Astronomically‐Paced Changes in Paleoproductivity, Winnowing, and Mineral Flux Over Broken Ridge (Indian Ocean) Since the Early Miocene

A significant shift in Earth's climate characterizes the Neogene, transitioning from a single‐ice‐sheet planet to the current bipolar configuration. This climate evolution is closely linked to changing ocean currents, but globally‐distributed continuous high‐resolution sedimentary records are needed to fully capture this interaction. The Ocean Drilling Program (ODP) Site 752, located on Broken Ridge in the Indian Ocean, provides such a Miocene‐to‐recent archive. We use X‐ray fluorescence (XRF) core scanning to build an eccentricity‐tuned age‐depth model and reconstruct paleoceanographic changes since 23 Ma. We find two intervals of enhanced productivity, during the early and middle Miocene (18.5–13.7 Ma) and late Pliocene/early Pleistocene (3–1 Ma). We also report a mixed eccentricity‐obliquity imprint in the XRF‐derived paleoproductivity proxy. In terms of grain size, three coarsening steps occur between 19.2–16 Ma, 10.8–8 Ma, and since 2.6 Ma. The steps respectively indicate stronger current winnowing in response to vigorous Antarctic Intermediate Water flow over Broken Ridge in the early Miocene, the first transient onset of Tasman Leakage in the Late Miocene, and the intensification of global oceanic circulation at the Plio‐Pleistocene transition. High‐resolution iron and manganese series provide a detailed Neogene dust record. This study utilized a single hole from an ODP legacy‐site. Nevertheless, we managed to provide novel perspectives on past Indian Ocean responses to astronomical forcing. We conclude that Neogene sediments from Broken Ridge harbor the potential for even more comprehensive reconstructions. Realizing this potential necessitates re‐drilling of these sedimentary archives utilizing modern drilling strategies.

restriction blocked the flow of warm Pacific water into the Indian Ocean (Kuhnt et al., 2004).The second was the final termination of the Tethyan Ocean (∼13.8 Ma, Bialik et al., 2019).This final closure of the Tethyan Ocean decoupled the Mediterranean Sea and the Indian Ocean, preventing Indian Ocean warm water from flowing west and entering the Atlantic Ocean at low latitudes.The closure of the Tethys shifted the inter-basinal connection between the Indian and Atlantic oceans to the mid-latitudes.Hence, surface waters were forced south around Africa, which is known as Agulhas Leakage (de Ruijter et al., 1999;Durgadoo et al., 2017;Gordon, 2003;Ohishi et al., 2017;Ridgway & Dunn, 2007).This configuration is marked by three inter-connected southern subtropical gyres, which together form the Southern Hemisphere Supergyre (Ridgway & Dunn, 2007;Speich et al., 2002).To a lesser extent, similar shifts in inter-basinal connections between the Pacific and Indian oceans occurred throughout the Neogene.The link between the Pacific and Indian oceans occurs mainly through routes the low-latitude Indonesian Throughflow (ITF, Gordon, 2005) and the middle-latitude Tasman Leakage (TL, Speich et al., 2001).The ITF transports water from the Pacific to the Indian Ocean at surface water depths (<300 m, van Sebille et al., 2014).Tasman Leakage occurs at intermediate water depths south of Australia (Middleton, 2002;Rosell-Fieschi et al., 2013;Speich et al., 2002;van Sebille et al., 2012van Sebille et al., , 2014) ) and was established ∼7 Myr ago (Christensen et al., 2021)-marking another connectivity shift to the middle-latitudes.
Neogene shifts in Indian Ocean dynamics are also interwoven with the evolution of the monsoon systems.Monsoonal winds drive ocean currents and contribute to upwelling and downwelling (Betzler et al., 2016;Bialik et al., 2020).For example, the abrupt intensification of South Asian Monsoon winds compounded by the onset of oceanic circulation and the global cooling during the Middle Miocene (Betzler et al., 2016).Hence, monsoonal winds influence ocean salinity, temperature, oxygen, and nutrient levels.Previous research (Dickens & Owen, 1994) has suggested the existence of an expanded oxygen minimum zone (OMZ) between 6.5 and 3 Ma, ranging from the northern Indian Ocean as far south as Broken Ridge.This interpretation is mainly based on observations of Mn depletions at Broken Ridge and benthic foraminifera assemblages (Dickens & Owen, 1994;Gupta et al., 2013).However, the timing and extent of this OMZ during the Late Miocene are poorly understood.Definitive evidence for the cross-hemispheric expansion of the Indian Ocean OMZ is currently lacking.
Over the last decade, the International Ocean Discovery Program (IODP) has primarily focused on the Indian Ocean marginal areas, particularly on regions influenced by monsoons (Clift et al., 2022).However, locations in the interior of the Indian Ocean also holds great potential for revealing shifts in the position of the Indian Ocean gyres, productivity patterns, and general changes in ocean circulation.Such regions were sampled by the Ocean Drilling Program (ODP) in the 1980's and 90's.For example, ODP Leg 121 drilled Site 752 on Broken Ridge.Broken Ridge is a west-northwest trending oceanic plateau in the central Indian Ocean that separated from the Kerguelen Plateau during the middle Eocene (∼42 Ma).Since then, it has drifted from a ∼55°S paleolatitude to its present-day location at ∼31°S (Berggren et al., 1985;Mutter & Cande, 1983).It was accompanied with subsidence but no large-scale tilting during its northward drift.This makes for a ∼100 m-thick package of Neogene sediments that accumulated on Broken Ridge.At ODP Site 752, this so-called horizontal cap consists of nannofossils ooze with foraminiferal sand (94.5%-97%CaCO 3 , Shipboard Scientific Party, 1989).While Site 752 has good recovery in its Neogene section (∼95.9%), the site has not yet been studied for paleoceanography on astronomical timescales.Moreover, due to its location and the intermediate water depth, this site can be used for tracing Antarctic Intermediate water (AAIW), and TL.Nevertheless, this sedimentary archive has the potential to reveal spatial and temporal changes in Indian Ocean paleoceanography since the Early Miocene.This is mainly because new techniques, like cm-scale X-Ray Fluorescence (XRF) core scanning, are standardly used nowadays but were not yet available at the time of coring.
In this study, we generate an astronomically-tuned age-depth model, based on the XRF-derived calcium-iron (Ca/Fe) ratios downcore.Thereby, it is important to acknowledge that the XRF data used in this study originate from a single hole (ODP Hole 752A), precluding the application of stratigraphic splicing techniques employed in contemporary IODP paleoceanographic studies.Nonetheless, the 95.9%-recovery achieved in Hole 752A enables us to develop an age-depth model based on the eccentricity scale.This age-depth model is then used to investigate time-series of other elemental proxies and to characterize oceanographic change at Broken Ridge since the Early Miocene (e.g., productivity, current winnowing, and dust transport).

