Climate-induced shift of deep-sea benthic foraminifera at the onset of the mid-Brunhes dissolution interval in the northeast tropical Indian Ocean

The mid‑Brunhes dissolution interval (MBDI; Marine Isotope Stage (MIS) 13 to 7; ~ 533–191 ka) is characterized by vari‑ ous paleoclimatic/paleoceanographic events in the world. We investigated fossil deep‑sea benthic foraminifera and sediment geochemistry at the onset of the MBDI (~ 670–440 ka) using Ocean Drilling Program (ODP) Site 758 and core GPC03 in the northeast tropical Indian Ocean (TIO), primarily focusing on the relationship between the pale‑ oceanographic conditions of the surface and deep oceans. Based on multi‑dimensional scaling, MDS axis 1 is related to the specific depth habitats of benthic foraminiferal fauna, possibly at the trophic level. In MDS axis 1, the difference between the two core sites was smaller from ~ 610 to 560 ka, whereas it was larger from ~ 560 to 480 ka. In contrast, MDS axis 2 may be related to the low food supply at episodic food pulses/relatively stable and low food fluxes. MDS axis 2 showed generally similar stratigraphic variations between the two cores during ~ 610–560 ka, but was differ‑ ent during ~ 560–480 ka. The proportion of lithogenic matter to biogenic carbonate was relatively low from ~ 610 to 530 ka under the highstand when sediment transport to the study area was reduced. Thus, both the depth gradi‑ ent in the distribution of benthic foraminiferal fauna and the lithogenic supply between the two cores changed coincidently across the MIS 15/14 (~ 570–540 ka) transition. Such paleoceanographic conditions across MIS 15/14 transition were attributed to the long‑term weakening of the wind‑driven mixing of surface waters, which might have been caused by the weakening of the Indian summer monsoon in the northeast TIO, possibly with the northward displacement of the InterTropical Convergence Zone in the Northern Hemisphere. In particular, the depth gradient in the distributions of benthic foraminiferal faunas represents the paleoceanographic linkage between the surface and deep oceans through particulate organic matter ballasting by calcareous plankton skeletons in addition to litho‑ genic matter, which changed transiently and significantly across MIS 15/14 transition close to the onset of the MBDI.


Introduction
The duration from Marine Isotope Stage (MIS) 15 to 11 (621-374 ka) is the transitional period between the mid-Pleistocene Transition (MPT, ~ 1250-700 ka; Pisias and Moore 1981) and the mid-Brunhes dissolution interval (MBDI) (MIS 13 to 7;.The MBDI represents global carbonate dissolution, lasting several hundred thousand years, centered around MIS 11 (~ 424-374 ka; Barth et al. 2018) which is characterized by paleoceanographic/paleoclimatic events recognized in the surface and deep oceans (e.g., Jansen et al. 1986;Farrell and Prell 1989).The mid-Brunhes event was defined as a climatic shift, evidenced by marine sediments and Antarctic ice cores, between MIS 12 and 11 at approximately 430 ka, showing an increase in the amplitude of glacial-interglacial cycles (EPICA community members 2004).Recently, the mid-Brunhes Transition (MBT) has been proposed as a two-stage (MBT-1 and MBT-2) global climate shift and these two stages were distinguished by global and regional paleoclimatic/paleoceanographic events (Ao et al. 2020).The first stage (MBT-1) at ~ 500 ka was generally characterized by the regional phenomena during MIS 13 (533-478 ka).In this period, asymmetrical temperature and precipitation between the hemispheres and northward displacement of InterTropical Convergence Zone (ITCZ) in the Northern Hemisphere were reported.By contrast, the second stage (MBT-2) at ~ 400 ka was characterized by more global extents of paleoclimatic/paleoceanographic events from MIS 11.Yu et al. (2017) suggested that MIS 14 (563-533 ka) was uniquely interglacial-like, showing more sluggish Atlantic Meridional Ocean Circulation in the North Atlantic, northward shift of the ITCZ, and southward displacement of westerlies in the southeastern Pacific.In addition, Barth et al. (2018) suggested several sequential steps of global climatic events during MIS 15-11.Thus, the transitional period from MIS 15 to 11 is important in delineating the onset of the MBDI, particularly, focusing on the spatiotemporal variation of paleoclimatic/paleoceanographic events.
