Abyssal Manganese Nodule Recording of Global Cooling and Tibetan Plateau Uplift Impacts on Asian Aridification

The impact of central Asian aridification on the low latitude North Pacific Ocean since the late Miocene remains unclear. To address this question, we systematically studied an abyssal manganese nodule from the northwestern Pacific Ocean, which is expected to be sensitive to eolian dust sourced from central Asia. Geochemical variations and the fossilized remains of magnetotactic bacteria within the studied nodule manifest two prominent Asian aridification events at ∼8–7 Ma and 3.6–0 Ma. These results suggest that central Asian aridification impacted both primary productivity and abyssal microbial activity in the NW Pacific Ocean via eolian dust inputs. In contrast to the Pliocene aridification event, the late Miocene event was associated with a primary productivity bloom that is not evident in coeval global primary productivity records, which indicates that the ∼8–7 Asian aridification event was likely due to NE Tibetan Plateau uplift rather than to global cooling.

be sensitive to eolian dust variations and to bottom current evolution (e.g., Jiang et al., 2021;Jones et al., 2000;Ling et al., 2005;Liu et al., 2020). Furthermore, manganese nodules can provide viable micro-environments for deep ocean microorganisms (Blöthe et al., 2015), as evidenced by abundant living microbial populations inside nodules (e.g., Blöthe et al., 2015;Shiraishi et al., 2016). Recently, magnetic remains of magnetotactic bacteria (MTB), known as magnetofossils (Chang & Kirschvink, 1989), have been reported to be the dominant magnetic component in manganese crusts and nodules (Jiang et al., 2020;Oda et al., 2018;Yuan et al., 2020). Therefore, NW Pacific manganese nodules may contain long-term records of Asian aridification and its impact on both primary productivity and abyssal microbial activity. Here, we investigate a NW Pacific hydrogenous manganese nodule to assess climatic imprints on its geochemical, isotopic, and magnetic records and to clarify the driving forces of central Asian aridification over the past ∼8 Ma.

Material and Methods
The manganese nodule studied here was collected from the NW Pacific Ocean (Figure 1a) at site 152°33.38′ E, water depth 5,583 m) by the research vessel Hai Yang Liu Hao in 2016. The nodule is 5.5 cm in diameter with concentric patterns in cross-section. A total of 17 slices (∼10 × 8 × 1.5 mm) were made by cutting with a low-speed diamond wire saw from surface to kernel in upper side of nodule ( Figure S1 in Supporting Information S1).

Magnetic Measurements and Microscopic Observation
To characterize the rock magnetic properties of the manganese nodule, 100 first-order reversal curves (FORCs; Pike et al., 1999) were measured for each sample with 300 ms averaging time using a Lakeshore Cryotronics vibrating sample magnetometer (VSM 8600) at the Southern University of Science and Technology (SUSTech), Shenzhen, China. An anhysteretic remanent magnetization (ARM) was imparted in a 100 mT peak alternating field with a superimposed 50 μT direct bias field using a Dtech D2000 instrument and was measured with an AGICO JR-6A spinner magnetometer. Green lines represent the pathway of the lower circumpolar deep water (LCDW) (Rella & Uchida, 2015). Red and black dashed lines indicate the positions of the boreal summer and winter intertropical convergence zone, respectively. (b) The data distribution is indicative of a hydrogenetic origin (Bonatti et al., 1972) (c) Age model for the studied nodule (red lines indicate the 90% credible interval, std = standard deviation), as well as 1,000 samples drawn from the posterior (blue lines).

10.1029/2021GL096624
3 of 10 After completion of bulk magnetic measurements, samples were powdered with a mortar and pestle. Magnetic particles were then extracted from the powders using the method of Li et al. (2012). Magnetic extracts were dispersed in ethanol; a small drop of the suspension (2 μl) was then dried on a double-layer carbon-coated copper grid for transmission electron microscope (TEM) observations. Samples were examined using a JEOL JEM-2100 TEM operated at 200 keV at the Institute of Geology and Geophysics, Chinese Academy of Sciences, Beijing, China.

