Mineralogy

Hydrogenetic ferromanganese crust consists mainly of vernadite; on the other hand, diagenetic and hydrothermal crust commonly show the presence of birnessite, todorokite, asbolane, and buserite (Cronan, Reference Cronan1999). Powder XRD patterns were collected from the black manganese phases, from the inner part to the outer part of HFMC. These XRD patterns indicate a pure vernadite nanophase with broad diffraction peaks at 2.43, 1.41, and 1.22 Å (Fig. 2). Previous studies determined the crystal structure of synthetic vernadite (δ-MnO₂) (Zhu et al., Reference Zhu, Farrow, Post, Livi, Billinge, Ginder-Vogel and Sparks2012; Manceau et al., Reference Manceau, Marcus, Grangeon, Lanson, Lanson, Gaillot, Skanthakumar and Soderholm2013) and natural vernadite (Lee et al., Reference Lee, Xu, Xu and Sun2019) using synchrotron X-ray diffraction and scattering techniques. A weak and broad (001) peak together with strong (hk) diffraction peaks indicates thin layers of vernadite (i.e. nanosheets or 2D crystals) (Fig. 2). The increasing intensity of the (001) diffraction peak from the outer to inner part suggests the improved crystallinity of vernadite and/or multiple repetitions along the stacking direction or c axis (Fig. 2). The ratio of the d spacing of the first two main diffraction peaks is ~1.723 (i.e. 2.43/1.41), close to $ \sqrt{3} $ , suggesting pseudo-hexagonal symmetry in the manganese octahedral layer.

Figure 2. Powder XRD patterns of the HFMC from the outer part to the inner part, including an image of the sample specimen. The broad (001) diffraction peak indicates thin layer nanophases along the c axis.

HFMC samples were studied through the BET surface area and BJH pore-size distribution analyses to determine the nanopore structures of the vernadite phase (Fig. 3). The multi-point BET surface areas of the inner part, middle part, and outer part were 376, 404, and 421 m² g^–1, respectively. The outer layers of HFMC have slightly larger surface areas than the inner portion, consistent with the outer layers having more poorly crystallized minerals due to their young-stage precipitates. The BJH pore-size distribution analyses revealed that the surface areas were dominated by nanopore surfaces (pore diameter <10 nm) (Fig. 3A). The BET surface area and BJH pore-size distribution analyses show a similar trend with hydrogenetic ferromanganese crusts from other seamounts in the northwest Pacific Ocean (Hein et al., Reference Hein, Koschinsky, Bau, Manheim, Kang and Roberts2000; Hein and Koschinsky, Reference Hein and Koschinsky2014).

Figure 3. Diagrams illustrating nanopore-size distribution determined by Barrett-Joyner-Halenda BJH adsorption (A) and cumulated surface areas (B) for HFMC samples from the inner part to the outer part.

TEM and combined EDS analyses provide information about the structure and chemistry of the vernadite nanophase in the HFMC (Fig. 4). Vernadite appears as aggregates of nanosized 2D crystals and the SAED pattern shows two broad diffraction rings at ~2.4 and ~1.4 Å (Fig. 4A), indicating the low crystallinity of vernadite. Natural vernadite nanophases showed the mixture component of ~7.2 Å vernadite and ~9.6 Å vernadite and interstratified phases are shown in Fig. S4 of the Supplmentary material (Lee et al., Reference Lee, Xu, Xu and Sun2019). The lattice fringes of vernadite are mostly curved and curled (Fig. 4; see also Fig. S4 in the Supplementary material). The curved layers were caused by the presence of vacancies and/or substitutions of other cations in Mn layers, which is associated with a high concentration of defects in the vernadite structure (Zhu et al., Reference Zhu, Farrow, Post, Livi, Billinge, Ginder-Vogel and Sparks2012; Yang et al., Reference Yang, Lee, Post, Xu, Wang, Xu and Zhu2018). TEM-EDS spectra indicate that the vernadite nanophase contains Na, Mg, Si, Cl, K, Ca, Ti, Fe, Ni, and Zn (Fig. 4C,D). Ni and Mg are preferentially enriched in the Mn-rich area, while P is associated with the Fe-bearing area of vernadite nanomineral (Fig. 4C,D).

