An environmental magnetic study was conducted to investigate temporal variations of the sources of magnetic minerals in the Pacific Ocean using sediment cores of Pleistocene age. The proportion of interacting (I) to noninteracting (N-I) single-domain components was estimated from first-order reversal curve diagrams, and the proportion of middle- (M) to low- (L) coercivity components was determined from decomposition of isothermal remanent magnetization (IRM) acquisition curves. The I and M components are interpreted to be carried by terrigenous maghemite, while the N-I and L components represent biogenic magnetite. An inverse correlation between the ratio of anhysteretic remanent magnetization susceptibility (χARM) to saturation IRM (SIRM) and the I/N-I ratio was confirmed for sediments across wide expanses in the Pacific Ocean. This implies that χARM/SIRM can be used to estimate the relative abundances of biogenic and terrigenous components in sediments. In the North Pacific, the relative abundance of the I and M components increases in glacial periods. Variations of χARM/SIRM and S ratio (S−0.1T) resemble each other and decrease in the same time intervals. These variations reflect increases of terrigenous input in glacial periods as eolian dust. On the Ontong-Java Plateau in the western equatorial Pacific, on the other hand, minima in χARM/SIRM and S−0.1T occur at glacial-to-interglacial transitions, and the relative abundance of the I and M components increases in these periods. Terrigenous material in this region is considered to be transported mainly from the New Guinea and Solomon Islands by the Equatorial Undercurrent. The higher proportion of the terrigenous component at glacial-to-interglacial transitions is coeval with carbonate preservation maxima reported from the Ontong-Java Plateau, suggesting a linkage between the two.
If you can't find a tool you're looking for, please click the link at the top of the page to "Go to old article view". Alternatively, view our Knowledge Base articles for additional help. Your feedback is important to us, so please let us know if you have comments or ideas for improvement.
 Recently, first-order reversal curve (FORC) diagrams [Pike et al., 1999; Roberts et al., 2000] have become widely used for characterizing magnetic mineral assemblages in paleomagnetism and rock magnetism [e.g., Weaver et al., 2002; Rowan and Roberts, 2006; Roberts et al., 2006; Wetter et al., 2007; Chen et al., 2007]. They provide information on the statistical distribution of coercivities (Hc) and local interaction fields (Hu) for the magnetic grain assemblage within a sample. Using FORC diagrams, Yamazaki  semiquantitatively estimated relative abundances of a noninteracting single-domain (SD) component (hereafter called the N-I component) and an interacting SD component (the I component) in North Pacific sediments. Together with decomposition of isothermal remanent magnetization (IRM) acquisition curves, it was estimated that the N-I component is mainly carried by biogenic magnetite and that the I component is carried by terrigenous maghemite. The ratio of the ARM susceptibility (χARM) to the saturation IRM (SIRM) of the samples decreases with increasing proportion of the I component, which can be explained by the dependence of ARM acquisition on magnetostatic interaction. This implies that the χARM/SIRM ratio does not necessarily reflect magnetic grain size; alternatively, it can be used to estimate the relative abundance of biogenic and terrigenous components in sediments.
