A Sediment Magnetic Record in the North Pacific Across the Mid-Pleistocene Transition and its Implication on Asian Dust Evolution

Eolian dust deposited in the North Pacific is an important archive of the evolutionary history of Asian interior source regions and climate system. Here, we present a ∼1 Myr sediment magnetic record from the central North Pacific to characterize eolian dust properties since the middle Pleistocene. For the studied sediments, magnetic components are mainly identified as biogenic magnetite and detrital magnetic minerals (dust and volcanic origins) based on coercivity analysis, microscopic observations, and sedimentological information. The detrital magnetic component is characterized by high coercivity (>100 mT) and shows a long-term increase in concentration since ∼1 Ma. In particular, the concentration shows a considerable increase at ∼0.8–0.7 Ma compared to the inorganic silicate fraction, indicative of magnetic mineral enrichment in detrital sediment fraction. At the same time, the coercivity distribution of the detrital component also decreases, which can be attributed to an increase in the ferrimagnetic mineral contribution. As the detrital sediments are primarily wind-blown particles, such ferrimagnetic enrichment implies a change in dust source materials after ∼0.8 Ma, which could be explained by the reorganization of atmospheric circulation and/or regional aridification in source regions across the mid-Pleistocene transition. The dust property change in source areas is likely to be synchronized across the North Pacific based on the similarity of the long-term trend of magnetic signals.


INTRODUCTION
In the North Pacific, eolian dust transported from inland Asia via westerly winds is the major terrigenous constituent in deep-sea sediments. During the late Cenozoic, the eolian dust deposition in the North Pacific has gradually increased, which has been suggested to be associated with increased dust production and transportation during the growth of the northern Hemisphere ice sheet and global cooling (Rea et al., 1998;Zhang et al., 2016). Such dust evolution can be induced by extreme environmental changes in source regions, such as through the reorganization of atmospheric circulation and aridity/weathering changes (Nie et al., 2018;Zhang et al., 2019;Abell et al., 2021), and thus, these processes fundamentally involve changes in mineralogical and chemical composition of eolian dust.
Marine sedimentary magnetic signal is a simple and useful tool for reconstructing past changes in eolian dust properties. In particular, high coercivity magnetic fractions have been widely used for semi-quantitative estimation of eolian dust content because hematite is a common weathering product in arid source regions (Doh et al., 1988;Larrasoaña et al., 2015;Zhang et al., 2018). However, the magnetic signals of eolian dust fraction in the North Pacific are still not fully understood. Although ferrimagnetic minerals (e.g., magnetite/maghemite) predominantly account for magnetic signals of North Pacific sediments, it is difficult to isolate them as Asian dust records due to post-depositional alteration and the addition of these minerals from other sources (i.e., biogenic magnetite and volcanic materials).
In this study, variations of detrital and biogenic magnetic signals were reconstructed for the last 1 million years from the central North Pacific, where the eolian dust fraction dominates the detrital sediments. The mid-Pleistocene transition (MPT) represents a period of major climate transition characterized by glacialinterglacial periodicity change from 41 to 100 kyr cycles due to ice sheet development and global cooling (1.2-0.7 Ma; Clark et al., 2006), and in situ loess/dust records have been vigorously studied to identify the climate evolution in the dust source regions during the MPT (Heslop et al., 2002;Zan et al., 2013;Tan et al., 2020). However, the North Pacific dust record during this period has been poorly investigated, although it can improve our understanding of the regional source evolution in Asia and dust transportation. Here, we identify variations in the mineral magnetic properties of detrital components from North Pacific sediments and evaluate the possible evolution of dust source materials across the MPT.