Figure 4. (A,B) Bright-field TEM images of vernadite nanophases from HFMC. Only the nanocrystals or flakes with (001) plane nearly parallel to the electron beam, show their lattice fringes. Inserted at the lower-left corner of (A) is a SAED pattern from the sample. A representative TEM-EDS spectrum (point analysis) of the relatively Fe-rich area (C) and Mn-rich area (D) of vernadite nanophases. Cu peaks are from the carbon-coated TEM Cu grid that holds the specimen.

The detailed crystal structure of the vernadite nanophase was analyzed using the X-ray PDF technique (Fig. 5). In the PDF pattern, the first major peak at 1.90 Å corresponds to Mn–O correlations in the first shell of the Mn layer. The second peak at 2.86 Å refers to Mn–Mn and O–O correlations in the edge-sharing Mn octahedra. The mixture model of vernadite (Lee et al., Reference Lee, Xu, Xu and Sun2019) is used for the PDF refinement, and the model consists of the 7 Å phase, the 10 Å phase, and the interstratified 7 Å/10 Å phases (Fig. 5; see also Fig. S4 in the Supplementary material). A comparison of the XRD and PDF patterns reveals that the sample does not include any 2-line or 6-line ferrihydrite as mechanical mixture (Fig. 6). If the iron is in the form of 2-line or 6-line ferrhydrite, the XRD pattern (Fig. 6A) will show a broad peak and shoulder at the low-angle side of the 2.43 Å and 1.41 Å peaks. The detection limit of synchrotron XRD can reach up to 0.01% by weight in crystalline mixtures (Thakral et al., Reference Thakral, Zanon, Kelly and Thakral2018). As the TEM-EDS spectra also support the suggestion that Fe is incorporated into natural vernadite nanophases at HFMC (Fig. 4C,D), the Fe-bearing vernadite structure is used for the refinement. Yang et al. (Reference Yang, Park, Son, Baik, Park, Kim, Yoon, Park and Kim2019a) reported the presence of Fe³⁺ in the vernadite from HFMC, as measured by electron energy loss spectroscopy (EELS) analysis.

Figure 5. (A) PDF refinement of Fe-bearing vernadite phase from the outer layer of the HFMC using two phases (7 Å and 10 Å) and three phases (7 Å, 10 Å, and interstratified 7 Å/10 Å) from 1.6–30 Å ranges. The gray lines are differences between experimental (black) and calculated (red) PDF patterns. (B) The calculation of major atomic correlations from the refined vernadite structure (Table 1).

Figure 6. (A) Experimental XRD pattern of vernadite in HFMC compared with natural, 6-line ferrihydrite (6LFh) and 2-line ferrihydrite (2LFh), indicating that the vernadite sample does not have 2- or 6-line ferrihydrite as a mechanical mixture. Peaks and shoulders will be visible at the low-angle side of the 2.43 Å and 1.41 Å peaks in the XRD pattern if the iron-phase is present as 2- or 6-line ferrihydrite. (B) Experimental PDF patterns of vernadite and 6-line ferrihydrite. The vernadite PDF pattern shows valleys instead of peaks where ferrihydrite peaks would typically appear, indicating that the sample does not contain any 2-line or 6-line ferrihydrite as a mechanical mixture. Experimental XRD and PDF pattern of vernadite in HFMC compared with natural 6-line ferrihydrite (6LFh) and 2-line ferrihydrite (2LFh), indicating the vernadite.