Site Description
ODP Hole 752A (Figure 1a) was drilled near the crest of Broken Ridge (30°53.475′S,93°34.652′E;1086.3 m water depth).Broken Ridge separated from the Kerguelen Plateau due to lithospheric extension and seafloor spreading that began before 42 Ma (Berggren et al., 1985;Mutter & Cande, 1983).Broken Ridge has moved north by about 20° of latitude since the Middle Eocene as part of the Indo-Australian plate.More than 1,500 m of sediments have accumulated on Broken Ridge, recording the oceanographic history of the Indian Ocean from the Late Cretaceous until today.Sediments from Broken Ridge consist of four major sections.The horizontal cap (fourth section) has been recovered at Site 752, 753, 754, 755, and1,141 (Shipboard Scientific Party, 1989, 2000).At Site 752, the horizontal cap consists of nannofossils and foraminiferal ooze with foraminiferal sand (94.5%-97%CaCO 3 ).The base of the horizontal cap includes some upper Eocene and upper Oligocene sediments, but it chiefly consists of Neogene carbonate oozes, deposited under pelagic conditions (Shipboard Scientific Party, 1989).For this study, we exclusively focus on the Neogene portion of the horizontal cap, clearly above the angular unconformity, studying sediments between cores 752A-1H and 752A-10H (0-91.70 m below sea floor (mbsf); Figure 1b).
Broken Ridge has always been located in intermediate water depths throughout the studied interval.The angular unconformity described above consists of two pebble layers at the bottom of the horizontal cap.Subsequent paleo-water depths gradually deepen up section.Sclater et al. (1985) calculated 316 m of subsidence for Broken Ridge since 22 Ma, based on an assumption that the Broken Ridge crust is 85 Myr old.However, the age of the crust is ∼100 Ma according to the Shipboard Scientific Party (1989).For that reason, we assume that Neogene subsidence rates at Broken Ridge must have been less than 316 m over the last 22 Myr.Moreover, the paleo-waterdepth reconstruction from benthic foraminifera suggests that water depths were already as deep as ∼800 m water around 23 Ma, with only a slight and gradual deepening afterward, to around a water depth of ∼1,000 m today (Driscoll & Karner, 1991).Paleobathymetric records by Shipboard Scientific Party (1989) also corroborate that Site 752 has remained at intermediate water depths throughout the Neogene.