MATERIALS AND METHODS
Samples were taken from the sediment core NPGP1401-2A (32.03°N, 178.59°E; water depth of 5,205 m; Figure 1) around the Hess Rise in the central North Pacific. The core is ∼7 m long and dominantly composed of homogenous brown clay. In this study, 312 discrete wet samples were taken using 7 cm 3 paleomagnetic cubes along the core depth, and freeze-dried sediments were also prepared at selected depths. Sediment ages were obtained from a modified age model using magnetic reversal stratigraphy and relative paleointensity correlations after Shin et al. (2019), which provided paleomagnetic data with stratigraphic correlations using rock magnetic data (Supplementary Figure S1). The bottom age and average sedimentation rate are approximately 1.1 Ma and 0.67 cm/ kyr, respectively.
Frontiers in Earth Science | www.frontiersin.org November 2021 | Volume 9 | Article 789584 4-cm depth interval. The extraction method and calculation were followed Hovan (1995): ISF content (wt%) weight of extracted ISF/weight of bulk sediment × 2.5 × 100%. First-order reversal curves (FORCs) and stepwise IRM measurements were also conducted for selected dry samples using a Princeton Measurement Corporation vibrating sample magnetometer (MicroMag 3900 VSM) at Korea Institute of Geoscience and Mineral resources. 156 FORCs were measured with a field increment of 1.5 mT and a maximum applied field of 1 T. The FORCs were smoothed using the FORCinel 3.0 processing software (Harrison and Feinberg, 2008) with the VARIFORC built-in function (Egli, 2013). The processed data were presented on FORC diagrams with B u (interaction field) and B c (coercivity) axes. IRM acquisition curves were obtained with 120 nonlinear steps up to a maximum applied field of 1 T. From the curves, IRM components were decomposed through their additive coercivity spectra (Kruiver et al., 2001).
Scanning electron microscopy (SEM) and transmission electron microscopy (TEM) observations were carried out on the magnetic mineral extracts for selected samples. The magnetic extracts mounted on a carbon tape were coated with carbon, and SEM observations were performed using a JEOL analytical Feld Emission (FE)-SEM (JSM-7610F) coupled with energy dispersive X-ray spectroscopy (EDS) at Gyeongsang National University, South Korea. For TEM experiments, the magnetic grains were deposited onto carbon-coated copper grids, and observations were made using a JEOL FE-TEM (JEM-2010F) equipped with EDS at Pukyong National University, South Korea.

Sediment Magnetism
The temporal variations of all the concentration-related magnetic parameters for the last ∼1 Myr are presented in Figure 2. Of these, χ and SIRM, representing total magnetic mineral concentration, commonly show cyclic fluctuations at glacial-interglacial timescales with higher values during interglacials ( Figures  2A,B). The cyclicity is not distinct in older parts (approximately >0.8 Ma) probably due to the smoothing effect at a lower sedimentation rate (Supplementary Figure S1). Over a long-term timescale, however, SIRM reveals distinct increases FIGURE 3 | IRM unmixing results and FORC diagrams for selected samples and temporal variation of IRM components. (A) IRM coercivity spectra and unmixing results; (B) FORC diagrams, and (C) comparison between SIRM and absolute concentration of IRM components. IRM components (Component 1-4) are presented as purple, green, blue, and red lines, respectively. The sum of components delineates the best-fitted coercivity spectra. Absolute IRM values (IRM percentage contribution (%) of component × SIRM) for the major components (Component 2 and 3) colored by yellow and orange, respectively.
Frontiers in Earth Science | www.frontiersin.org November 2021 | Volume 9 | Article 789584 4 since ∼0.8 Ma across the MPT, while χ fluctuates within a relatively uniform range. Unlike χ, which is affected by all magnetic components, including paramagnetic minerals, SIRM reflects the concentration of remanence carrying magnetic minerals. Hence, the different long-term trends between χ, and SIRM could indicate changes in magnetic composition after ∼0.8 Ma. As a record, χ fd% remains relatively constant and low values ranging 3-7% ( Figure 2F), indicating a weak grain-size effect on χ (Dearing et al., 1996).
To diagnose the variation of magnetic mineral concentration, SIRM was separated into lower and higher coercivity fractions on the criterion of 100 mT (IRM soft and HIRM 100 , respectively). IRM soft accounts for ∼70-85% of SIRM and displays distinctive glacial-interglacial changes without a long-term increasing trend ( Figure 2C). On the other hand, HIRM 100 , comprising ∼15-30% of SIRM, shows a clear long-term increasing trend with diminished cyclic short-term variation ( Figure 2D). Notably, HIRM 100 increases gradually over the last 1 Myr with a dramatic increase across the MPT between 0.9 and 0.7 Ma and then a gradual decrease until ∼0.5 Ma and increase afterward. The variation of IRM soft percentage in SIRM is displayed as S 100 , which ranges at 0.7-0.85 ( Figure 2E). The S 100 shows glacial decreases and a long-term decrease, which reflects the relative decrease/increase of the low/high coercivity fractions, as shown by IRM soft and HIRM 100 , respectively. All these results indicate that low and high coercivity fractions account for the short and long-term changes in magnetic signals, respectively.