Based on the weighted residue (R _w) values with parameters (Fig. 5A), the refined vernadite structure shows an acceptable agreement, and the results are listed in Table 1. Fe is involved in the Mn layer sites of the vernadite structure. We also tested the Fe³⁺ adsorbed on the vacancy sites of the vernadite structure, but it did not improve the goodness of fit. The Fe³⁺ could be incorporated into Mn layers of vernadite or adsorb the triple corner-sharing complex above or below vacancy sites of the octahedral sheet, which is similar to Mn³⁺ incorporated into both sites (Manceau et al., Reference Manceau, Marcus, Grangeon, Lanson, Lanson, Gaillot, Skanthakumar and Soderholm2013). The interlayer cations (i.e. Na⁺, K⁺, Ca²⁺, and Mg²⁺) were tested in PDF refinement and did not improve the fitting of the results. This suggests that this PDF analysis is not sensitive enough to precisely determine the interlayer cations in the vernadite structure (Zhu et al., Reference Zhu, Farrow, Post, Livi, Billinge, Ginder-Vogel and Sparks2012; Lee et al., Reference Lee, Xu, Xu and Sun2019). During the PDF refinement, the mismatched residuals are mainly derived from the 2.87 Å and 3.46 Å peaks (Fig. 5). The mismatch could be caused by local distortions and aperiodic features of manganese octahedral layers (Xu et al., Reference Xu, Hausner, Harrington, Lee, Strongin and Parise2011).

Table 1. The structure of vernadite from two-phase PDF refinement of Fig. 5A

7 Å vernadite: a=4.962 (6) Å, b=2.833 (4) Å, c=7.20 (2) Å, β=97.3 (2)°; 10 Å vernadite: a=4.961 (6) Å, b=2.833 (4) Å, c=9.60 (4) Å, β=98.4 (3)°. U = isotropic displacement parameters, Occ = occupancy, IL = interlayer.

Crystal chemistry

EPMA analyses were performed to determine the major element chemistry and elemental correlations of the HFMC (Fig. 7; see also Table S1 in the Supplementary material). Ca and Na are major elements in the vernadite interlayer and they correlate positively with Mn (Fig. 7A,B). The Fe/Mn ratios in the HFMC range from 0.31 to 0.69 (wt.% of atoms) and are negatively correlated (Fig. 7C). The lack of a link between Ca and P (Fig. 7D) suggests that Ca is not in the form of a phosphorus mineral (e.g. apatite). XRD patterns of HFMC also indicate that no phosphorus mineral is present in the sample. The absence of the phosphorus phases from the studied samples is also shown by the synchrotron XRD patterns of HFMC (Fig. 2). P and Fe have a positive correlation, while Ni and Co have a positive correlation with Mn (Fig. 7E–J). The atomic correlations observed among Mn, Fe, P, Ni, and Co are in agreement with the findings from the TEM-EDS analysis. (Fig. 4C,D). Similar to this, Hein et al. (Reference Hein, Koschinsky, Bau, Manheim, Kang and Roberts2000) observed that Ni and Co are preferentially enriched in the Mn-rich HFMC.

Figure 7. Plots of Na, Ca, Fe, P, Ni, and Co vs Mn and Fe from EMPA analyses of the outer part of HFMC. The full results of the major element analysis are shown in Table S1 in the Supplementary material.

Previous studies have shown the enrichment of REE in ferromanganese crusts from the Magellan Seamount in the northwest Pacific Ocean (Bolton et al., Reference Bolton, Ostwald and Monzier1986; Hein et al., Reference Hein, Koschinsky, Bau, Manheim, Kang and Roberts2000; Jiang et al., Reference Jiang, Lin, Yao and Guo2011; Novikov et al., Reference Novikov, Mel’nikov, Bogdanova and Vikent’ev2014). The total REE from the HFMC sample studied was plotted vs P in Fig. 8. The ΣREE has a positive correlation with P, indicating that phosphate and the REE concentrations are linked in the HFMC. The phosphate is positively correlated with Fe as evidenced by both the TEM-EDS (as shown in Fig. 4C,D) and the EPMA data analysis (Fig. 7D). This implies that Fe-rich vernadite might act as a host layer for the incorporation of phosphate, which in turn could locally enrich REE.