Due to differences in insolation, the sea surface temperature (SST) is relatively high in modern tropical oceans.The thermal contrast between the tropical and polar regions is a primary driving force of meridional atmospheric circulation such as the Hadley Circulation (e.g., Farrell 1990).In addition, the east-west gradient of the SST in the tropical ocean often causes interannual variations in atmospheric circulation (i.e., Walker Circulation) such as the El Niño-Southern Oscillation (ENSO) in the equatorial Pacific Ocean (Messie and Chavez 2013).Similar to modern circulation in the high SST tropical oceans (e.g., Fedorov et al. 2015), both the long-term (millions of years scale order) meridional and zonal atmospheric circulations might have been present in the geological past.For example, since the middle Pleistocene, the different decreasing SST trend between the eastern and western equatorial Pacific Ocean has resulted in an ENSO-like asymmetric zonal pattern along the equator through the intensity changes of the trade winds (e.g., Lie and Herbert 2004).The eastern tropical Indian Ocean (TIO), located in the Indo-Pacific Warm Pool (IPWP), plays an important role as a heat source that controls the global climate (e.g., De Deckker 2016), revealing that the spatiotemporal variation of paleoclimatic/paleoceanographic events related to long-term atmospheric circulation change in the TIO during the transitional period (MIS 15-11) is a possible driver of global climatic change.
Because deep-sea biota are sensitive to environmental conditions, as well as plankton and neritic organism, they have been used to evaluate the biodiversity of deep-sea microfossils in response to climatic change (e.g., Yasuhara et al. 2017).Benthic foraminifera provide crucial information on past trophic conditions (e.g., Gooday 2003;Jorissen et al. 2007).Benthic foraminiferal data have also been utilized to evaluate the ballasting effect of particulate organic matter (POM) on food delivery to the seafloor by mineral grains such as plankton skeletons (e.g., Griffith et al. 2021).Takata et al. (2019) studied the benthic foraminiferal faunas in the central equatorial Pacific Ocean and reported that the shift in the ballasting of POM from calcareous to siliceous plankton across ~ 300 ka.Takata et al. (2022) investigated the benthic foraminiferal faunas in the northeast TIO and found that changes in the ballasting of POM by calcareous plankton skeletons occurred at around 370-210 ka.Such a paleoceanographic event may be related to a longterm change in the asymmetric zonation of wind-driven upwelling between the eastern and western TIO, similar to the modern Indian Ocean Dipole mode (Takata et al. 2024).Since the ballasting of POM by mineral grains is an important mechanism for linking the paleoceanographic change between the surface and deep oceans, the ballasting effects of POM can be evaluated by the variation in benthic foraminiferal fauna in the deep-sea environment, which implies that atmospheric circulation changes influence the surface ocean conditions.
In this study, we investigated benthic foraminiferal fauna and measured the major elemental compositions of bulk sediments in terms of the depth transect at Ocean Drilling Program (ODP) Site 758 and core GPC03 in the northeast TIO, adjacent to the IPWP, during MIS 16-12 corresponding to the transitional period between the MPT and MBDI.The main objective of this study was to trace the paleoceanographic event at the onset of the MBDI in the northeast TIO by evaluating the ballasting effect of POM by mineral grains.

Methods
Oceanography in the Bay of Bengal is characterized by the distinct seasonality of wind and precipitation by the Indian monsoon, and the consequent large amount of freshwater discharge, transporting lithogenic particles from the Ganga-Brahmaputra-Meghna (GBM) River system.Rixen et al. (2019) observed seasonal variations in particulate organic carbon (i.e., POM) flux.The summer flux was high at Station SBBT (Southern Bay of Bengal Trap), which is located at a latitude similar to that of our study area (Fig. 1).In addition, Zhang et al. (2022) summarized carbonate sedimentation with water depth in the Indian Ocean, whereas the Bay of Bengal is exceptional area on the relationship between carbonate sedimentation and water depth because of dilution by the abundant lithogenic matter supply from the GBM River system.The Indian summer monsoon is driven by the different heating between the warmer Asian continent and the cooler Indian Ocean with the seasonal movements of the ITCZ (e.g., Webster et al. 1998).
