles in Figure F30 for Holes U1359A–U1359C and the top and bottom of Hole U1359D. The directions of the best-fit lines from step-wise AF demagnetization of discrete samples (e.g., as in Fig. F29) are shown as red triangles and the data from the archive halves after demagnetization and deletion of disturbed sedimentary sections are shown as blue circles. In general, there is excellent agreement between the discrete sample measurements and those from the archive halves, which lends support to an interpretation of the data as to polarity: negative inclinations are normal and positive inclinations are reversed.

Anisotropy of magnetic susceptibility

In addition to remanence analyses, we measured the AMS (including the bulk susceptibility, χb) on all discrete samples (Fig. F31). Samples were taken to avoid to the extent possible disturbed intervals and all but a few (those with V3 inclinations less than ~60° and distinct maximum and intermediate eigenvalues) had characteristic sedimentary fabrics.

There are two pronounced zones of very low bulk magnetic susceptibility (Fig. F31C), one between ~90 and 125 mbsf coincident with a zone of very low remanent intensity (Fig. F30) and a second between ~210 and 275 mbsf. These will be referred to as the Pliocene and Miocene low-susceptibility zones, respectively.

All discrete samples that were subjected to step-wise AF demagnetization were also given an ARM in an alternating field of 100 mT with a direct-current bias field of 50 µT and subjected to progressive IRM acquisition experiments (see “Paleomagnetism” in the “Methods” chapter). We show plots of the progressive acquisition of isothermal remanent magnetization (IRM) for the two low-susceptibility zones and the surrounding sediment (Pliocene and Miocene high-susceptibility zones) in Figure F32. The Pliocene samples from high-susceptibility zones (Fig. F32A) and the low-susceptibility zone (Fig. F32B) are quite different from one another. High-susceptibility sediments display an approach to saturation near 200 mT (characteristic of magnetite), whereas Pliocene low-susceptibility sediments are, for the most part, not saturated by even 1 T, indicating a very high coercivity. Note also the difference in magnitude in the strengths of the IRMs, the low-susceptibility samples being more than an order of magnitude weaker than the high-susceptibility ones. Such high coercivities are characteristic of hematite, a surprising mineral to find in such pyrite-rich, olive-green sediments.

The high-susceptibility Miocene samples are quite similar to those of the Pliocene. The low-susceptibility Miocene samples, however, are on the whole different from the low-susceptibility Pliocene samples. The strongest sample is virtually identical to high-susceptibility samples, displaying the characteristics of a magnetic mineralogy dominated by magnetite. The two weakest samples appear to acquire no stable IRM. The middle sample, though it does not saturate by 200 mT, does appear to reach saturation by ~400 mT, behavior that is more characteristic of maghemite than hematite. These samples do not appear to have a hematite mineralogy but may have either magnetite, an oxidized version of magnetite (e.g., maghemite), or be virtually nonmagnetic.

Discussion

Drop in intensity

The Pliocene low-susceptibility zone is associated with a profound drop in intensity at ~95 mbsf and return to higher values at ~125 mbsf (Fig. F30A, F30B, F30C). A sudden loss of remanence with depth has been reported before (see e.g., Ruddiman, Sarnthein, Baldauf, et al., 1988; Kroenke, Berger, Janecek, et al., 1991; Mayer, Pisias, Janecek, et al., 1992; Barker, Camerlenghi, Acton, et al., 1999; O’Brien, Cooper, Richter, et al., 2001) and has been attributed to dissolution of magnetite during diagenesis (e.g., Tarduno, 1995; Florindo et al., 2003). The situation in the Pliocene low-susceptibility zone is different, however, because the remanence, at least in Hole U1359B, is measurable and stable; it is just extremely weak. The behavior of the IRM acquisition curves (Fig. F32A, F32B) suggests that the magnetization is dominated by hematite within the Pliocene low-susceptibility zone. The magnetite-dominated samples above and below this zone could well have hematite in them, but magnetite is two orders of magnitude stronger and could mask the presence of hematite.