Figure 8. A plot of total REE vs P concentration of the studied HFMC, showing a positive correlation. The data are listed in Table S2 of the Supplementary material, reported by Xue (Reference Xue2007) and Zhang et al. (Reference Zhang, Sun, Xue, Liu, He and Lu2009).

Several previous studies have reported the source and enrichment of PGE in the ferromanganese crusts (Hein et al., Reference Hein, Koschinsky, Bau, Manheim, Kang and Roberts2000; Ren et al., Reference Ren, Glasby, Liu, Shi and Yin2007; Hein and Koschinsky, Reference Hein and Koschinsky2014). The concentration of Pt can be enhanced up to 3000 ppb on the Pacific seamount HFMC. Furthermore, regions exhibiting an abundance of Pt are typically associated with hydrogenetic origins (Ren et al., Reference Ren, Glasby, Liu, Shi and Yin2007). The studied HFMC sample has Pt concentrations of up to 567.9 ppb and 629.26 ppb of ΣPGE, as also reported in previous studies by Sun et al. (Reference Sun, Xue, He, Zhang, Shi, ShengWei and Lu2006) and Xue (Reference Xue2007). Plots of Pt vs Ni and Cu are shown from the inner to the outer side of the HFMC sample from the Magellan Seamount in Fig. 9. ΣPGE and Pt exhibit a positive correlation with Ni and Cu, according to Fig. 9. The correlation indicates that the platinum element exhibits similar behavior to copper and nickel during the formation of the HFMC. Hein et al. (Reference Hein, Koschinsky, Bau, Manheim, Kang and Roberts2000) reported that PGE is correlated with siderophilic Co and Ni. In addition, the middle and inner parts are richer in Pt and ΣPGE (see Table S3 in the Supplementary material), which is likely due to the vernadite structure (i.e. 7 Å vernadite vs 10 Å vernadite, as shown in Fig. S4 of the Supplementary material). The inner part of the HFMC contains a greater quantity of 10 Å vernadite compared with 7 Å vernadite. Given that 10 Å vernadite possesses a larger interlayer gap than 7 Å vernadite, it might be more conducive to the incorporation of PGE (Lee et al., Reference Lee, Xu, Xu and Sun2019).

Figure 9. Plots of the average concentration of total Pt vs Ni and Cu of HFMC samples from the Magellan Seamount. The graph includes 232 data points: 75 in the outer part, 87 in the middle part, and 70 in the inner part. The data are listed in the Table S3 of the Supplementary material, as reported by Xue et al. (Reference Xue, Xiaoming, Gaowen, Shengwei, Hongfeng and Mei2005) and Sun et al. (Reference Sun, Xue, He, Zhang, Shi, ShengWei and Lu2006). Black circle = outer part, red square = middle part, green cross = inner part.

The elemental correlations indicate that REE and P were more likely to be absorbed in the Fe-rich areas, while PGE, Co, and Ni were found to be more concentrated in the Mn-rich areas. These correlations align with previous geochemistry studies conducted on Fe-Mn crusts from the Magellan Seamounts (Glasby, Reference Glasby2006; Novikov and Bogdanova, Reference Novikov and Bogdanova2007; Manceau et al., Reference Manceau, Lanson and Takahashi2014) and Fe-Mn crusts from other localities (Xue, Reference Xue2007; Zhang et al., Reference Zhang, Sun, Xue, Liu, He and Lu2009; Hein and Koschinsky, Reference Hein and Koschinsky2014; Yang et al., Reference Yang, Post, Wang, Xu, Geiss, McCurdy and Zhu2019b). The crystal structure and chemical composition of vernadite in HFMC may be the key factors that control the distribution of minor elements such as P, Ni, and Cu with trace elements (e.g. REE and PGE). Figure 10 consists of a schematic model of vernadite, highlighting the differences between the Fe-rich and Mn-rich areas in the vernadite.