 In this study, I demonstrate the usefulness of the new approach, the combination of the I/N-I ratios from FORC diagrams and χARM/SIRM, for paleoceanographic applications. For this purpose, I investigate magnetic properties of deep-sea sediments over wide areas of the Pacific Ocean. I first apply the approach to North Pacific sediments to examine whether temporal changes in a terrigenous component can be better resolved. In the North Pacific, sedimentological and paleoceanographic studies revealed that a major source of sediments is eolian dust transported from arid regions of the Asian continent [Rex and Goldberg, 1958; Blank et al., 1985], and that the eolian flux varies with glacial-interglacial changes [e.g., Hovan et al., 1989, 1991]. An environmental magnetic study of nonfossiliferous pelagic clay (red clay) in the North Pacific [Yamazaki and Ioka, 1997] demonstrated that the S ratio (S−0.3T), which is a proxy for magnetic mineralogy, reflects variations in the contribution of eolian dust. However, it was not possible to resolve orbital-scale variations because sedimentation rates of the red clays are low. Next, I apply the method to investigate temporal changes of a terrigenous component in the western equatorial Pacific using sediment cores from the Ontong-Java Plateau. Previous paleoceanographic studies on the Ontong-Java Plateau mainly focused on carbonate production/preservation changes on orbital time scales using sediments recovered during Ocean Drilling Program (ODP) Leg 130 and its precursor site survey cruises [Wu et al., 1991; Yasuda et al., 1993; Zhang et al., 2007]. Temporal and spatial changes of sediment sources in the equatorial Pacific sediments are not yet well understood, although some geochemical studies conducted at 140°W, to the east of Ontong-Java Plateau, discussed sediment sources [Lacan and Jeandel, 2001; Anderson et al., 2006; Ziegler et al., 2008]. Intensive rock magnetic studies on magnetic property changes associated with reduction diagenesis were conducted using Ontong-Java Plateau sediments [Tarduno, 1994, 1995; Tarduno and Wilkison, 1996; Tarduno et al., 1998; Smirnov and Tarduno, 2000, 2001], and these results are applicable to paleoenvironmental studies because reduction diagenesis is associated with productivity variations induced by climatic changes. These studies, however, did not address the sources of magnetic minerals except for a possibility of enhanced biogenic magnetite production just above the Fe redox boundary.
2. Sediment Cores and Age Control
 Four cores from widely separated regions in the Pacific Ocean were used in this study; gravity core NGC65 from the North Pacific, two gravity cores (NGC36 and NGC88) from the northern slope of the Ontong-Java Plateau in the western equatorial Pacific, and piston core KR9912-PC5 from the top of the Manihiki Plateau in the central South Pacific (Figure 1). The positions and water depths of the coring sites, and lengths of the cores are listed in Table 1. Core NGC65 is from the North Pacific Subtropical Gyre, and consists of brownish siliceous clay. Cores NGC36 and NGC88 occur in a belt of enhanced productivity along the equator, and consist of nannofossil foraminiferal ooze. The sediment color in these cores gradually changes with depth from light yellow to light gray. The Manihiki Plateau lies in the South Pacific Subtropical Gyre, and core KR9912-PC5 is composed of pale yellowish white foraminiferal ooze. The paleomagnetism of these cores was reported by Yamazaki et al.  (NGC36), Yamazaki  (NGC65), Yamazaki  (NGC36 and NGC88), and Yamazaki and Oda  (KR9912-PC5).
 Age control for core NGC65 is based on correlation of relative paleointensity variations with the stacked paleointensity curve Sint-800 of Guyodo and Valet  (Figure 2). This core spans the last ∼650 ka, with an average sedimentation rate of about 10 m/Ma. Oxygen isotope (δ18O) analysis could not be applied because the coring site lies below the carbonate compensation depth (CCD). Relative paleointensity data for this core were presented by Yamazaki ; this work was done before publication of the Sint-800 stack, so the age of the core was estimated from the resemblance of S ratio (S−0.3T) variations to the standard deep-sea δ18O curve. No phase difference between the two was assumed. However, it is undesirable to base the age model on the S ratio in the present study because the aim is to discuss temporal paleoenvironmental variations using rock magnetic parameters. Thus, the age model for the core has been redeveloped using relative paleointensity variations. The paleointensity-based age model does not differ significantly from the S−0.3T-based age model, which supports the assumption that there is no phase lag between S−0.3T and δ18O.
 The age models for Ontong-Java Plateau cores, NGC36 and NGC88, were determined by combined δ18O stratigraphy and relative paleointensity (Figure 3). The age model for core NGC36 was estimated from the δ18O stratigraphy, which was published by Yamazaki et al.  and Yamazaki , but the target curve has been changed to the LR04 stack [Lisiecki and Raymo, 2005] and age tie points have been slightly modified in the present study. Intercore correlation between cores NGC36 and NGC88 is based on relative paleointensity variations [Yamazaki, 2002]; δ18O ages for core NGC36 were transferred to core NGC88 down to a depth of ∼4.3 m. Below 4.3 m, relative paleointensity variations were correlated to Sint-800 to constrain the age model. The ages of the sediments at the bottoms of cores NGC36 and NGC88 are approximately 540 and 730 ka, with average sedimentation rates of about 9 and 8 m/Ma, respectively.