XRF Core Scanning
The bulk elemental composition of ODP Hole 752A sediments was measured on the archive-half core surfaces using XRF core scanning using an Itrax XRF scanner (Löwemark et al., 2019) at the Kochi Core Center (KCC), Kochi, Japan.Measurements were taken at a spatial resolution of 2 cm, with an X-Ray beam size of 0.2 × 20 mm, a Molybdenum X-Ray source energy of 30 kV (55 mA, no filter), and a 10 s count time for each measurement.Element intensities (in counts) were obtained by processing raw X-ray spectra using the ITRAX software for the elements Al, Si, P, S, Cl, Ar, K, Ca, Ti, V,Cr, Mn, Fe, Ni, Cu, Zn, Ga, As, Se, Br, Rb, Sr, Y, Zr, Cd, Sn, Cs, Ba, Ta, W, Re as well as the coherent (coh) incoherent (inc) scatter of the Molybdenum X-ray source.

Paleoproductivity Index
To reconstruct changes in paleoproductivity (export productivity), we utilize a group of metallic micronutrients (Cd, Cu, Ni, Zn, Cr and Br).These elements are usually depleted at the surface ocean by biology, we argue their accumulation in the sediment could be used to trace productivity (Steiner et al., 2017;Tribovillard et al., 2006).To generate a paleoproductivity index we summed up the XRF counts of all productivity-related elements and multiplied by 1,000.To account for the background, we then divided the product by Ca counts.This paleoproductivity index is designed to capture fluctuations in the relative strength of the nutrient-limited biological pump versus the alkalinity pump.
We recognize that counts for the selected metallic micronutrients are generally low.Nevertheless, we argue that each of those elements carries a paleoproductivitiy signal (and is therefore not pure noise) because of three main reasons.First, for all these elements, an exposure time of 10 s is sufficient to obtain reliable results according to the practical guidelines in Löwemark et al. (2019).Second, the covariance matrix (Figure S1 in Supporting Information S1) of Cd/Ca, Cu/Ca, Ni/Ca, Zn/Ca, Cr/Ca and Br/Ca shows that these productivity-related elements are mostly positively correlated, and hence likely to be reflecting the same paleoproductivity signal.Especially Zn, Cr, Ni, and Cu seem to be elements with mutually consistent (and therefore non-random) signals.Third, and most importantly, our paleoproductivity index exhibits co-variation with shipboard C org measurements (Figure S2 in Supporting Information S1).

Current Winnowing Proxy
Here, we use the total counts received by the XRF core-scanner detector as a measure of sediment bulk density.Total counts record shows a good agreement with the shipboard GRA bulk density (Figure S3 in Supporting Information S1), which supports that total counts record relates to the sediment bulk density and can be used as proxy for the current winnowing.We also checked the paleoproductivity index with this XRF-derived current winnowing proxy, which show no correlation between these two records.

Grain Size Analysis
Grain size measurements of 123 samples throughout the last 23 million years were conducted in the Particle-Size Laboratory at MARUM, University of Bremen with a Beckman Coulter Laser Diffraction Particle Size Analyzer LS 13320.Samples were first diluted (dilution factor: >25), and the samples were boiled with ∼0.3 g tetra-sodium diphosphate decahydrate (Na 4 P 2 O 7 * 10H 2 O) to ensure fully disaggregated prior to the measurements (McGregor et al., 2009).All preparation steps and measurements were carried out with deionized, degassed and filtered water (filter mesh size: 0.2 μm) to reduce the potential influence of gas bubbles or particles within the water.The particle-size distribution result of a sample consists of 116 size classes (from 0.04 to 2,000 μm).The calculation of the particle sizes relies on the Fraunhofer diffraction theory and the Polarization Intensity Differential Scattering for particles from 0.4 to 2,000 μm and from 0.04 to 0.4 μm, respectively.The reproducibility was checked regularly when analyzing the full sample by replicate analyses of three internal glass-bead standards and is found to be better than ±0.7 μm for the mean and ±0.6 μm for the median particle size (1σ).The average standard deviation integrated overall size classes is better than ±4 vol% (note that the standard deviation of the individual size classes is not distributed uniformly).All provided statistic values are based on a geometric statistic.

Biostratigraphy
Biostratigraphic age control points from the shipboard report (Shipboard Scientific Party, 1989) were updated to GTS2020 (Table 1, Raffi et al., 2020).During the expedition, 4-to-5 cm thick biostratigraphic samples were taken from every core catcher (i.e., every ∼9 m).The depth error shown in Table 1 was calculated by taking the stratigraphic distance between the proposed stratigraphic depth of the datum in the shipboard report and the stratigraphic depth of the corresponding biostratigraphic sample.The biostratigraphic age-depth tie-points (Table 1) suggest relatively stable sedimentation rates throughout the studied interval, ranging between 2.9 and 5.7 m/Myr.