Figure 10. A schematic model for P, Ni, Cu, REE, and PGE enrichment of the vernadite structure between the Fe-rich area and the Mn-rich area. The Mn octahedral sites are shown in purple and the Fe octahedral sites in red.

The elemental correlations might also be influenced by the chemical hardness and softness of the metal cations and their associated ligands (Xu et al., Reference Xu, Xu and Wang2017). The cations of REE are hard Lewis acids, which prefer to bind with a hard base like phosphate anions to form compounds like monazite (Ce, La, Nd)PO₄. The phosphorous tetrahedra at the Fe-bearing Mn layer help to build a strong bond with the REE, which could be crucial for the REE enrichment of HFMC (Fig. 10). The soft Lewis acids Cu, Co, Ni, and Pt cations establish strong covalent connections with the Mn(IV)-dominated octahedral layers of HFMC because Mn(IV) forms covalent bonds with nearby oxygen (Fig. 10). Pt can be adsorbed directly above the Mn octahdedra, similar to Cu in chalcophanite and Cu-adsorbed birnessite (Post and Appleman, Reference Post and Appleman1988; Yang et al., Reference Yang, Park, Son, Baik, Park, Kim, Yoon, Park and Kim2019a).

Paleoclimatic records

The BSE images of a cross-section of the HFMC reveal chemical variations along the direction of growth at a microscopic scale (Fig. 11). The micro-bands display a variety of textures, alternating between columnar and botryoidal formations. Their thickness varies, ranging from less than a micron to several microns. The varied chemical characteristics of the microbands – such as their varying brightness in BSE mode – are connected to elements (e.g. Ni, Zn, Co, Na, Mg, Al, Si, P, S, Cl, K, Ca, and Ti) incorporated in the vernadite nanophases. The bands rich in Mn are linked with Mg and Ni, while the bands rich in Fe are associated with P. This is based on the SEM elemental maps (as shown in Fig. 11D–H) and EDS spectra (see Fig. S5 in the Supplementary material). These findings align with the results from EPMA and TEM-EDS (Figs 4 and 7).

Figure 11. (A–C) The BSE images and (D–H) elemental maps from the studied HFMC sample. A small yellow box at top of the image (A) is the area for X-ray EDS mapping. (B) Enlarged images of the lower part of (A) indicated by a yellow arrow (C–H) showing oscillation of micro-sized layers along the growth direction. The bright layers in BSE images are related to the enrichment of Mn, Ni, and Mg.

The paleoclimatic records were preserved by the HFMC, which crystallized from seawater that had undergone environmental and chemical changes over millions of years (Smith et al., Reference Smith, Staudigel, Watts and Pringle1989; Frank et al., Reference Frank, O’Nions, Hein and Banakar1999; Glasby et al., Reference Glasby, Ren, Shi and Pulyaeva2007; Lee and Xu, Reference Lee and Xu2016). The age and associated geological occurrence of the HFMC from the Magellan Seamounts have been documented (Pulyaeva, Reference Pulyaeva1997; Glasby et al., Reference Glasby, Ren, Shi and Pulyaeva2007; Melnikov and Pletnev, Reference Melnikov and Pletnev2013). Pulyaeva (Reference Pulyaeva1997) and Melnikov and Pletnev (Reference Melnikov and Pletnev2013) recognized three units (I–III), based on the distribution of calcareous microfossils and benthic foraminiferal assemblages: layer I (inner part) in the Late Paleocene–Late Eocene, layer II (middle part) in the Middle Eocene–Late Miocene, and layer III (upper part) in the Pliocene and Quaternary. Three units are also present from the outside to the inner section of our examined HFMC sample (Fig. 1; see also Fig. S2 in the Supplementary material). As the outer part preserved the laminations (Fig. 11), this study focused on them to properly understand paleoclimatic changes.