 Core KR9912-PC5 covers the last ∼2.2 Ma, as estimated from magnetostratigraphy and correlation between the LR04 δ18O stack and magnetic susceptibility variations [Yamazaki and Oda, 2005]. Sedimentation rates decrease upward, and the average rate during the Brunhes chron is about 7.4 m/Ma. This core will be used to compare spatial variations of magnetic properties, but not temporal changes, because the age model was not well enough resolved compared to other cores.
 Magnetic hysteresis and FORC measurements were conducted using an alternating gradient magnetometer (AGM, Princeton MicroMag 2900). For core NGC65 from the North Pacific, 55 specimens including 36 published by Yamazaki  were measured. For cores NGC36 and NGC88 from the Ontong-Java Plateau, 36 and 50 specimens, respectively, were selected from the entire depth ranges of the cores. A total of 22 specimens were measured for core KR9912-PC5 from the Manihiki Plateau. The field spacing between measurements was set to 0.5 mT. A total of 191 FORCs were measured, with Hc between 0 and 60 mT, and Hu between −15 and 15 mT. The narrow field spacing was adopted to precisely depict the shape of the peak near the Hc axis that represents a noninteracting SD grain assemblage, although this might amplify noise. The maximum applied field was 1.0 T. The averaging time spent at each data point was 200 ms for specimens from cores NGC36, NGC65, and NGC88. For specimens from core KR9912-PC5, which have extremely weak magnetizations, a 400 ms averaging time was used. A smoothing factor (SF) [Roberts et al., 2000] of 3 was used for all specimens. The FORCIT software of Acton et al.  was used for data processing.
 For semiquantitative estimation of the relative contribution of the I and N-I components from the FORC diagrams, the same procedure as that of Yamazaki  was adopted: curve fitting of a cross section that parallels the Hu axis and that crosses the peak of Hc (Figure 4). It is assumed that the profile consists of three components, the single-domain (SD) N-I and I components, and a multidomain (MD) component, each with a Gaussian distribution of Hu. The standard deviation of each component was fixed to be 0.7 mT (N-I component), 6 mT (I component), and 23 mT (MD component), following Yamazaki , and the relative abundance of the three components was varied by trial and error so as to achieve an optimal fit, which is similar to the fitting method in the IRM decomposition process of Kruiver et al. . The MD component is represented nearly as a straight line on these cross sections. The lower half (negative Hu) of the FORC diagrams was used for the component fitting.
 IRM acquisition curves were measured using the AGM. One hundred measurements were made at equidistant field steps on a log scale ranging from 3 mT to 1.4 T. The IRM acquisition curves were decomposed into magnetic coercivity components using the method of Kruiver et al.  assuming that the IRM acquisition curves are a linear addition of components represented by cumulative log-Gaussian functions.
 Magnetic parameters for the four cores including magnetic susceptibility, ARM, SIRM, and S−0.3T have been published along with the respective paleomagnetic results [Yamazaki et al., 1995; Yamazaki, 1999, 2002; Yamazaki and Oda, 2005]. For S−0.1T, measurements were newly conducted on cores NGC36 and NGC88 using a pulse magnetizer (2G Enterprises model 660). First, an IRM was imparted at 2.5 T for discrete samples, and then an IRM of 0.1 T was successively imparted in the direction opposite to the initial IRM. The definition of Bloemendal et al.  was used to calculate S ratios. For core NGC65, the values of S−0.1T were substituted by those measured for piston core KR0310-PC1. The two cores were taken at the same location, and precise correlation between the cores was conducted using magnetic susceptibility variations [Yamazaki and Kanamatsu, 2007].
 In order to obtain further information on magnetic mineralogy and grain size, low-temperature magnetic measurements were made on dried specimens from cores NGC36 and NGC88 using a low-temperature SQUID susceptometer (Quantum Design MPMS-XL5). An IRM was imparted to the specimens in a field of 2.5 T after cooling to 6 K in a zero field, which was followed by measurement of the IRM up to 300 K. For some specimens, the following measurement sequence was conducted before the thermal demagnetization of a low-temperature IRM: an IRM was imparted at 300 K in a 2.5 T field, and magnetization changes were measured when cycling the temperature between 300 K and 6 K in a nearly zero field. For core NGC65, low-temperature magnetic measurements have been published by Yamazaki .