Time-Series Analysis
Spectral analyses in this study were carried out using the multitaper method (MTM) with five 2π-tapers (Thomson, 1982) and robust AR(1) background estimation (Meyers, 2012), as implemented in the R-package astrochron (Meyers, 2014).We emphasize that the analyzed XRF data set originates from a single hole is characterized by up to 1.92-m-long datagaps in-between cores 752A-6H and 752A-7H.In present-day paleoceanographic studies, this issue is alleviated by constructing a stratigraphic splice using multiple holes.To better illustrate the effect of these datagaps, and to account for possible changes in sedimentation rate, we also carry out wavelet-transform time-series analysis.Wavelet analysis was conducted using R-package biwavelet (Gouhier et al., 2013).Depth-to-time conversion and bandpass filtering were carried out using functions "tune" and "bandpass" from the R-package astrochron.

Astrochronology
We adopt an astrochronologic approach to reconstruct the age-depth model (Table 1).First, we detrended the Ca/ Fe depth series by cutting off the frequency smaller than 0.05 cycles/m, and apply wavelet and spectral analysis to the Ca/Fe series in the depth domain.The power spectrum of the Ca/Fe depth series (Figure 2b) shows a spectral peak ∼0.63 cycles/m (∼1.6 m/cycle), which may reflect the imprint of the long 405-kyr eccentricity cycle given the biostratigraphic estimate of long-term average sedimentation rate around 5.7 m/Myr.We suspect that Ca/Fe is climatically controlled by aridity, which in turn could have driven changes in eolian flux.The same periodicity can be discerned from the wavelet transform, at least between 0-33 mbsf and 44-65 mbsf (white dashed line on Figure 3a).While the MTM power spectrum does not show a spectral peak that could be associated with 100-kyr eccentricity, the wavelet spectrum exhibits a bifurcation pattern at ∼0.4 m cycles, and markedly higher periodicities between 33 and 44 mbsf (white circles on Figure 3a).The bifurcation pattern is in an agreement with the long-eccentricity amplitude modulation of the 100-kyr cycles.The wavelet analysis (Figure 3a) thus suggests relatively stable sedimentation rates, in agreement with the biostratigraphic estimate, except for the interval between 33 and 44 mbsf, where sedimentation rates might have been somewhat higher.The wavelet  Note.Samples for biostratigraphy have been taken from every core-catcher in the studied interval between cores 752A-1H and 752A-10H.We estimate the depth error by reporting the stratigraphic distance between the datum and the corresponding sample.transform (Figure 3a) also suggests significant variability at periodicities in-between the interpreted 100-kyr and 405-kyr eccentricity bands.This spectral power could tentatively be ascribed to the 173-kyr term in obliquity amplitude modulation.
In a next step, a broad Gaussian bandpass filter is applied to the Ca/Fe series.The filter is centered on the peak at ∼0.63 cycles/m (bandwidth: 0.4-0.95cycles/m).We fine-tuned the XRF-derived Ca/Fe series according to the above-described astronomical interpretation by correlating maxima in the Ca/Fe ∼0.63 cycles/m filter to eccentricity maxima and vice versa (gray connecting lines between Figures 3a and 3b).This strategy can be adopted between 0 and 15.1 Ma, but the older interval (15.1-23.0Ma) does not exhibit the clear ∼405kyr imprint needed for astronomical tuning.Therefore, we use the biostratigraphic tie-points in the >15.1 Ma interval.We do not include any phase-lags while tuning.The phase relationship between the proxy and the astronomical solution is in agreement with the findings by Vervoort et al. (2021) for pelagic carbonates.This phase-relationship implies that more arid climates around the Indian Ocean correspond to eccentricity minima (e.g., De Vleeschouwer et al., 2018;Duesing et al., 2021;Liu et al., 2021).The power spectrum (Figure 2c) of the astronomically-tuned Ca/Fe time-series shows an obliquity signal, reaching to the 99% confidence level.We present the Ca/Fe time-series along the final astronomically-tuned age model in Figure 3d.

Elemental Geochemistry
The XRF-derived elemental geochemistry data set generated for ODP Hole 752A consists of 31 elements, of which several have the potential to serve as a proxy for specific environmental parameters (Löwemark et al., 2019).In the following subsections, we will discuss the XRF-derived elements according to their environmental interpretation.