Considering that (1) the lithology shows that the Pliocene low-susceptibility zone has much higher diatom abundance than the overlying sediments (see “Lithostratigraphy”) and (2) trace element data show higher Ba/Al ratios and lower MnO and Na concentrations (see “Geochemistry and microbiology”), it seems likely that the low-susceptibility zone was a period of high biotic productivity. Dissolution of magnetite has been associated with higher productivity (e.g., Hartl et al., 1995) and it is likely that the magnetite originally in the sediments has disappeared through diagenesis. The recovered sediments at Site U1359 are unusual because of the persistence of a magnetic remanence in the form of hematite and the reappearance of high intensities at depth. Secondary hematite is generally found in red pigmented rocks, which these sediments emphatically are not and it would be quite surprising to have hematite forming diagenetically along with such reduced forms as pyrite. We surmise that the hematite is detrital. Our preliminary interpretation is that both magnetite and hematite are present throughout the sequence, but the magnetite is largely gone from the Pliocene low-susceptibility zone, leaving primarily hematite. Alternatively, there is an influx of detrital hematite during the Pliocene low-susceptibility zone, indicating a difference in provenance for the detrital fraction. Further investigations are necessary to resolve the issue of the origin of the magnetic minerals.

Interestingly, the Miocene low-susceptibility zone does not appear to have a hematite-dominated magnetic mineralogy, supporting the change in provenance hypothesis for the Pliocene low-susceptibility zone. The Miocene low-susceptibility zone also appears to have an extremely weak but measurable magnetic remanence.

Correlation of magnetostratigraphy to the geomagnetic polarity timescale

Comparison of the paleomagnetic records from the three holes was greatly facilitated by the composite depth scheme based on physical properties (see “Stratigraphic correlation and composite section”). The inclinations for Holes U1359A–U1359C are shown placed on the composite depth scale in Figure F8A. Data from Hole U1359D are shown in Figure F8B and F8C.

We constructed a composite polarity log for Holes U1359A–U1359C by using the most complete and best resolved intervals from the three holes. The hole from which the composite was constructed is indicated by the colored solid, dashed, or dotted lines to the left of the composite. The major normally magnetized intervals are labeled N1–N12 for convenience.

The composite log can be correlated to the GPTS of Gradstein et al. (2004), shown to the right. Our correlation tie points are listed in Table T12. Starting from the top, the three polarity intervals N1–N3 correspond to Chrons C1n, C1r.1r, and C1r.2r, respectively. The next normal package downhole (intervals N4–N6), starting at ~45 mcd, is not easily matched to Chron C2n without a major change in sediment accumulation rate and/or a hiatus; the inferred part of Chron C1r is too short. A micropaleontological sample taken from Section 318-U1359B-5H-2 at 131 cm is in the T. vulnifica Zone, with an age of 2.17–2.49 Ma (see “Biostratigraphy”). This rules out a tie of the top of interval N4 to the top of Chron C2n and suggests instead a tie to Chron C2An at 2.58 Ma, as shown in Figure F8. It appears that both Chrons C2An.1r and C2An.2r are found in Holes U1359C and U1359A, respectively. There is a strong constraint on hole to hole correlation based on the physical properties in this interval, demanding that these two reverse intervals be different polarity zones as shown in the composite. The reverse interval below interval N6 is a straightforward match to Chron C2Ar and normal intervals N7–N10 correspond to Subchrons C3n.1n–C3n.4n, respectively. Intervals N11 and N12 are the two normal parts of Chron 3An. This correlation is in general agreement with the available radiolarian identifications but conflict with the diatom age estimates within Chron C3n.

The stratigraphic interval from 160 to 220 mbsf in Hole U1359D is shown in Figure F8B. The polarity intervals N11 and N12 are the same as those in Figure F8A and correspond to Chron C3An. The radiolarian stratigraphy (see “Biostratigraphy”) suggests that the interval comprising polarity intervals N13–N15 is highly condensed, ranging in age from ~7 to 9 Ma. Based on these constraints, we correlate Interval N15 to Chron C4An and intervals N13 and N14 to Chron C4n.