 Typical FORC diagrams for the studied samples (Figure 5) have a narrow ridge along the Hc axis with a small vertical spread, which indicates weak magnetostatic interaction. There are some differences in the vertical spread at the base of the ridge; for example, the spread of the specimen from core KR9912-PC5 is the smallest among the four examples, although the difference among the other three is not visually clear in Figure 5. The ratios of the I to N-I components were obtained from the results of the component fitting to cross sections of the FORC distributions. When plotting the I/N-I component ratio versus the χARM/SIRM ratio, data from the Manihiki and Ontong-Java Plateaus inversely correlate in the same way as data from the North Pacific sediments [Yamazaki, 2008] (Figure 6a). Specimens from the Manihiki Plateau (core KR9912-PC5) have small I/N-I ratios, as indicated by the small vertical spread on the FORC distributions in Figure 5. Data from core NGC65 from the North Pacific and cores NGC36 and NGC88 from the Ontong-Java Plateau cluster in a similar region on the I/N-I versus χARM/SIRM diagram; the latter cores tend to have slightly higher χARM/SIRM ratios than the former, although both have similar I/N-I ratios. The strong inverse correlation suggests that the χARM/SIRM ratio is dominantly controlled by the strength of magnetostatic interactions in pelagic sediments across wide expanses of the Pacific Ocean, not just the North Pacific.
 Typical results of IRM component analyses (Figure 5) indicate the presence of two dominant components, a low-coercivity (L) component with a mean coercivity of ∼40 mT and a middle-coercivity (M) component with a mean coercivity of ∼100 mT. In addition, a very low coercivity component (mean coercivity of ∼15 mT) and a high-coercivity component (mean coercivity of ∼0.6 T) are required for optimal fitting. Within-core and intercore variability of the mean coercivity and the dispersion parameter (DP) are listed in Table 2. Differences in the shape of IRM acquisition curves can be explained by varying the relative abundance of the four components, in particular the L and M components, without changing significantly the mean coercivity and DP. This result matches the conclusion of Yamazaki  from North Pacific sediments, which indicates that the L and M components are also the dominant magnetic carriers in equatorial and South Pacific sediments.
Table 2. Within-Core and Intercore Variability of the Mean Coercivity and Dispersion Parameter for the Low- and Middle-Coercivity Components Derived From IRM Acquisition Curvesa
DP, dispersion parameter; L, low-coercivity component; M, middle-coercivity component.
39.2 ± 1.9 (1σ)
0.223 ± 0.005
103.3 ± 5.2
0.271 ± 0.008
36.5 ± 0.6
0.230 ± 0.004
98.7 ± 5.8
0.262 ± 0.005
35.2 ± 0.7
0.219 ± 0.004
96.6 ± 10.2
0.269 ± 0.003
38.3 ± 0.1
0.201 ± 0.006
105.2 ± 2.2
0.259 ± 0.003
 The I/N-I component ratio is roughly proportional to the M/L component ratio (Figure 6b), as was pointed out by Yamazaki  for North Pacific sediments. This suggests that the L component broadly corresponds to the N-I component, and the M component to the I component. However, Ontong-Java Plateau sediments have somewhat lower M/L ratios than North Pacific core NGC65, although both have similar I/N-I ratios. The two major sources of magnetic minerals in Pacific sediments are biogenic and terrigenous. Yamazaki  interpreted the L and N-I components to be carried by biogenic magnetite on the basis of the small DP and χARM/SIRM and coercivity values that correspond to the biogenic soft (BS) component of Egli . The presence of biogenic magnetites in Ontong-Java Plateau sediments is supported by TEM observation of magnetic extracts from core NGC36 [Yamazaki et al., 1995]. The M and I components correspond to terrigenous maghemite.