Unmixing of Magnetic Components
IRM component analysis enables the isolation of different magnetic coercivity components. In all samples, four coercivity components are commonly identified ( Figure 3A). Component 1 exhibits a low coercivity fraction with skewed coercivity distribution, probably due to the thermal relaxation effect of SP grains (Heslop et al., 2004;Zhang et al., 2018). Component 2 is characterized by a narrow dispersion (∼0.2) at a mean coercivity (B h ) of ∼35 mT, corresponding to biogenic magnetite (Egli, 2004). Component 3 is recognized as detrital ferrimagnetic grains (e.g., magnetite and maghemite) showing a peak dispersion at B h of ∼100 mT. Component 4 at B h of ∼260 mT indicates detrital antiferromagnetic minerals, such as hematite and goethite. Components 2 and 3 are the major contributors of SIRM, accounting for approximately 44 and 34%, respectively. FORC diagrams display distinct central ridge along the B u 0 axis and surrounding triangular spread ( Figure 3B), which confirms the main contribution from non-interacting single-domain (SD) biogenic magnetite and vortex state detrital grains, respectively, (Roberts et al., 2000. The relative contribution of biogenic and detrital magnetic components shows temporal changes in both IRM and FORC distributions ( Figures 3A,B). To estimate the concentration changes, the absolute IRMs of Components 2-4 were calculated as IRM percentage contribution (%) of component × SIRM, and the IRM variations are presented in Figure 3C; Supplementary Figure S2. For the major components, Component 2 generally shows glacialinterglacial fluctuations, resembling IRM soft , while Component 3 has a variation analogous to that of HIRM 100 (Figures 2C,D). Component 4 accounts for only a small percentages (3-7%) of overall SIRM variation owing to the weak magnetization of antiferromagnetic minerals. This component shows a gradual long-term increase in concentration over the last 1 Myr, which is roughly similar with HIRM 300 (high coercivity fraction of >300 mT) variation (Supplementary Figure S2). However, compared to Component 3, Component 4 exhibits a less pronounced increase around ∼0.8 Ma ( Figure 3C; Supplementary Figure  S2). Such magnetic unmixing results confirm that low coercivity biogenic magnetite and high coercivity detrital ferrimagnetic minerals mainly control the bulk magnetic variations at glacial-interglacial and longer-term timescales, respectively.

Electron Microscope Observation
From SEM observations of the magnetic extracts for selected samples, magnetic grains of various sizes and morphologies were observed ( Figure 4A). On a scale of tens of microns (generally >10 μm), the grains often have an angular, irregular, and vesicular texture and show magnetite inclusions ( Figure 4B), which corresponds to the characteristics of volcanic ashes reported in the North Pacific (Shin et al., 2020). Smaller grains are dominated by rounded and subrounded (titano) magnetite grains, representing reworked and abraded eolian particles ( Figure 4C).
To confirm the presence of biogenic magnetite, TEM observations were made on magnetic extracts of four samples at glacial and interglacial intervals. From all selected samples, biogenic magnetite was mostly identified as equant type in morphology and bullet and elongated prismatic types occasionally with a size of ∼50 nm ( Figures 4E,F). The dominance of equant-type biogenic magnetite is consistent with our IRM unmixing results that showed the low coercivity peak (∼35 mT) of the biogenic component (i.e., soft biogenic magnetite; Egli, 2004) in all samples ( Figure 3A).