The Magellan Seamounts HFMC average growth rate ranged from 1 to 3 mm Myr^–1 (Glasby et al., Reference Glasby, Ren, Shi and Pulyaeva2007; Melnikov and Pletnev, Reference Melnikov and Pletnev2013). According to the Xue (Reference Xue2007) study, the outer portion of the HFMC sample that was examined in the study has a growth rate of 1.27 mm Myr^–1. This was determined by correlating the concentration of cobalt with the growth rate. There are various elemental fluctuations observed in the ferromanganese crust. Some areas, such as those depicted in Fig. 12A, maintain their texture from the top layer, making them suitable for paleoclimate research. Figure 12A shows the growth rate in relation to the ages of the HFMC layers. Furthermore, the layers and ages of the HFMC have been correlated with global sea levels, average global temperatures, and CO₂ concentrations dating back to 270,000 y ago. The bright layers (Ni-bearing Mn-rich bands) are highly correlated with the rise in global mean temperature, CO₂ concentrations, and sea level (Fig. 12B). The results suggest that the upper region of the HFMC is very responsive to changes in sea level and global temperature.

Figure 12. (A) The BSE image of the surfacial part of the HFMC sample linked to the ages based on the growth rates (1.27 mm Myr^–1). (B) The global sea level, global mean temperature, and CO₂ concentration for the past 270 kyr (modified from Hansen et al., Reference Hansen, Sato, Kharecha, Russell, Lea and Siddall2007).

The oxygen minimum zone (OMZ) and dissolved oxygen concentration in seawater play a major role in controlling the growth and chemistry of the laminations in HFMC (Frank et al., Reference Frank, O’Nions, Hein and Banakar1999; Hein et al., Reference Hein, Koschinsky, Bau, Manheim, Kang and Roberts2000; Hein et al., Reference Hein, Mizell, Koschinsky and Conrad2013). In the OMZ, where it crosses with the oxidation of dissolved Mn²⁺ and oxygen-rich Antarctic water, a poorly crystalline vernadite phase is more likely to occur (Hein et al., Reference Hein, Mizell, Koschinsky and Conrad2013; Yang et al., 2019). The chemistry of the layers is associated with the redox shifts, or variations in the concentration of dissolved oxygen, which occur during the development of the HFMC. The rhythmic behavior of growth patterns suggests changes in the regional oceanographic setting, including the oxygen minimum zone and sea level, according to Marino et al. (Reference Marino, González, Lunar, Reyes, Medialdea, Castillo-Carrión, Bellido and Somoza2018). The Mn-rich vernadite bands (higher Mn/Fe ratio) originate in relatively oxidizing (more dissolved oxygen) environments, in contrast to the Fe- and P-bearing vernadite bands, which are formed in reducing environments (lower productivity of Mn(IV) oxide). Lu et al. (Reference Lu, Ling, Zhou, Jiang, Chen and Zhou2005) reported on the chemical variations of layers of the Fe-Mn crust from the Central North Pacific region and found the drop in the Mn/Fe ratio linked with significant cooling events and a decrease in sea level.