 There are marked temporal variations of magnetic properties on glacial-interglacial time scales in the North Pacific (Figure 7) and Ontong-Java Plateau cores (Figures 8 and 9). The variations of χARM/SIRM are inversely correlated with changes in the I/N-I ratio, as expected from the relationship observed in Figure 6. Fluctuations of S−0.1T resemble those of the M/L ratio (inverted), which results from the mean coercivity of ∼100 mT for the M component. An exception is that S−0.1T in core NGC88 from the Ontong-Java Plateau has a long-term decreasing trend toward the present that is not accompanied with M/L ratio changes (Figure 9).
 In North Pacific core NGC65, magnetic property variations generally mimic the δ18O curve (Figure 7). In glacial periods, χARM/SIRM and S−0.1T are generally lower and the I/N-I ratio and the M/L ratio are generally higher, which indicates an increased proportion of terrigenous eolian maghemite. This was confirmed by spectral analyses; χARM/SIRM and δ18O variations have significant coherency at the ∼100 ka period with little time lag (Figure 10). Magnetic susceptibility also varies on glacial-interglacial time scales. It tends to be higher in interglacial periods; however, the correlation is not as good as for the former four magnetic parameters (Figure 7). For example, no clear magnetic susceptibility peak occurs within Marine Isotope Stage (MIS) 11 or 13.
 In cores NGC36 and NGC88 from the Ontong-Java Plateau, on the other hand, the χARM/SIRM and S−0.1T minima (maxima in the I/N-I and M/L ratios) occur at glacial-to-interglacial transitions, and do not correspond to either glacials or interglacials (Figures 8 and 9). The spectral analyses show that the variations of χARM/SIRM and δ18O have significant coherency at the ∼100 ka period with a phase angle of about 90°, which corresponds to a time lag of about 25 ka (Figure 10). Peaks or troughs in magnetic susceptibility do not occur consistently at glacial-to-interglacial transitions. Instead, magnetic susceptibility minima roughly correspond to glacial periods, but the correlation is not strong. Compared with the δ18O curve, magnetic susceptibility variations are dominated by short-wavelength features, possibly with orbital precessional periodicity.
 Significant variations in magnetic mineralogy and grain size were not detected using the low-temperature magnetometry. All samples from each core have almost identical curves, regardless of magnetic property changes such as χARM/SIRM ratios (Figure 11). Thermal demagnetization of an IRM acquired at 6 K (curves with the largest magnetization decay with warming) did not reveal an intensity drop at ∼120 K indicative of the Verwey transition for magnetite. However, the presence of magnetite is suggested by the slight irreversibility of low-temperature cycling curves of a 300 K IRM (curves with lower normalized magnetizations); the magnetization during cooling was slightly larger than during warming above ∼100 K. It is inferred that the magnetite has undergone significant oxidation, because partial maghemitization can partially or completely suppress the Verwey transition [Özdemir et al., 1993; Cui et al., 1994]. The relative abundance of ultrafine magnetic minerals can be estimated from the memory ratio: the ratio of the IRM imparted at 6 K to the remanent magnetization remaining after being warmed to 300 K [Hunt et al., 1995]. This reflects unblocking of the remanence carried by magnetic grains that are in a superparamagnetic (SP) state at 300 K but in an SD state at lower temperatures, although continuous unblocking might occur even for grains in an SD state when oxidized. The memory ratio measured on the three cores did not vary significantly throughout the cores. The mean and standard deviation are 0.556 ± 0.009 (n = 16) for core NGC65, 0.544 ± 0.015 (n = 17) for core NGC36, and 0.465 ± 0.027 (n = 20) for core NGC88. This indicates uniform magnetic grain sizes in the sense of unvarying relative abundances of SP grains to SD and larger grains. This suggests that the effect of thermal activation for the difference in the vertical spread of the FORC diagrams [Egli, 2006a] due to grain size variations would not be significant in these cores.