Synchronous Magnetic Signals in the North Pacific
Our results indicate that sedimentary magnetic signals come from biogenic and detrital components, showing different temporal variation patterns (Figure 3). In pelagic sediments, the relative variation of the two magnetic components can be indicated by χ ARM /SIRM, because χ ARM sensitively responds to SD grains with weak magnetostatic interactions, such as biogenic magnetite (Egli, 2004;Yamazaki, 2008). That is, higher/lower ratios are equivalent to more biogenic/detrital contributions, respectively. For the studied sediments, χ ARM / SIRM exhibits distinct glacial-interglacial fluctuations and a Frontiers in Earth Science | www.frontiersin.org November 2021 | Volume 9 | Article 789584 long-term decreasing trend ( Figure 5B), as shown in biogenic and detrital IRM components (i.e., Components 2 and 3), respectively, ( Figure 3C). In the older part (<0.6 Ma), the glacial-interglacial variability of χ ARM /SIRM is more evident when compared directly to the LR04 record than to the alternating marine isotope stages shaded in gray ( Figures  5A,B). The IRM ratio of the two components (Component 2/ Component 3) shows similar variation to χ ARM /SIRM, confirming that the χ ARM /SIRM variation reflects their relative contribution ( Figure 5B). As Component 3 has a coercivity of ∼100 mT, S 100 also exhibits similar variations, reflecting the relative abundance of biogenic and detrital components ( Figure 2E) (e.g., Yamazaki, 2009). The glacial-interglacial fluctuation of biogenic magnetite can be associated with several environmental factors. In the northwest Pacific, it has been proposed that magnetic minerals dissolve under reducing conditions during some glacial periods due to climate-related cyclic redox changes at glacial-interglacial timescales (Korff et al., 2016). The intervals for magnetic mineral dissolution are identified as abrupt decreases in ARM/SIRM (marked as red lines in Figure 5C) because fine-grained magnetite (e.g., biogenic magnetite) dissolves preferentially due to its large surface-to-volume ratio (Karlin and Levi, 1983). Our χ ARM /SIRM displays similar glacial decreases but does not show prominent drops throughout the studied period as in the SO202-39-3 core ( Figure 5C), indicative of weak or little dissolution effect at our study site. From the TEM observations, we could not find any direct evidence on diagenetic alteration of biogenic magnetite (e.g., partially etched grains; Yamazaki, 2020) in all the samples (Figure 4). Thus, we interpret that most of magnetic grains in our sediments, including biogenic magnetite, have not been subjected to reductive dissolution. Prior to the postdepositional alteration, the production of biogenic magnetite may have controlled its concentration on glacial-interglacial timescales, as it has been suggested to respond to climate factors, such as iron and organic carbon supply (Roberts et al., 2011). However, the biogenic magnetite concentration could also be sensitive to changes in the sedimentation rate changes. For example, enhanced eolian dust input during glacials may lower the apparent concentration of biogenic magnetite (Yamazaki, 2009). Although the cause of the biogenic magnetite signal is not conclusive with the presented data, further detailed studies on biogenic magnetite flux would be helpful to resolve this.
Despite some ambiguity about the cause of glacialinterglacial variation, χ ARM /SIRM (or ARM/SIRM) variations compiled in the North Pacific mimic each other since ∼1 Ma, indicating simultaneous long-term variation of detrital and biogenic magnetic signals (Figures 5B,C). In detail, the χ ARM /SIRM has a large valley within the time span of ∼0.8-0.5 Ma centered around 0.7 Ma, reflecting the increased detrital contribution during this time period ( Figure 5B). Such temporal features match well with the variation of HIRM 100 , showing only detrital contribution (Figures 3C, 5B). Such similarity strongly suggests that the increased detrital magnetic contribution during 0.8-0.5 Ma is synchronized regionally in the North Pacific.