Geochemical Measurements
Powders (200-mesh size) of nodule slices made after completion of bulk magnetic measurements were subjected to the following geochemical measurements. The bulk major element composition was measured using X-ray fluorescence spectrometry (ZSX Primus). Rare earth element (REE) and minor element measurements were made using an inductively coupled plasma mass spectrometer (ICP-MS) at the Instrumental Analysis and Research Center, SUSTech. To establish an age model for the studied nodule, four slices (L1, L4, L9, and L15) were selected for beryllium (Be) isotope measurements ( 10 Be/ 9 Be) at the GFZ German Research Centre for Geosciences, Potsdam, Germany. Analytical and geochronological methods used are similar to those of Usui et al. (2007).
Lead (Pb) and neodymium (Nd) isotopic compositions were analyzed for each slice after conventional digestion and purification (Jones et al., 2000) using a Neptune plus multichannel ICP-MS housed at the Beijing Createch Testing Technology Co., Ltd. Pb isotopes are used to determine detrital component provenance in North Pacific Ocean sediments, manganese nodules, and crusts (Christensen et al., 1997;Jones et al., 2000). Nd isotopes from nodules and crusts have been used to reconstruct deep-ocean ventilation (e.g., Ishizuka et al., 2006;Haynes et al., 2020). The normalization formula for Nd isotopes (expressed as ε Nd ) is: where, CHUR represents the "chondritic uniform reservoir" and 143 Nd/ 144 Nd CHUR is 0.512638.

Age Model
A Fe-Mn-10(Cu + Co + Ni) ternary diagram for the nodule indicates that it has a hydrogenetic origin, meaning that ambient seawater was the metal source for building the nodule (Figure 1b, Tables S1 and S2 in Supporting Information S1). Therefore, the nodule should have negligible diagenetic and hydrothermal inputs (Bonatti et al., 1972) that would complicate interpretation of the archive recorded during nodule growth.
For hydrogenetic manganese nodules, the Co flux from seawater is considered constant (Halbach et al., 1983;Manheim & Lane-Bostwick, 1988), which can be used to date nodules based on their growth rate (Frank, O'Nions, et al., 1999;Manheim & Lane-Bostwick, 1988;McMurtry et al., 1994). There are several empirical equations that relate growth rate with Co contents. Although they share the same mathematical form, their coefficients are different, which could yield significantly distinct estimates for the same sample (Yi et al., 2020). In addition, the Co-based method can only provide relative ages if additional constraints from absolute age points are lacking.
To resolve these issues, we measured the Co content of each slice of the sample to construct a Co-based age model with three Be isotope ages as calibration points. Rather than directly employing the reported empirical equations, we determined the coefficients of an empirical equation using the constraints from Be ages by a Bayesian approach (Figure 1c), which is implemented using the python probabilistic package PyMC3 (Salvatier et al., 2016; Appendix S1 in Supporting Information S1). The resulting age model suggests that the nodule started to form at ∼8.5 Ma with a mean growth rate of ∼2.9 mm/Myr. This approach allowed us to address the uncertainty in the age estimates that is critical for low-resolution paleoclimatic records.

Geochemical Characteristics
Phosphorus (P) is used to indicate organic matter input to the nodule, as described in Appendix S2 in Supporting Information S1. REEs have unique geochemical characteristics due to their 4f electron configurations (Henderson, 1984), among which cerium (Ce) changes into a tetravalent state under oxic conditions in marine environments. Based on the Post-Archean Australian Shale-normalized (PAAS) REE pattern (Pourmand et al., 2012), a strong positive Ce anomaly ( Figure S2a in Supporting Information S1) supports a hydrogenetic origin for the nodule and indicates an oxic ambient environment throughout its formation. Normalized Ce, where, N indicates PAAS normalization (Piper, 1974), is used to reveal subtle bottom water oxygenation variations during nodule formation. It suggests that bottom waters became less oxic from ∼9 to ∼7 Ma ( Figure S2e in Supporting Information S1). In addition, our P and δCe records have opposite trends ( Figure S2 in Supporting Information S1), which suggests that oxygenation variations indicated by δCe were likely related to organic matter input.
ε Nd values for the nodule range from −4.5 to −6.1 (Table S1 in Supporting Information S1), which indicates negligible radiogenic volcanic input (Frank, 2002); a local Nd isotope overprint (e.g., continental inputs or arc weathering) can be excluded in this region. The sampling site is located under the pathway of the lower circumpolar deep water sourced from Antarctic bottom water ( Figure 1a). Therefore, ε Nd should mainly reflect a watermass signal and indicate ventilation changes.
Pb in the North Pacific Ocean sediments is derived mainly from leaching of central Asian eolian dust (Jones et al., 2000). Pb isotope signals ( 208 Pb/ 206 Pb vs. 207 Pb/ 206 Pb and 206 Pb/ 204 Pb vs. 207 Pb/ 204 Pb) in the nodule slices resemble those of sands from the central Asian desert and bulk Asian loess ( Figure S3 in Supporting Information S1). Also, adjacent lzu-Bonin and Mariana arc volcanoes do not contribute significantly to the Pb isotope composition of the nodule. This indicates that detritus in the studied nodule is from Asian eolian dust, and that the nodule records Asian aridification.