 It is known that in the North Pacific the major source of terrigenous materials is eolian dust, and that the flux of eolian dust was greater in glacial periods [Hovan et al., 1989, 1991]. A rock magnetic study of pelagic red clay [Yamazaki and Ioka, 1997] demonstrated that the S ratio can be an indicator of eolian input. Magnetic property changes in core NGC65 (Figure 7) indicates that the contribution of the terrigenous component (the I and M component) increased in glacial periods, which can be explained by increased eolian dust flux in glacials. It has already been reported that variations in χARM/SIRM and S ratio closely resemble each other in North Pacific sediments [Yamazaki, 1999; Yamazaki and Kanamatsu, 2007], but the explanation for this correlation is not straightforward if the two are independent proxies representing magnetic grain size and mineralogy variations. If we consider that χARM/SIRM mainly reflects magnetostatic interactions, however, the correlation can be explained by postulating that the eolian component contains partially oxidized titanomagnetites with strong within- and/or between-grain magnetostatic interactions. The χARM/SIRM variations are too large to be explained by changes in the strength of magnetostatic interactions that caused only by differences in magnetic mineral concentration [Egli, 2006b], but stronger magnetostatic interactions could occur in grains with ilmenite lamellae of volcanic or plutonic origin [Evans et al., 2006], and/or through formation of grain aggregations during transportation and deposition. In addition to glacial-interglacial changes, a long-term trend can be recognized in the magnetic properties: an upward decrease of χARM/SIRM and S−0.1T and an increase of the I/N-I and M/L ratios. This suggests a long-term increase in eolian dust flux, which is likely to have been caused by increased aridity of the Asian continent and/or by increased intensity of westerlies associated with intensification of the East Asian winter monsoon since 2.6 Ma [An et al., 2001]. A similar long-term increase in eolian dust flux was reported in the eastern Mediterranean Sea, which was attributed to an increased aridity of the Sahara caused by weakening of the African summer monsoon [Larrasoaña et al., 2003].
 In the western equatorial Pacific, it is estimated from magnetic property changes in cores NGC36 and NGC88 that the contribution of the terrigenous component (the I and M component) increased at glacial-to-interglacial transitions. The magnetic property changes are synchronous with fluctuations in carbonate dissolution intensity on the Ontong-Java Plateau [Wu et al., 1991; Zhang et al., 2007]. Carbonate preservation maxima occur at glacial-to-interglacial transitions, which are known as deglacial preservation spikes. The synchronous changes suggest a relationship between terrigenous flux and the carbon cycle although causality is unknown. A possible source of sediments on the Ontong-Java Plateau is terrigenous material that originates from Papua New Guinea and that is transported eastward along the equator by the Equatorial Undercurrent [Lacan and Jeandel, 2001; Ziegler et al., 2008]. Thus, flux of terrigenous material that originated from Papua New Guinea may have increased at glacial-to-interglacial transitions. Eolian flux variations cannot be the cause of the magnetic property changes because at low latitudes eolian dust inputs are much smaller than in the North Pacific [Duce et al., 1991], and because they are not synchronous; the flux was greater in glacial periods in the equatorial Pacific at 140°W as well as in the North Pacific [Anderson et al., 2006]. An alternative possibility for the cause of magnetic property changes may be decreased production of biogenic magnetites at glacial-to-interglacial transitions. Changes in ocean environment that caused the fluctuation of carbonate preservation might also have affected bacterial populations in the surface sediments.
 Temporal variation patterns of carbonate preservation vary within the equatorial Pacific; there may be a time lag in carbonate preservation variations between the Ontong-Java Plateau at ∼160°E and the central equatorial Pacific at 140°W. In the latter region, high CaCO3 contents correspond to glacial periods [Chuey et al., 1987], or to the later part of glacial periods [Farrell and Prell, 1989], rather than to glacial-to-interglacial transitions on the Ontong-Java Plateau. Asian eolian dust represents the majority of the terrigenous component at 140°W [Ziegler et al., 2008], and the flux was greater during glacial periods [Anderson et al., 2006]. Synchronous changes in CaCO3 content and eolian flux at 140°W, though there is a time lag between 140°W and the Ontong-Java Plateau, again suggests a relationship between terrigenous flux and the carbon cycle. The contribution of material from Papua New Guinea would be smaller at 140°W than on the Ontong-Java Plateau because of the greater distance from source, and there is no remarkable difference in its flux between the present interglacial and the last glacial maximum [Ziegler et al., 2008].