Long-Term Variation of Detrital Magnetic Signals Since 1 Ma
As detrital magnetic components in the studied sediments are characterized by high coercivity, mainly of ∼100 mT (Figure 3), HIRM 100 can be used as a proxy for the concentration of detrital magnetic minerals. Such detrital magnetic signal in bulk sediments is generally recognized as terrigenous contributions unless there is significant source change or alteration of magnetic minerals (Yamazaki, 2009). However, for the studied sediments, HIRM 100 and ISF show different long-term trends ( Figures  5D,I); HIRM 100 shows a striking increase during 0.8-0.7 Ma, while the ISF content shows a stable high value (∼60%) in older (>0.7 Ma) sediments. After then, they show comparable variations with a decrease at ∼0.5 Ma. To highlight the differences between the detrital magnetic fraction and ISF, dilution effects on HIRM 100 by the non-terrigenous fraction were corrected: HIRM corr HIRM 100 /(100-water content %)/ISF% ( Figure 5E). The HIRM corr shows a distinct increase after ∼0.8 Ma, indicating that magnetic minerals have been gradually enriched in detrital sediments.
We also assessed the coercivity distribution of the detrital magnetic fraction (i.e., high coercivity fraction) using the L-ratio (Liu et al., 2007). The L-ratio shows a dramatic decrease during ∼0.8-0.7 Ma, along with increasing HIRM 100 , and a gentle increase at ∼0.5 Ma ( Figure 5F). Liu et al. (2007) suggested that the L-ratio is indicative of the coercivity distribution of hematite when a hard magnetic fraction (>100 mT) is carried by hematite, but the studied sediments cannot be interpreted as such because the hard fraction is dominantly contributed by ferrimagnetic minerals of ∼100 mT (e.g., magnetite/maghemite; Figure 3). In this study, the L-ratio decrease reflects the greater input of the ∼100 mT materials rather than antiferromagnets of ∼300 mT, which can be confirmed by IRM Component 3 showing a more pronounced increase in concentration since ∼0.8 Ma compared to Component 4 ( Figure 3C; Supplementary Figure S2). Thus, the concurrent (opposite) variation between the L-ratio and HIRM 100 indicates relative softening and enrichment of detrital magnetic minerals due to ferrimagnetic input over the last ∼1 Myr, especially after ∼0.8 Ma.
In summary, the concentration of magnetic minerals in the detrital sediment fraction significantly increases after ∼0.8 Ma, which is also accompanied by a shift in the coercivity distribution (i.e., relative magnetic softening). Based on the coercivity component analysis, we attribute such magnetic changes to ferrimagnetic mineral enrichment. Interestingly, such ferrimagnetic enrichment accounts for the long-term increase of SIRM since ∼0.8 Ma ( Figure 2B), whereas χ fluctuates within a relatively uniform range (Figure 2A). Considering the relatively high ISF content in the lower part of the core ( Figure 5I), the difference may be due to mineralogical change affecting χ signals, such as contribution of non-ferromagnetic materials (e.g., paramagnetic minerals). Therefore, the increase in ferrimagnetic signals could be reflected better by SIRM than by χ in the study area. Nevertheless, as shown by similar trends in χ ARM /SIRM (or ARM/SIRM) records ( Figures 5B,C), the timing of this notable magnetic change (i.e., ∼0.8 Ma) is likely synchronized across the North Pacific.