For the faunal analysis of benthic foraminifera, 2-6 g of freeze-dried sediment were processed following the protocol of Takata et al. (2022).More than 200 benthic foraminiferal specimens were picked, using a binocular microscope, in the > 105 µm fraction from split aliquots (1/2 to 1/32) of the washed residues (> 63 µm) that was followed by Nomura (1995).There are several opinions about the suitable size fraction and counting number for faunal analysis of benthic foraminifera (e.g., Boltovskoy and Wright 1976;Thomas 1985).Some scientists have adopted the > 150 µm fraction for benthic foraminiferal fauna, providing convenient identification, whereas other scientists have stated that the > 63 µm is more useful to catch the occurrences of the smaller taxa (e.g., phytodetritus species; Thomas 1985).In addition, many scientists have thought that the 100-300 counts are suitable for the reliable faunal analysis, as Boltovskoy and Wright (1976) mentioned.Nonetheless, there are variable ideas on the suitable size fraction and counting number of benthic foraminifera.In this study, we adopted the > 105 µm fraction and the 200 counts, following the same scheme of our previous studies (Takata et al. 2022(Takata et al. , 2024)).It allows the combination of the previous results with this study, which highlights the difference in paleoceanography between our study period and the younger span.The specimens were identified, and counted, following taxonomic assignments on Boltovskoy (1978), Corliss (1979), van Morkhoven et al. (1986), Jones (1994), andNomura (1995), with the generic classification of Loeblich and Tappan (1987).We combined the new 13 data of core GPC03 with the published 34 data between 841 and 595 cm by Takata et al. (2022).
We examined the structure of community for each sample: rarefaction (diversity E [Sn]; n = 50), Shannon-Wiener function (Hʹ) (Shannon and Weaver 1949), and the evenness of Buzas and Gibson (1969).We conducted non-metric multi-dimensional scaling (MDS) for a data matrix of relative abundance (25 taxa and 63 samples), following protocol and configuration of Takata et al. (2022).Calculation of the rarefaction and non-metric multi-dimensional scaling were performed, using the statistical programming environment R (R Development Core Team 2020) with the function from the Vegan Community ecology package (Oksanen et al. 2019), whereas Shannon-Wiener function (Hʹ) and the evenness were calculated via Microsoft Excel software.
For 34 samples from ODP Site 758, the concentrations of 11 major elements (Na 2 O, MgO, SiO 2 , Al 2 O 3 , P 2 O 5 , S, K 2 O, CaO, TiO 2 , MnO, and Fe 2 O 3 ) were measured from a briquette, using an energy-dispersive X-ray fluorescence (XRF) elemental analyzer (MESA-500W, Horiba Co., Ltd.), following protocols and analytical conditions of Takata et al. (2024).The weight percentages of the 11 major elements were quantified using the standard regression method, based on the 12 standards provided by Dr. H. Fukusawa (former professor at Tokyo Metropolitan University, Japan).As chlorides were not removed from the sediment samples by rinsing with distilled water, the Na 2 O data may be slightly overestimated.
In order to objectively evaluate the variations in the 10 elements (except for P 2 O 5 ), we conducted a R-mode principal component (PC) analysis using the "prcomp" function in the statistical programming environment R (R Development Core Team 2020) based on a data matrix of the 10 elements and 33 samples.A sample at 8.48 mcd of ODP Site 758, which is relatively high SiO 2 probably due to one of thin ash layers (Shipboard Scientific Party 1989), was excluded from the principal component analysis.The correlation coefficient was used to calculate the diagonal matrix.
The δ 18 O and δ 13 C values of benthic foraminifera (Cibicidoides wuellerstorfi; > 250 µm fraction) from 31 samples of ODP Site 758 were measured at the Marine Core Research Institute of Kochi University (Japan).Two to five specimens were analyzed using an Elementar isoprime precisION with the Multicarb preparation system, reacting the samples with 100% phosphoric acid at 90 °C for 10 min.The data were calibrated to the VPDB standard using IAEA-603 and JCp-1 (Geological Survey of Japan) standards.The analytical precision (± 1σ) was ± 0.16‰ for δ 18 O and ± 0.13‰ for δ 13 C. Takata et al. (2022) reported isotopic data for C. wuellerstorfi in core GPC03.In this study, the isotope values of C. wuellerstorfi (> 250 µm fraction) were additionally measured for new 13 samples of core GPC03 at the Oregon State University (USA) following the protocol of Takata et al. (2022).The analytical precision (± 1σ) was ± 0.02‰ for δ 18 O and ± 0.04‰ for δ 13 C.