Magnetic Properties of the Manganese Nodule
Abundant magnetic nanoparticles were extracted from a powdered nodule sample. The nanoparticles have a restricted 35-120 nm sizes, with dominantly cubo-octahedral morphology (>90%) and a minor proportion of elongated prisms ( Figure S4a-c in Supporting Information S1). High-resolution TEM images and electron diffraction patterns ( Figure S4d-g in Supporting Information S1) suggest that the particles are magnetite. FORC diagrams for bulk samples have a prominent central ridge along the B c axis (B i = 0; Figure S4h-i in Supporting Information S1) that is diagnostic of magnetically non-interacting or weakly interacting uniaxial single domain particles (e.g., Egli et al., 2010). These results are consistent with previous studies and suggest that magnetite magnetofossils dominate the magnetic particles in western Pacific manganese nodules and crusts (Jiang et al., 2020;Yuan et al., 2020). The magnetofossil morphologies and the oxic seafloor environment indicated by the positive Ce anomaly are consistent with previous studies (e.g., Jiang et al., 2020), and suggest that the magnetofossils are associated with MTB species that prefer oxic conditions (Yamazaki & Shimono, 2013;Yamazaki et al., 2019). ARM is used here as a quantitative magnetofossil proxy because it is indicative of fine magnetic particle concentrations (Banerjee et al., 1981).

Impacts of Asian Aridification on the NW Pacific Since the Late Miocene
P 2 O 5 variations recorded by the nodule since ∼8 Ma are comparable to those of the eolian mass accumulation rate (MAR) at ODP Site 885/886 (Rea et al., 1998), dust flux from central Asia in the Japan Sea as indicated by clay mineral compositions (Shen et al., 2017), and loess accumulation rate in North China (Figure 2; Guo et al., 2002), particularly during nodule growth stages 1 (∼8-7 Ma) and 3 (∼3.6-0 Ma). Eolian dust is an important P source to the abyssal oceans (Paytan & McLaughlin, 2007). The nodule site is located along the Westerly path, as are ODP Site 885/886, the Japan Sea, and North China. There are similar trends among dust MAR, clay mineral ratio, loess, and P (Figures 2b-2e), therefore, indicate that they are inherently linked and regulated by central Asian aridity. This inference is supported strongly by previous and our own Pb isotope evidence ( Figure S3 in Supporting Information S1; e.g., Jones et al., 2000;Ling et al., 2005).
Across the nodule, we observe two striking intervals with enhanced eolian dust and P (indicative of primary productivity) at ∼8-7 Ma and ∼3.6-0 Ma, respectively. δCe, which is a proxy for redox conditions, indicates that the seafloor became less oxic with enhanced dust fluxes (Figure 2f), which can be ascribed to increased organic matter deposition associated with primary productivity. Furthermore, the covarying MTB proxy (Figure 2g) supports this idea given that suboxic conditions restrain aerophilic MTB biomineralization in seafloor nodules (Jiang et al., 2020). In addition, the two intervals were also recorded by another side of the studied nodule ( Figure  S1 in Supporting Information S1) as indicated by green lines (Figures 2e-2g), which confirms the reliability of paleoenvironment recording in manganese nodule (Appendix S3 in Supporting Information S1).
As primary productivity in the study area is unlikely to arise from increased upwelling as indicated by ε Nd values (Figure 2h), we suggest that iron fertilization by dust is responsible for the enhancement. Therefore, covariation among dust MAR, P, δCe, and magnetofossils indicates that dust supply from central Asia to the NW Pacific regulated local primary production and may have resulted in productivity blooms since, the late Miocene. This suggests that prominent Asian aridification at ∼8-7 Ma and 3.6-0 Ma profoundly impacted NW Pacific Ocean biogeochemical cycling and abyssal microbial activity.
P content and nodule growth rate during stage 1 were higher than during stage 3, while eolian dust flux was comparable during these two stages (Figure 2). This can be explained by ocean stratification in the study area.
No obvious water mass variation occurred during stage 1, as indicated by ε Nd values (Figure 2), and enriched seafloor organic matter inputs probably reflect surface productivity. The stage 3 (∼3.6-0 Ma) environment was characterized by global cooling, which generally contributed to sluggish ocean circulation. Furthermore, ε Nd increased during nodule formation in stage 3 (Figure 2h), consistent with previous studies (e.g., van de Flierdt et al., 2004), which indicates attenuated deep-ocean ventilation or enhanced southward intermediate water flow (Kender et al., 2018). Either mechanism could have enhanced ocean stratification to reduce organic carbon and nutrient transfer to the seafloor from the euphotic zone.  (Guo et al., 2002). (c) Japan Sea Asian eolian dust flux based on illite/smectite fluxes (Shen et al., 2017). (d) Eolian dust mass accumulation rate from ODP Site 885/886 (Rea et al., 1998). (e)-(g) Bulk P 2 O 5 and redox proxy (δCe) across the nodule (Tables S1 and S2 in Supporting Information S1). Anhysteretic remanent magnetization proxy for the relative magnetofossil abundance in nodule slices (Red and green lines indicate the proxies from upper and lower parts of nodule, respectively [ Figure S1 in Supporting Information S1], and blue lines represent their average value) (h) ε Nd proxy for bottom water ventilation variations during nodule formation. A sudden ε Nd increase at ∼5 Ma was probably associated with Panama gateway closure, as suggested by Frank, Reynolds, and O'Nions (1999), when Atlantic seawater with low ε Nd (∼−10 to −14; Albarède & Goldstein, 1992) stopped entering directly into the Pacific Ocean.