Paleoproductivity
The paleoproductivity index fluctuates at relatively high values in the Early and Middle Miocene before sharply decreasing around 14 Ma (Figure 5a).This pattern shows the same trend as the low-resolution shipboard record of organic carbon content (Shipboard Scientific Party, 1989; Figure 5a), which again, approved the validation of this paleoproductivity proxy (Figure S2 in Supporting Information S1 shows the correlation between the paleoproductivity proxy and C org ).The index rebounds after the middle Miocene minimum and then maintains a rather stable level until the Late Pliocene.None of these shifts correspond to the Biogenic Bloom event in the Indian Ocean and elsewhere (Dickens & Owen, 1999;Grant & Dickens, 2002;Hermoyian & Owen, 2001;  Karatsolis et al., 2022).Nor do we observe particularly elevated productivity levels around the time of the Biogenic Bloom (Figures 5a and 5b).This is consistent with the very heterogeneous nature of the Biogenic Bloom (Pillot et al., 2023): the Biogenic Bloom appears to have been highly magnified in already high productivity zones, whereas ODP Hole 752A was located in the center of a low productivity oceanic gyre throughout the studied interval.ODP Hole 752A is also far away from any continental margin area or upwelling zone.All these factors combined make ODP Hole 752A less suited to capture the surface productivity increase of the Biogenic Bloom.

Paleoproductivity Pacing the Eccentricity Forcing
The multitaper power spectrum of the paleoproductivity index displays statistically-significant peaks at frequencies that are reminiscent of 405 and ∼100-kyr eccentricity forcing.Elevated spectral power can also be observed at higher frequencies, possibly hinting at an imprint of obliquity and precession (Figure 6).However, the 2-cm sampling resolution (∼5 kyr temporal resolution) may not be adequate to interpret the highest-frequency peaks (precession) with confidence.We interpret the paleoproductivity reconstruction to be driven by eccentricity and obliquity on astronomical timescales.Therewith, we make a similar inference as Karatsolis et al. (2022), who constructed and analyzed a low-latitude global paleoproductivity composite between 7 and 3 Ma.Our Broken Ridge paleoproductivity reconstruction shows similar dynamics to the Karatsolis record on eccentricity timescales before 4.4 Ma.Following 4.4 Ma, however, the behavior diverges (Figure 5b).Karatsolis et al. (2022) report a sudden collapse of paleoproductivity (inferred to be the end of the Biogenic Bloom) at 4.4 Ma, whereas paleoproductivity levels at ODP Hole 752A remain unchanged before and after 4.4 Ma.The divergence between the two records reiterates the fact that ODP Hole 752A lies outside any of the oceanic productivity belts, and therefore exhibits an independent paleoproductivity signal.This signal is influenced by the same orbital parameters (eccentricity and obliquity)-but is not impacted by the large-scale shifts in nutrient availability occurring at oceanic margins.It would also suggest that the productivity gradients between oceanic margins and Broken Ridge during the Biogenic Bloom may have been steeper.
We suggest that the paleoproductivity patterns as recorded by ODP Hole 752A were primarily influenced by the northward tectonic movement of the Broken Ridge plateau, climate-driven changes in the latitude of the subtropical front, and astronomically-paced changes in global ocean circulation.During the Early and Middle Miocene, Broken Ridge was situated at a latitude of ∼40°S and we find enhanced paleoproductivity during globally-warm eccentricity maxima.This phase-relationship is illustrated by the negative correlation between our productivity index and the benthic foraminiferal δ 18 O record (note the flipped isotope axis in Figure 5c; coss-wavelet analysis between the productivity index and δ 18 O in Figure S4 of Supporting Information S1).This observation is in an agreement with the eccentricity-pacing of ocean, climate and carbon cycle dynamics in the Middle Miocene, as described in Holbourn et al. (2007).At that time, low eccentricity orbits might have favored organic carbon burial and global cooling.Eccentricity maxima on the other hand, especially precession-driven maxima in austral summer insolation, resulted in a smaller Antarctic ice sheet (De Vleeschouwer et al., 2017).With a warmer planet and smaller ice sheet came less vigorous ocean circulation, lower calcium carbonate saturation states in the ocean and a net flux of CO 2 from the ocean to the atmosphere.These warmer conditions have also been inferred to be associated with more intense monsoons (Holbourn et al., 2007).For ODP Hole 752A, we infer a higher paleoproductivity index during eccentricity maxima (Figures 4b and 4g), suggesting a simultaneous enhancement of the biological pump and possibly a less efficient alkalinity pump.This model is best illustrated by the rapid rise in paleoproductivity index after 18.5 Ma (Figure 5a): This interval coincides with a rapid increase in eccentricity forcing after a long-term 2.4-Myr eccentricity minimum (Figures 4b and 4g) and a rapid warming after one of the coolest intervals of the Early-Middle Miocene (Holbourn et al., 2015).