The orbitally-tuned age models at ODP Site 758 and core GPC03 were established by oxygen isotope stratigraphy.Isotope data at our study sites were standardized (zero mean and unit variance) prior to their correlation to global stack LR04 (Lisiecki and Raymo 2005).Stratigraphic correlation was performed under hidden Markov model (HMM) probabilistic algorithm (Lin et al. 2014;Ahn et al. 2017).This personal computer-based technique provides age model uncertainties of oxygen isotope alignment from every probabilistic match, as shown in Fig. 2. The assessments of age model uncertainty at ODP Site 758 and core GPC03 help heavily on our discussion on up-downs of mass accumulation rate (MAR).
For ODP Site 758, we calculated the MAR of Al 2 O 3 and CaO at ODP Site 758 as follows: Al 2 O 3 -MAR or CaO-MAR (g cm −2 kyr −1 ) = %(Al 2 O 3 or CaO)/100 × LSR × DBD, where LSR is the linear sedimentation rate (cm kyr −1 ) and DBD is the dry bulk density (g cm −3 ) of the sediment.The DBD was obtained from the Shipboard Scientific Party (1989)  samples based on core depths (mbsf ) of the same hole (Hole A).We also calculated the benthic foraminiferal accumulation rate (BFAR; Herguera and Berger 1991), following the equation of Herguera and Berger (1991): BFAR (# cm −2 kyr −1 ) = (the number of benthic foraminifera per unit gram of bulk sediment [# g −1 ]) × LSR × DBD.

Results
The δ 18 O and δ 13 C values of Cibicidoides wuellerstorfi at ODP Site 758 ranged between 2.7-3.9‰ and − 0.5-0.4‰,respectively (Table S1; Fig. S1).On the other hand, those of core GPC03 ranged between 2.6-4.6‰ and − 0.9-0.5‰,respectively.Based on the matching of the oxygen isotope profiles to the LR04 stack, the age-depth curves with 95% confidence intervals for ODP Site 758 and core GPC03 are shown in Fig. 2. Fossil benthic foraminifera were found in all samples in this study.The abundances of benthic foraminifera per unit weight at ODP Site 758 and core GPC03 ranged from 89 to 301 and 63 to 407, respectively (Fig. 3).Oridorsalis umbonatus, Gyroidinoides sp.A, Pullenia spp., Epistominella exigua, Nuttallides umbonifer, and Eilohedra levicula were common constituents in both cores, auxiliary to Uvigerina dirupta, Siphouvigerina ampullacea, Globocassidulina subglobosa, Cibicidoides mundulus, and Astrononion echolsi (Figs. 3,4,5).Obvious features on destruction or abrasion of the benthic foraminiferal tests were not observed both at ODP Site 758 and core GPC03 (Figs. 4,5), similar to the observation of Takata et al. (2022Takata et al. ( , 2024)).These features imply that the faunas are unlikely to be affected by downward transport from other areas to the core sites.In addition, severe carbonate dissolution was not detected on the benthic foraminiferal tests (Figs. 4,5).
The faunal associations of benthic foraminifera were generally similar between the two cores, whereas some taxa differed in their stratigraphic distributions between the cores.For example, the relative abundances of uvigerinids, including Siphouvigerina ampullacea, were more common and O. umbonatus was less common at ODP Site 758 than in core GPC03 (Fig. 3).Both rarefaction (E[S 50 ]) and Shannon-Wiener (Hʹ) of benthic foraminiferal fauna show frequent fluctuations in ODP Site 758 and GPC03 (Fig. 6).In addition, the species diversity indices at ODP Site 758 were partially higher than those of core GPC03, especially at ~ 630 ka and ~ 490 ka.The evenness of Buzas and Gibson (1969) was not markedly different between the two cores but varied significantly with depth (Fig. 6).
Two MDS axes were recognized in the benthic foraminiferal faunas.The scores of MDS axis 1 at ODP Site 758 were generally positive, whereas those of core GPC03 were negative (Fig. 6), suggesting a depth gradient in the distribution of benthic foraminiferal fauna between the two cores.The scores of MDS axis 2 showed similar variation patterns between the two cores during MIS 15, whereas they showed different variation pattern during MIS 14-13 (Fig. 6).MDS axis 1 was positively correlated with Gyroidinoides sp.A and negatively correlated with O. umbonatus and Pullenia jarvisi, whereas MDS axis 2 is positively correlated with C. mundulus and negatively correlated with E. exigua and N. umbonifer (Table S2).