Eolian Dust Evolution Across the MPT
For North Pacific sediments, an increase in the high coercivity fraction in the bulk sediment can be attributed to enhanced dust input from inland Asia during cold periods because eolian dust is the main terrigenous constituent (Doh et al., 1988;Yamazaki and Ioka, 1997). In this study, however, HIRM 100 reflects magnetic mineralogical change in the detrital fraction as well as its quantitative change, taking into account the increase of HIRM corr and decrease in L-ratio, especially at ∼0.8-0.7 Ma. Such magnetic mineralogical changes in the studied sediments implies changes in the source materials.
We can first consider the supply of ferrimagnetic minerals from volcanic materials. In the North Pacific, terrigenous sediments are generally a binary mixture of Asian dust and volcanic materials supplied from nearby arc regions (e.g., Japan and Kamchatka) (Pettke et al., 2000). Recent magnetic studies have shown that volcanic materials are characterized by abundant high coercivity ferrimagnetic minerals as well as strong magnetic signals (Zhang et al., 2018;Shin et al., 2020). However, the volcanic contribution is indicated as low at ∼10% for the studied sediments in the Sr-Nd isotope compositions and grain size analysis (Lee et al., 2019). End-member (EM) analysis of grain-size distribution reveals two main contributors (EM1, mode size at 3.5 μm; EM2, mode size at 11 μm), probably corresponding to fine dust and coarse volcanic particles, respectively, (Supplementary Figure S4; Lee et al., 2019). Our SEM observation of magnetic extracts agrees that volcanic particles mainly exist in a few tens of micron in size ( Figure 4B). The volcanic component (EM2) shows only minor temporal changes until 0.6 Ma ( Figure 5H). These sedimentological data confirm that our detrital magnetic signals are primarily of eolian origin, and the volcanic contribution is negligible with respect to temporal variation. Therefore, it is reasonable to conclude that the magnetic mineralogical change after ∼0.8 Ma is due to changes in dust source materials, including their provenance.
The isotope compositions of detrital sediments are a useful indicator of provenances variability in Asian dust records over the past millions of years (Chen and Li, 2013). In the studied sediments, the Sr-Nd isotopic compositions are relatively invariant and designate central Asian deserts (e.g., the Taklimakan desert) as the main provenance ( Figure 5G; Pettke et al., 2002;Lee et al., 2019). However, a small increase in ε Nd values after ∼0.8 Ma ( Figure 5G), together with the magnetic enrichment ( Figure 5E), could support changes in dust contribution from different sources. For example, global cooling during the MPT could have changed the dust transport pathway due to climate-related latitudinal shifts of westerlies (see Figure 1; Schiemann et al., 2009;Nagashima et al., 2011). As observed in the present spring conditions, meridional dispersion Frontiers in Earth Science | www.frontiersin.org November 2021 | Volume 9 | Article 789584 of Westerly jet path could occur during cold periods together with a southward shift of the main jet axis (Schiemann et al., 2009;Abell et al., 2021). In such Westerly jet condition, dust supply may have been enhanced from the high-latitude deserts on the Central Asia orogenic belt (e.g., the Mongolian Gobi), which have relatively high ε Nd values (Zhao et al., 2014;Lee et al., 2019). Another potential mechanism for changes in dust source materials is regional aridification and desertification in the Asian interior during the MPT (Rea et al., 1998). The extensively exposed drylands of Central Asia, where isotopic compositions are similar (Chen et al., 2007;Zhao et al., 2014), may have supplied dust with a slightly increased ε Nd composition and more abundant magnetic minerals. In addition, the relative enrichment of ferrimagnetic component could also be associated with a decrease in chemical weathering intensity in various source regions during intensified aridification (e.g., Deng et al., 2006;Zan et al., 2010). Taken together, our magnetic data suggest changes in dust sources in the central North Pacific at ∼0.8 Ma, but it is not possible to present conclusive sedimentological or geochemical evidence in this study to support our conclusions derived from magnetic data. The source change had been likely caused by the reorganization of atmospheric circulation and/or regional aridification changes associated with climate changes during the MPT, which need to be elucidated with further investigation on magnetic mineral distributions in various source regions of Asia.

CONCLUSION
To identify the climate dependence of eolian dust properties, sediment magnetic signals for the last ∼1 Myr were investigated in the central North Pacific. As a result of the IRM component and FORC analyses, the bulk magnetic signals are decoupled to biogenic and detrital origins showing lower and higher coercivity than 100 mT, respectively. SEM and TEM observations confirm the presence of biogenic magnetite and detrital magnetic grains (eolian dust and volcanic particles). The biogenic and detrital components mainly present glacialinterglacial cyclicity and long-term variation in their concentrations, respectively. Based on similar long-term trends in χ ARM /SIRM (or ARM/SIRM) records, variations in biogenic and detrital magnetic components are likely synchronized in the North Pacific. As a proxy of the concentration of the detrital magnetic component, HIRM 100 shows a long-term increase since 1 Ma, with a significant increase at ∼0.8-0.7 Ma. As ISF shows relatively low content after ∼0.7 Ma, the HIRM 100 increase at ∼0.8-0.7 Ma reflects magnetic mineral enrichment in detrital sediments. In this period, the L-ratio also concurrently decreases with the HIRM 100 increase, indicating relative softening of detrital magnetic minerals (i.e., relative abundance of ferrimagnetic minerals). Combined with sedimentological information, such ferrimagnetic enrichment after ∼0.8 Ma suggests a change in dust source materials because the contribution of volcanic materials is generally low and negligible. The source change could be induced by the reorganization of atmospheric circulation and/or enhanced aridification in the Asia interior across the MPT.

DATA AVAILABILITY STATEMENT
The original contributions presented in the study are included in the article/Supplementary Material, further inquiries can be directed to the corresponding author.