No Significant Increase in Paleoproductivity Since the Late Miocene
The globally-warm period of the Early and especially the Middle Miocene (Miocene Climatic Optimum, MCO) was abruptly terminated by a major increase in Antarctic ice volume and a global sea level fall at 13.9 Ma.This global cooling (Middle Miocene Climatic Transition, MMCT) coincided with a dramatic decline of productivity over Broken Ridge.This event is paired with a rapid rise and fall in oceanic turnover (Crampton et al., 2016).Afterward, productivity slightly recovered and remained rather stable throughout the Late Miocene and early Pliocene.This stability is likely the result of the northward tectonic movement of Broken Ridge on the one hand, and the simultaneous northward shift of the Southern Hemisphere climate belts on the other hand (Groeneveld et al., 2017).Throughout this interval, Broken Ridge was probably situated slightly north of the subtropical front.The most recent increase in paleoproductivity occurs around 3.5 Ma and coincides with generally cool climate conditions across the Southern Hemisphere and a more northerly position of the subtropical front (De Vleeschouwer et al., 2022).This shift in the latitudinal position of the climate belts leads to a contraction of the Indian Ocean gyre, and positioned the subtropical front north of Broken Ridge.Thus, Broken Ridge would be closer to the higher-productivity area, which is outside of the oligotrophic Indian Ocean gyre.

Current Winnowing
The Neogene portion of Site 752 is mainly foraminiferal and nannofossil oozes (Shipboard Scientific Party, 1989).Those two sedimentary components generate a bimodal grain size distribution with two distinct modes, at ∼10 and ∼150 μm, throughout the studied interval (Figure 7b).While the relative strength of these two modes are marked by considerable variability through time, their position in terms of grain size is stable.This pattern indicates that calcareous nannofossils and foraminifera are the two major parts of the sediments, with calcareous nannofossils making up the bulk part of the finest sediment (<63 μm) and foraminifera making up the coarser part of the sediment (100-350 μm).
We find, however, strong variations in the fine fraction proportion with respect to the total sediments.These variations likely reflect periods of stronger and weaker winnowing, with stronger winnowing leading to a removal of the finest fraction (House et al., 1991).Although there are other factors that may have influence on the grain size, they play only a minor role here over Broken Ridge (House et al., 1991).The most common among these factors is dissolution.However, the paleodepths were always above the lysocline (Peterson & Prell, 1985;Rea & Leinen, 1985), and foraminifera are well preserved throughout the Neogene.Moreover, there is no dissolution pulse or sea-level change resemblance to the grain size record (House et al., 1991).Therefore, we argue that the grain size record mainly reflects variations in current energy.The grain-size variations are also registered by the core bulk density, with a strong negative correlation between both (note the reversed y-axis for total counts on Figure 7a).The total counts reflect sediment bulk density, which is thus inversely correlated with grain size and current winnowing intensity.
The total counts time-series and the percentage of the grain size >63 μm series both suggest a decrease in winnowing between 23-19.2Ma, as well as between 15-13.7 Ma (Figure 7a).In Figure 7b, these intervals of weakening winnowing are marked by shifts in the relative importance, away from the coarser mode, toward the finer mode in the grain size distribution.The 15-13.7 Ma declining trend culminates in the finest sediment grain size distribution observed in the studied interval between 13.7 and 10.8 Ma (Figure 7).There are three periods during which winnowing becomes stronger and grain size distribution coarser: 19.2-16, 10.8-8, and ∼2.6 Ma to the recent.The relative importance of the foraminifera mode increases during these higher current energy intervals (Figure 7b).
Current winnowing was strong in the earliest Miocene portion of the record, then gradually decreased until 19.2 Ma.We relate this relatively strong winnowing to Antarctic Intermediate Waters (AAIW) flowing over Broken Ridge.In the earliest Miocene, two intermediate water masses were present in the Indian Ocean, with Tethyan Intermediate Waters (TIW) coming in from the northwest and occupying the intermediate water depth range in most of the western Indian Ocean.As a consequence, AAIW was forced to spread into the eastern Indian Ocean with vigorous flow over Broken Ridge (Fine, 1993;Wyrtki et al., 1971).The decline in winnowing energy over Broken Ridge between 23 and 19.2 Ma coincided with the initial disconnection between the Indian Ocean and the Tethys Sea due to the restriction of the Mesopotamian gateway (Bialik et al., 2019).This restriction resulted in a weakening of the low-latitude circumglobal ocean circulation, and TIW was gradually being cut off from its source.As a consequence, AAIW had the opportunity to occupy a larger part of the intermediate water depth range in the Indian Ocean, and expand to the west.In other words, AAIW could have dissipated its kinetic energy over an increasingly large area, with a reduction in current winnowing over Broken Ridge as a result.
In contrast, the subsequent interval between 19.2 and 16 Ma was characterized by a marked increase in current winnowing.We attribute this increase in winnowing to an eastward shift in the AAIW flow, which would then be again more focused over Broken Ridge.As a result, Broken Ridge would have been increasingly exposed to more energetic oceanic currents.However, the driving factor that caused AAIW shift eastward remains unclear.
A second decline in both winnowing proxies is observed between 15-13.7 Ma, which corresponds with the final closure of the Mesopotamian Seaway around 13.8 Ma (Bialik et al., 2019).This closure led to the complete termination of low-latitude circumglobal ocean circulation, and also to a further weakening of Indian Ocean circulation and a reduction in winnowing over Broken Ridge.Following this evolution, a "quiet period" is discerned between 13.7 and 10.8 Ma.The "quiet period" is marked by higher percentage of finer sediments and a more distinctive nannofossil peak (Figure 7b).The onset of the "quiet period" at Broken Ridge somewhat surprisingly coincides with the MMCT and the establishment of the West Antarctic ice sheet (Levy et al., 2016).Those significant changes in the global climate system led to a stronger meridional temperature gradient and intensification of global ocean circulation.Our results from Broken Ridge thus suggest that the ocean's behavior around this time was actually more heterogenous then previously considered.While most of the surface and upper intermediate water depths currents see remarkable strengthening (Eberli & Betzler, 2019), ODP Site 752 on Broken Ridge experienced relatively weak intermediate currents.We note that this interval is also very stable in terms of productivity due to a simultaneous and similarly large northward tectonic and climate-belt shift.
Winnowing increases again between 10.8 and 7.5 Ma, coinciding with the first short-lived occurrences of TL (Christensen et al., 2021).This body of intermediate water originates from the Pacific Ocean and pass directly by Broken Ridge.Hence, it provides the energy to re-enhance winnowing over Broken Ridge.After 7 Ma, TL no longer occurs as short-lived occurrences but is permanently installed over Broken Ridge, making for continuously-high winnowing between 7.5 and 3 Ma.Finally, at the Pliocene-Pleistocene transition, there is another notable increase in winnowing strength.This evolution coincides with the intensification of glacial-interglacial cycles and the invigoration of global ocean circulation (Kleiven et al., 2002).