Among the elemental compositions of the bulk sediments at ODP Site 758, SiO 2 and CaO were the dominant components and Al 2 O 3 and Fe 2 O 3 were associated (Fig. 7; Table S1).The stratigraphic variation pattern of %CaO was generally opposite to those of %SiO 2 , %Al 2 O 3 , %TiO 2 , and %Fe 2 O 3 (Fig. 7).According to the principal component analysis, the contributions of PC axes 1, 2, and 3 were 55.9%, 16.1%, and 11.4%, respectively.Due to the similarly low contributions of PC axes 2 and 3, we focused only on PC axis 1 in this study.PC axis 1 was characterized by positive loadings of SiO 2 , Al 2 O 3 , TiO 2 , and Fe 2 O 3 and negative loading of CaO (Table 1).The former combination consists of a common elemental component in lithogenic matter such as illite and smectite, whereas the latter represents carbonates, particularly biogenic skeletons.These results suggest that the variation of PC axis 1 depends on the proportion of lithogenic detrital particles and biogenic carbonates.The PC axis 1 scores were normally more positive during MIS 17-16 and MIS 13-12, whereas they were more negative during MIS 15-14 (Fig. 7).With respect to PC axis 1 loadings, positive scores represented a greater contribution of lithogenic matter, whereas negative scores symbolized more biogenic carbonate contributions.

Faunal composition
We considered the ecological issues on the two MDS axes, based on ecological information on modern benthic foraminifera from Supplementary file 5 of Takata et al. (2022) and the literatures listed in Takata et al. (2024).Based on the correlation between the score on MDS axis 1 and the relative abundance of each taxon (Table S2), MDS axis 1 is negatively correlated to Oridorsalis umbonatus and Pullenia jarvisi and positively correlated to Gyroidinoides sp. A. These cosmopolitan taxa live at various water depths and trophic conditions (e.g., van Morkhoven et al. 1986).Peterson (1984) reported the spatial distribution of modern benthic foraminifera in the eastern equatorial Indian Ocean.According to his faunal list, both O. umbonatus and Gyroidinoides orbicularis (likely equivalent to our Gyroidinoides sp.A) are widely In the present study, the relative abundance of O. umbonatus was common during MIS 17-11 (Fig. 3).Takata et al. (2022) reported that its occurrence generally decreases after MIS 11.Thus, the past habitats of these species might have been different during MIS 16-12 compared to their modern distributions.It may be difficult to specify the exact reason for their different occurrences during MIS 16-12 in the two cores, just based on modern ecological knowledge.In contrast, Siphouvigerina ampullacea was slightly positively correlated with MDS axis 1 (0.46) (Table S2).Other Uvigerina spp.were also more common at ODP Site 758 than at core GPC03 (Fig. 3).Uvigerinids are usually regarded as eutrophic taxa (e.g., Jorissen et al. 1997).Hence, although the interpretation of the ecological issue of MDS axis 1 is still difficult, the negative score on MDS axis 1 was related to deeper water habitats of benthic foraminiferal fauna, especially at deep site (core GPC03) during MIS 16-12, whereas a positive score was related to shallower water habitats possibly with a relatively higher food supply.

MDS axis 2 was negatively correlated with
Epistominella exigua and Nuttallides umbonifer and positively correlated with Cibicidoides mundulus (Table S2).Epistominella exigua is known as a common species with an episodic food supply (Gooday 1994;  Smart et al. 1994;Thomas et al. 1995;Thomas and Gooday 1996;Sun et al. 2006;Thomas 2007;Takata et al. 2016).Many studies have reported that Nuttallides umbonifer occurred at abyssal depths between the lysocline and the calcium carbonate compensation depth (CCD) (Bremer and Lohmann 1982;Mackensen et al. 1990Mackensen et al. , 1993Mackensen et al. , 1995;;Schmiedl et al. 1997).This species also tolerates an extremely low food supply (Schmiedl et al. 1997) in the greater depths where the organic matter reaches only a very small portion from surface ocean to the seafloor in general (Martin et al. 1987;Rigwell 2003;Henson et al. 2012;Griffith et al. 2021).Nuttallides rugosa (probably equivalent to N. umbonifer) was reported at ~ 1916 m water depth of the western Arabian Sea (Kurbjeweit et al. 2000).However, this species is commonly associated with low and seasonal food supplies (Gooday 1994(Gooday , 2003;;Sun et al. 2006), possibly due to its tendency to reproduce in such season.De and Gupta (2010) found N. umbonifera and E. exigua at the abyssal depths of the eastern and central Indian Ocean, which is characterized by carbonate-corrosive cold bottom waters and a low and pulsed food supply.However, C. mundulus is common under oligotrophic conditions (e.g., van Morkhoven et al. 1986), probably with a non-episodic (i.e., relatively stable) food supply, compared to E. exigua.Thus, a negative score on MDS axis 2 is likely related to an extremely low food supply with episodic food pulses, whereas a positive score may indicate a low but relatively stable food supply.