 The occurrence of magnetic dissolution during reduction diagenesis has been reported in carbonate sediments on the Ontong-Java Plateau [Tarduno, 1994; Tarduno and Wilkison, 1996]. The Fe redox boundary occurs within several meters of the seafloor, and can be detected through a sudden downward decrease of coercivity (Hc) and coercivity of remanence (Hcr). The depth of the boundary is controlled by organic carbon flux. In cores NGC36 and NGC88, however, no stepwise change in Hc and Hcr was observed (Figure 12). A gradual upward increase in coercivity with minor fluctuations is recognized, which corresponds to observed S−0.1T variations, and is caused by changes in high-coercivity mineral supply. In these cores, iron and sulfate reduction occur below the bottoms of the cores, at 5.2 and 6.2 m for cores NGC36 and NGC88, respectively (Figure S4), so these sediments did not undergo reductive diagenesis. The water depths of cores NGC36 and NGC88 are close to those of ODP Sites 805 and 803 of Tarduno and Wilkison , respectively. The dependence of the Fe redox boundary depth on water depth that they found on the Ontong-Java Plateau predicts the occurrence of Fe redox boundaries at about 3.4 and 4.0 m below the seafloor in cores NGC36 and NGC88, respectively. However, the sedimentation rates for the studied cores are 9 and 8 m/Ma, which are significantly smaller than those of the cores studied by Tarduno and Wilkison , 17 and 10 m/Ma. This difference is probably responsible for the deeper Fe redox boundary in our cores. The variations in magnetic hysteresis parameters that are indicative of reduction diagenesis were observed in other cores taken from the Ontong-Java Plateau by the Geological Survey of Japan, AIST; such cores were not used in this study. The results of low-temperature magnetic measurements also show no evidence for reduction diagenesis. Lack of the Verwey transition indicates that predepositional maghemitization has been preserved and that oxic conditions have been maintained throughout the cores. Reduction diagenesis can enhance the contribution of SP grains [Tarduno, 1995; Rowan et al., 2009], but the rather uniform low-temperature memory ratios suggest that enhancement of SP grains did not occur in these cores.
 For Manihiki Plateau core KR9912-PC5, data points are distributed near the upper left-hand corner on the plot of χARM/SIRM and I/N-I ratios (Figure 6), and a small M component is obtained from the decomposition of IRM acquisition curves. These observations imply that the terrigenous component is small in this area. This is consistent with the fact that the eolian dust flux in the South Pacific is much smaller than in the North Pacific [Duce et al., 1991]. This area, which belongs to the South Pacific Gyre, is far from land, and hence the contribution of noneolian terrigenous material is also estimated to be small. Biogenic magnetite is therefore likely to be a dominant source of magnetic minerals in this area.
 1. The interpretation of Yamazaki  that the χARM/SIRM ratio is dominantly controlled by the strength of magnetostatic interactions has been confirmed for sediments across wide expanses in the Pacific Ocean, and is not limited to the North Pacific. The χARM/SIRM ratio can be used to estimate the relative abundance of terrigenous and biogenic components.
 2. In the North Pacific, the proportions of the interacting to noninteracting SD components and the middle- to low-coercivity components (with mean coercivities of ∼100 mT and ∼40 mT, respectively) increase in glacial periods. Variations of χARM/SIRM and S ratio (S−0.1T) resemble each other, and decrease over time. These variations reflect increases of terrigenous input in glacial periods as eolian dust.
 3. On the Ontong-Java Plateau, χARM/SIRM and S−0.1T minima occur at glacial-to-interglacial transitions, and the relative abundance of the interacting and middle-coercivity components increases in these periods, indicating increased proportions of terrigenous to biogenic components. These are coeval with carbonate preservation maxima in this area, which suggests a possible linkage between the carbon cycle and terrigenous input.
 4. The above mentioned results demonstrate usefulness of rock magnetic proxies for investigating sources of deep-sea sediments and paleoclimatic and paleoceanographic changes in the source and depositional areas.
 I thank Emi Kariya for help with measurements and members of the paleomagnetism laboratory of the Geological Survey of Japan, AIST, for discussion. Constructive comments of reviewers, Mike Jackson and Andrew Roberts; the Editor, John Tarduno; and the Associate Editor, Catherine Kissel, greatly improved the manuscript.