Fe-and Mn-Enriched Dust
The presence and abundance of minerals and trace metals in Broken Ridge sediments has been used in several past studies to make inferences over global climate aridity and ocean water oxygenation (e.g., Dickens & Owen, 1994;Hovan & Rea, 1991, 1992).In the framework of this paper, we highlight that the XRF-derived Fe and Mn depth-series agree remarkably well with these previously published records (Figure 8).The XRF-derived iron (Fe) series exhibits similar trends as the eolian component record provided in Hovan and Rea (1991; Figure 8a), with particularly high values during the Middle Miocene and a significant decrease after ∼15 Ma.While Fe counts slightly recover around 3 Ma, both proxy records remain at a low level since the Late Miocene.The manganese (Mn) time-series shows a good agreement with the data generated by Dickens and Owen (1994), which exhibits a similar evolution, with highest counts in the Early and Middle Miocene and much lower values after that.The decline of Mn (Figure 8b) and Mn/Sc ratio in the sediment between 6.5 and 3 Ma has been described to an expansion of the OMZ by Dickens and Owen (1994).However, in the light of our winnowing reconstruction presented-above, the overall decrease in sedimentary Mn content may potentially be driven by the long-term increase in current energy.
The decline of Fe and Mn in the Early and Middle Micoene is intriguing, as simultaneously elevated Fe and Mn counts hint toward submarine hydrothermal activity and the Amsterdam-St.Paul hotspot was only ∼500 km away from Broken Ridge in the Early Miocene (Janin et al., 2011;Maia et al., 2011;Resing et al., 2015).However, Hovan and Rea (1991) did not find volcanogenic materials in their investigated smear slides from ODP Hole 752A.Hence, we adhere to their original interpretation that the mineral component of Broken Ridge Neogene sediments is mostly eolian in nature, and that changes in mineral flux reflect major climatic shifts in the southern African dust source.For example, Hovan and Rea (1991) interpreted the major Early and Middle Miocene spikes as well as the minor peak at ∼3 Ma (Figure 8a) as indicators of the aridification of the dust source region.A more LYU ET AL. 10.1029/2023PA004761 13 of 15 elaborate analysis of mineral sediments provenance and its implications is beyond the scope of this paper, and therefore, requires for future research from ODP Site 752; however, we indicate that the Neogene eolian sediments over Broken Ridge was composed of an Fe-and Mn-enriched dust type.