Processes Driving Central Asian Aridification Since the Late Miocene
The late Miocene was a time of significant climatic, tectonic, and environmental change (∼8-5.5 Ma; e.g., Guo et al., 2002;Holbourn et al., 2018;LaRiviere et al., 2012;Li et al., 2014;Rea et al., 1998;Zhang et al., 2014), which makes it challenging to distinguish driving forces responsible for central Asian aridification between global cooling and Tibetan Plateau uplift (Li et al., 2014;Zachos et al., 2008). NE Tibetan Plateau uplift at ∼8-7 Ma has been considered a potential candidate for aridification based on synchronism between uplift and Asian aridification (Li et al., 2014;Shen et al., 2017). However, other studies (Zhang et al., 2018;Matsuzaki et al., 2020) ascribe the aridification to global cooling evidenced by a prominent global sea-surface temperature (SST) drop, especially in mid-and high-latitudes (Figure 3b; Herbert et al., 2016), and a cooling event at ∼ 7.8 Ma from the most recent marine benthic δ 18 O megasplice of Westerhold et al. (2020).
Our data provide insights into the main mechanisms. Increased eolian inputs at ∼8-7 Ma may have enhanced NW Pacific Ocean primary productivity (Figures 2c and 2d). To our knowledge, there was no clear primary productivity change along the equatorial Pacific Ocean or in other oceans during this period, which would have been coeval with the enhanced productivity proposed here ( Figure 3); this implies that the proposed NW Pacific event was not global in nature. In addition, the Taklimakan Desert is suggested to have formed at ∼7 Ma (Sun et al., 2009, which is likely related to Tibetan Plateau uplift and resulting rain-shadow effect at ∼8-7 Ma. Therefore, central Asian aridification at ∼8-7 Ma was probably a regional event that cannot be explained by global cooling; rather it was probably associated with NE Tibetan Plateau uplift (Figure 4a).
Despite its local origin, late Miocene central Asian aridification may have had an important impact on a global climate transition. This event may have enhanced global cooling that is evident in lower SST at ∼7.5-5.5 Ma (Figure 3b), which is considered to be a response to a significant late Miocene atmospheric CO 2 decline associated with biological carbon pumping (Diester-Haass et al., 2006;Herbert et al., 2016). Enhanced NW Pacific productivity at ∼8-7 Ma suggested by our data could have played a significant role in local or regional carbon cycling and may have acted as a forerunner for global cooling through declining atmospheric CO 2 at ∼7.5-5.5 Ma. Global cooling detected by the late Miocene SST change is not fully registered by marine benthic δ 18 O (∼7.5-7 Ma; Figure 3. Representative global and regional environmental change records since the late Miocene (a) Benthic foraminiferal δ 18 O megasplice (Westerhold et al., 2020). (b) Representative sea surface temperature records (Herbert et al., 2016). (c) Asian eolian dust flux to the North Pacific Ocean (Rea et al., 1998). (d) Bulk P across the studied nodule (Red and green lines indicate the proxies from upper and lower parts of nodule, respectively [ Figure S1 in Supporting Information S1], and blue lines represent their average value. Shading indicates 90% credible interval based on Monte Carlo simulations). (e)-(g) Global ocean primary productivity indicated by TOC and primary productivity carbon (Diekmann et al., 2003;Diester-Haass et al., 2006;Gupta et al., 2015;Stickley et al., 2004). Figure 3a), which can be explained if cooling did not fall below the threshold for permanent continental ice sheet development (Herbert et al., 2016).
A similar controversy exists regarding the driving forces(s) of central Asian aridification since the late Pliocene (Guo et al., 2020;Liu et al., 2015;Lu & Guo, 2014) because rapid northern hemisphere ice sheet expansion (Xiong et al., 2003;Zachos et al., 2008) was synchronous with marked NE Tibetan Plateau uplift Li et al., 2014) since ∼3.6 Ma. Prominently enhanced global primary productivity has also been documented since ∼3.6 Ma compared with the ∼8-7 Ma period (Figure 3), which indicates that cooling impacted global environments profoundly during this period. This indicates that NW Pacific Ocean primary productivity was most responsive to global cooling, although NE Tibetan Plateau uplift may also impacted NW Pacific productivity (Figure 4b). These global to regional results suggest that global cooling likely played an overwhelming role in central Asian aridification compared to Tibetan Plateau uplift since ∼3.6 Ma.