The entire record of Hole U1359D is shown in Figure F8C. The biostratigraphy (see “Biostratigraphy”) suggests that the bottom of Hole U1359D is ~12.5 Ma in age, supporting a correlation of polarity interval N17 to Chron C5An, although Chron C5An.1r is missing. Polarity interval N15 can be comfortably correlated to Chron C5n, although there are numerous reverse polarity intervals within it that are not in the standard GPTS. Polarity interval N16 could be Subchron C5r.2n, although it appears a bit higher than expected.

Missing and mislocated polarity intervals can be explained by the highly variable sedimentation expected on a levy. The additional polarity intervals observed in Chron C5n could be either the elusive “tiny wiggles” of Blakely (1974) and Cande and LaBrecque (1974) or the unremoved influence of ice-rafted debris (IRD). To address the latter possibility, we have deleted all intervals associated with visible IRD and those associated with large intensity spikes. Small IRD could influence the record, contributing short perturbations that could be mistaken for geomagnetic field behavior. However, some of the reverse polarity intervals within interval N15 are quite long, spanning a meter or more, and cannot be attributed to hidden IRD. Moreover, a discrete sample from one of the reverse polarity intervals (marked with an arrow in Fig. F8C) also shows a stable reverse polarity (Fig. F30D) and is not contaminated by IRD. Bowles et al. (2003) investigated the issue of “tiny wiggles” within Chron C5n and concluded that if reverse polarity zones exist at all, they must be very short in duration and probably not global in nature. Therefore, although we suspect that at least some of the reverse polarity intervals within interval N15 are in fact geomagnetic in origin, they are probably only locally expressed (excursions as opposed to reversals of the entire dipole field). We surmise that we have recovered these excursions in intervals of extremely high sediment accumulation (100 m/m.y. on average and perhaps much higher for brief periods of time) for which we have very high sampling density (every 5 cm).

Trends in anisotropy of magnetic susceptibility data

The AMS data shown in Figure F31 reflect changes in sedimentary fabric with depth. In simple sedimentary systems, dewatering and compaction of sediment results in a monotonic increase in the degree of anisotropy of an essentially oblate fabric with a vertical axis of minimum susceptibility (V3 in the terminology used here; see “Paleomagnetism” in the “Methods” chapter). The degree of anisotropy is reflected by the difference between the minimum and maximum eigenvalues (plotted in Fig. F31A and referred to as τ3 and τ1, respectively). Instead of a monotonic increase in the difference between these two values, rapid swings between high and low degrees of anisotropy (0 to 20 mcd), intervals of monotonic increase in anisotropy with depth (e.g., 20–25, 90–95, and 115–130 mcd), and intervals of monotonic decrease in anisotropy with depth (e.g., 25–30, 40–50, and 95–100 mcd) occur. Moreover, there are several intervals with very low anisotropy (indicated by stars). Two of these intervals correspond to the two low-susceptibility zones in the Pliocene and Miocene already discussed. The other intervals also correspond to brief intervals of low susceptibility. In fact, with one exception at ~475 mcd, whenever we encountered low susceptibility, we also found a low degree of anisotropy. At least in the Pliocene, these intervals of low susceptibility are also associated with high diatom abundances and may well represent zones of high productivity.

There are several possible explanations for the variations in anisotropy degree with depth at Site U1359. First, there are changes in magnetic mineralogy as discussed earlier, especially with respect to the high- and low-susceptibility regions. If the hematite, for example, is diagenetic in origin, it would not have a sedimentary fabric (e.g., Tauxe et al., 1990), so the changes in anisotropy could result from changes in origin of the magnetic phases. Second, pronounced changes in bioturbated versus laminated beds could also explain alternations between highly oblate and virtually isotropic sedimentary fabrics. Finally, there may be differences in excess pore water caused by the more clay rich versus more diatom rich layers, creating compaction disequilibria of the sort observed by Schwehr et al. (2006) (Fig. F33). The solution to this interesting problem must await further shore-based analyses.