The majority of earlier studies have revealed large-scale paleoclimatic periodic variation linked to the HFMC. Hein et al. (Reference Hein, Bohrson, Schulz, Noble and Clague1992) found an association between composition variations with a million-year periodicity and changes in the ocean circulation pattern. Han et al. (Reference Han, Jin, Yang, Fietzke and Eisenhauer2003), Ren et al. (Reference Ren, Glasby, Liu, Shi and Yin2007), and Asavin et al. (Reference Asavin, Daryin and Melnikov2015) noted rhythmic variations in the Fe-Mn crust in the Western Pacific, which suggested that the 0.2–1.0 Myr cycles of chemical variations were controlled by Milankovitch cycles related to orbital eccentricity, obliquity, and precession, respectively. The cross-section of HFMC exhibits extremely small-scale compositional oscillations (sub-micron layers), which are related to millennial-scale paleoclimatic oscillations that took place during the Middle–Late Pleistocene and Holocene periods, including a glacial termination event (~126 kyr ago). These oscillations are visible in our high-resolution BSE images (Fig. 13). The observed 2 μm bands indicate cycles of ~1600 y, according to the growth rate (1.27 mm Myr^–1) derived from cobalt content (Fig. 13). Another HFMC sample (similar depth) that was taken from the Magellan Seamount regions similarly had fine-scale compositional oscillations in the outer section (same depth) (see Figs S1 and S6 in the Supplementary material).

Figure 13. High-resolution BSE image of HFMC with intensity profile from α to β, indicating the sub-micron and micron-sized oscillating bands. (A) The area is a magnification of the yellow box in Fig. 12A. (B) The profile shows the ~2 μm intervals, indicating cycles of ~1600 y based on the growth rate (1.27 mm Myr^–1). The Ni-rich layer (corresponding to a glacial termination at ~126 kyr BP) is indicated by a green arrow.

Millennial-scale rapid climate changes for the Pleistocene and Holocene epochs have been widely reported in sediments, ice cores, and submarine anoxia (Dansgaard et al., Reference Dansgaard, Johnsen, Clausen, Dahl-Jensen, Gundestrup, Hammer, Hvidberg, Steffensen, Sveinbjörnsdottir and Jouzel1993; Bond et al., Reference Bond, Showers, Cheseby, Lotti, Almasi, DeMenocal, Priore, Cullen, Hajdas and Bonani1997; Hyodo et al., Reference Hyodo, Bradák, Okada, Katoh, Kitaba, Dettman, Hayashi, Kumazawa, Hirose and Kazaoka2017). There was also a series of rapid millennial-scale climatic variations during the last glacial era, according to Turney et al.’s (Reference Turney, Kershaw, Clemens, Branch, Moss and Fifield2004) interpretation of peat humification investigations. In the deep-ocean environment, rhythmically interbedded limestone and marl layers revealed millennial-scale climatic oscillations (<2100 y) (Elrick and Hinnov, Reference Elrick and Hinnov1996). Despite the characteristics being observed in many marine sediments, glacial deposits, and terrestrial deposits, the causes of such millennial-scale climate variability are still not fully understood (Elrick and Hinnov, Reference Elrick and Hinnov1996; Clark et al., Reference Clark, Webb and Keigwin1999). This is due to limitations in the resolution of the chronology, poor preservation, and additional variations, including those caused by local climatic, oceanographic, tectonic, geographic, and biological factors. Several studies suggest that the millennial paleoclimatic changes are mainly caused by ice sheet melting modulated by solar activity contributing to the global climate system (Elrick and Hinnov, Reference Elrick and Hinnov1996; Bond et al., Reference Bond, Showers, Cheseby, Lotti, Almasi, DeMenocal, Priore, Cullen, Hajdas and Bonani1997; Turney et al., Reference Turney, Kershaw, Clemens, Branch, Moss and Fifield2004; Wang et al., Reference Wang, Cheng, Edwards, He, Kong, An, Wu, Kelly, Dykoski and Li2005; Hyodo et al., Reference Hyodo, Bradák, Okada, Katoh, Kitaba, Dettman, Hayashi, Kumazawa, Hirose and Kazaoka2017). Ice-sheet growth and decay may be related to fine-scale paleoclimatic variations found in our examined HFMC samples (Figs 11–13). The slow growth rates and HFMC sensitivity to variations in sea level are key factors in recording fine-scale paleoclimatic changes. More ferromanganese crusts from other sites, depths, and sources need to be examined in future studies in order to understand the fine-scale paleoclimate variations.