Conclusions
Geochemical and magnetic data from an abyssal NW Pacific Ocean manganese nodule provide a rich palaeoenvironmental record for the past ∼8 Ma. These records trace eolian dust inputs from central Asia to the NW Pacific Ocean. Two prominent intervals with increased dust fluxes at ∼8-7 Ma and ∼3.6-0 Ma are associated with enhanced primary productivity and decreased magnetotactic bacteria at the seafloor, which highlight the . Conceptual model for links between climate change and abyssal microbial activity for two scenarios (OMZ = oxygen-minimum zone, OC = organic carbon) (a) Central Asian aridification sdue to NE Tibetan Plateau uplift at ∼8-Ma. Enhanced eolian input to the NW Pacific triggered increased primary productivity, which decreased seafloor oxygenation due to increased organic matter degradation. This environmental change impacted magnetotactic bacteria biomineralization. (b) Central Asian aridification due to global cooling since ∼3.6 Ma. The same link from Asian aridification to abyssal NW Pacific microbial activity also occurred during this period. Organic matter and nutrient input to the seafloor may have also been restrained by ocean stratification, with vegetation dominated by C4 plants during this period.
impact of central Asian aridification on regional biogeochemical cycling. Our results suggest that central Asian aridification at ∼8-7 Ma was mainly forced by Tibetan Plateau uplift rather than global cooling, with global cooling providing the driver of central Asian aridification since 3.6 Ma.

Data Availability Statement
The data presented here can be found in Supporting Information S1 and at http://doi.org/10.5281/zenodo.4882952. of Marine Resources and Coastal Engineering, the Australian Research Council Discovery Projects DP200100765, and the National Natural Science Foundation of China (Grant 41920104009) for financial supports, and the National Institute of Polar Research (NIPR) through Advanced Project (KP-7 and KP306) and JSPS KAKENHI grants (15K13581, 16H04068, 17H06321 and 18K13638). This research used samples provided by COMRA. The authors thank staff of the Instrumental Analysis Research Center (SUSTech) and Institute of Geology and Geophysics, Chinese Academy of Sciences (Beijing) for laboratory assistance.