Conclusions
The main patterns observed in the XRF records of ODP Hole 752A coincide with major Neogene climate evolutionary steps.The MCO is characterized by high variability and generally high productivity index, with productivity of ODP Hole 752A pulsing at the same eccentricity beat as the dynamics of the global climate, ocean and carbon cycle.Following 13.9 Ma and the expansion of the Antarctic ice sheet, ODP Hole 752A experienced a decline in productivity and current winnowing, in conjunction with the northward shift of latitudinal climate belts and oceanographic reorganization.The productivity index stabilized at moderate levels during the Late Miocene and Pliocene.This stabilization is likely due to the combined effect of a northward tectonic movement of the site and a gradual northward shift of the climate belts.In the last 3 Myr, increased productivity and current winnowing proxies is observed.These changes are interpreted to be the result of invigorated ocean circulation at the time of intensifying northern hemisphere glaciation.The XRF-derived Fe and Mn records have particularly high values during the Middle Miocene then decline significantly after ∼15 Ma.These series provide a high-resolution record, probably reflecting eolian dust, in good agreement with previous low-resolution smear-slide analyses (Hovan & Rea, 1991).These higher resolution dust records form the basis for future provenance analyses and a more detailed investigation of eolian dust fluxes.The above-described features are constrained in time by an astronomically-calibrated age-depth model.This provides a reference framework for future paleoclimatic and paleoceanographic studies.
However, to further improve the precision and accuracy of future paleoceanographic studies, particularly those focusing on Milanković (obliquity and precession) timescales, obtaining higher-resolution and fully-continuous sedimentary records is essential.This can be achieved by re-drilling Broken Ridge with state-of-the art coring techniques and strategies.This includes the use of the advanced piston corer and the construction of a stratigraphically complete composite section by splicing (at least) three different holes at the same site.

Figure 2 .
Figure 2. Spectral analysis of Ca/Fe series.(a) Spectral power distribution of the Ca/Fe time-series along the biostratigraphic datums (red dots) with error bars showing depth-uncertainties and the astronomically-tuned age model (blue); (b) Spectral power distribution of the Ca/Fe depth-series.The gray line represents the 99% of confidence interval and the light gray line represents the 95% of confidence interval; (c) Spectral power distribution of the Ca/Fe time-series along the astronomically-tuned age model.The gray line represents the 99% of confidence interval and the light gray line represents the 95% of confidence interval.

Figure 3 .
Figure 3. Astrochronology.(a) Wavelet analysis of detrended Ca/Fe depth-series.The dash line indicates the power at ∼1.6 m/cycle, while the circles indicate the power at ∼0.2 m/cycle.(b) Detrended Ca/Fe depth-series (yellow) and its ∼1.6 m/ cycle filter (gray), and red dots represent the stratigraphic position of biostratigraphic datum; (c) La11 Eccentricity solution as a tuning target; (d) Detrended Ca/Fe time-series according to the astronomically-tuned age-depth model (yellow) and La11 Eccentricity solution (gray).The astronomically-tuned age-depth model is refined by tuning maxima in Ca/Fe depth series into eccentricity maxima; (e) Sedimentation rate.

Figure 7 .
Figure 7. Winnowing.(a) Summary of all XRF-derived elements records (Total counts, beige curve) and grain size record (black curve).Thick arrows mark increasing (green) and decreasing (brown) of current winnowing.TL, Tasman Leakage.(b) Grain size distribution in the logarithmic scale.The black arrow points out an interval with mass transport deposit (Figure S5 in Supporting Information S1).MTD, Mass transport deposit.

Figure 8 .
Figure 8. Fe and Mn enriched dust.(a) XRF-derived Fe (green) compared to the extracted dust content reconstruction by Hovan & Rea (brown, Hovan & Rea, 1991); (b) XRF-derived Mn (green) compared to Mn concentrations as measured by Dickens & Owen (light brown dots,Dickens & Owen, 1994).Arrows point the significant decline in Fe and Mn after ∼15 Ma.

Table 1
Calcareous Nannofossil Datums (Shipboard Scientific Party, 1989)Used for Depth-Age Correlation at Ocean Drilling Program Hole 752A