In summary, during our study period (MIS 16-12), MDS axis 1 was related to the specific depth habitats of benthic foraminiferal faunas possibly with trophic conditions in the tropical Indian Ocean, whereas MDS axis 2 may be related to food supply conditions.Thus, benthic foraminiferal fauna during MIS 16-12 seemed to be mainly influenced by water depth and episodic states of the food supply from the surface waters.

Depth gradient in the distributions of benthic foraminiferal faunas
Rarefaction (E[S 50 ]) and the Shannon-Wiener index (Hʹ) at ODP Site 758 decreased slightly from late MIS 14 to early MIS 13, whereas those of core GPC03 did not vary distinctly (Fig. 6).The long-term changes in the species diversity of benthic foraminifera differed between the two sites.This suggests that the long-term changes in species diversity of benthic foraminifera occurred mainly at the shallow site (ODP Site 758; 2925 m water depth) rather than at the deep site (GPC03; 3650 m water depth).The difference in MDS axis 1 between the two cores was generally large during glacial periods (MIS 16 and 14; Fig. 8).In particular, MDS axis 1 during MIS 15 was similar, whereas that during MIS 14-13 was different (Fig. 6).Thus, the large difference in MDS axis 1 between the two cores seems to indicate an increased depth gradient in the distribution of benthic foraminiferal fauna.In contrast, MDS axis 2 showed a generally similar pattern of variation (small variation in the difference in MDS axis 2 in Fig. 8) between the two cores before MIS 15 (Fig. 6), whereas it was different (large variation in the difference in MDS axis 2 in Fig. 8) during MIS 14-13 (Fig. 6).Thus, based on both the structure of the community and the faunal composition represented by MDS axis 1, the depth gradient in the distribution of benthic foraminiferal faunas between ODP Site 758 and core GPC03 increased during MIS 14-13.The lateral distance between ODP Site 758 and core GPC03, located at ~ 5ºN is quite short (~ 168 km).Although we cannot dismiss the possibility of lateral distance on the oceanographic difference, it is reasonable to assume that the depth gradient in the distribution of benthic foraminiferal faunas seems to result from different water depths (~ 730 m difference) between the two sites.Such a depth gradient in the distribution of benthic foraminiferal faunas is difficult to comprehensively interpret based on modern ecological knowledge.However, this may be partly explained by the changes  (Spratt and Lisiecki 2016), the differences of the scores of MDS axis 1 and 2 between ODP Site 758 and core GPC03, the relative abundance of Globigerina bulloides at ODP Site 758 (Chen and Farrell 1991), and relative abundance of G. bulloides (Bhadra and Saraswat 2022) and stable oxygen isotope variations of seawater that was estimated by oxygen isotope values of planktonic foraminifera and TEX 86-sea surface temperature at IODP Site U1446 (Clemens et al. 2021)   in water mass during the MIS 16-12.Peterson and Prell (1985) reported that the Composite Dissolution Index (CDI) was more negative (i.e., more carbonate-corrosive condition) in the equatorial Indian Ocean during MIS 13-11.The boundary between the Indian Deep Water (IDW) and Indian Bottom Water (IBW) is currently roughly consistent with the lysocline (~ 3900-3800 m water depth) at present (Peterson 1984) where the CDI is almost 0 (Peterson and Prell 1985).Thus, during MIS 13-12, the boundary between the IDW and the IBW might be located shallower than the core GPC03 site (3650 m water depth), compared to the present day.Thus, the depth gradient in the distribution of benthic foraminiferal faunas between ODP Site 758 and core GPC03 was enhanced during MIS 13-12.Howe and Piotrowski (2003) reported stratigraphic changes in ε Nd at ODP Site 929 (western equatorial Atlantic) over the last ~ 800 ka (Fig. S2).The ε Nd value in the Atlantic Ocean represents the proportion of AABW to NADW.This fluctuation in ε Nd clearly indicates the orbital-scale glacial-interglacial variations, similar to LR04 stack of benthic δ 18 O values (Lisiecki and Raymo 2005).However, they did not show a marked long-term tendency during the MBDI.On the other hand, Barth et al. (2018) reported a larger gradient of benthic Δδ 13 C values between the North Atlantic and South Atlantic during the "pre-MBT (mid-Brunhes Transition)" than "post-MBT" (Fig. S2).They regarded the larger gradient of benthic Δδ 13 C values as the northward penetration of AABW and the lower gradient as the southward penetration of NADW.They recognized that the gradient of benthic Δδ 13 C values between the North Atlantic and South Atlantic changed at the beginning of MIS 12 (~ 436 ka), and this transition appears to be a gradual change during MIS 14-12 than an abrupt one (Fig. S2).This onset timing is generally consistent with our observation of an enhanced depth gradient of the MDS axes across MIS 14.Although careful interpretation is necessary, based on the Δδ 13 C values between the North Atlantic and South Atlantic which indicate the different proportion of deep-water formation between the AABW and NADW, the enhanced depth gradient in the distribution of benthic foraminifera across MIS 14 might be related to the global deep-water circulation.
In summary, the specific depth distribution of benthic foraminifera represented by MDS axis 1 between ODP Site 758 and core GPC03 seems to be related to the shallower position of the boundary between the IDW and IBW in the Indian Ocean during MIS 16-12, compared to the modern ocean.In addition, changes in the depth distribution of benthic foraminifera across MIS 14 may reflect changes in global deep-water circulation.

Ballasting effect of particulate organic matter by mineral grain in the northeast tropical Indian Ocean prior to the onset of the mid-Brunhes dissolution interval
Sediment trap studies have reported the ballasting effect of particulate organic matter (POM) by mineral grains in the water columns of modern ocean (Francois et al. 2002;Henson et al. 2012;Rixen et al. 2019).Ballasting of POM by biogenic carbonate grains (calcareous plankton skeleton) may increase the transfer efficiency of POM from the surface waters to the seafloor (Francois et al. 2002).Benthic foraminiferal data have been used to evaluate the ballasting effects of POM by mineral grains (Diester-Haass and Faul 2019; Griffith et al. 2021;Takata et al. 2019Takata et al. , 2022)).Takata et al. (2022) reported a positive correlation between BFAR and carbonate-MAR in core GPC03 from ~ 600-370 ka (MIS 15-11) and after ~ 210 ka (mid-MIS 7), which suggests that calcareous plankton skeletons (probably mainly calcareous nanoplankton) affected the transfer efficiency of POM through the ballasting effect of POM in the northeast TIO.In our study interval , the correlation between the BFAR and CaO-MAR was positive (r = 0.66, P < 0.01) at ODP Site 758 (Fig. 8).Since the contribution of Ca-bearing silicates (e.g., illite) was very minor compared to biogenic carbonate (Takata et al. 2022), the positive correlation implies that calcareous plankton skeletons seem to promote the better transfer of sinking POM in the water column as ballast at ODP Site 758 during our study period .
In our study area, lithogenic matter was affected mainly by riverine input from the GBM Rivers (e.g., Zhang et al. 2022).In addition to calcareous plankton skeletons, lithogenic matter is thought to be an efficient ballasting material for POM sinking (Rixen et al. 2019).Takata et al. (2024) reported a positive correlation between %N.umbonifer and the contribution of lithogenic matter during the last ~ 450 ka, confirming the increasing ballasting effect of POM by lithogenic matter in the northeast TIO.At ODP Site 758, the correlation between BFAR and Al 2 O 3 -MAR was positive (r = 0.79, P < 0.01) during MIS 16-12 (Fig. 8).In addition to calcareous plankton skeletons, lithogenic matter such as illite-containing aluminum seemed to intensify the transfer efficiency of sinking POM in the water column as ballast at ODP Site 758 during MIS 16-12.Particularly, during MIS 15-14, PC axis 1 on the major elemental compositions related to lithogenic matter and biogenic carbonate was largely negative (Fig. 8), suggesting that the contribution of lithogenic matter was relatively low, compared to biogenic carbonate.Because this period is characterized by a relatively high sea level (Spratt and Lisieki 2016;Fig. 8), the low contribution of lithogenic matter is due to the distance from the mouths of the GBM Rivers to our study area.
The variation in PC axis 1 of the major elemental compositions showed short-term fluctuations during MIS 15 (Fig. 8).The timing of the negative scores of PC axis 1 in this period generally corresponded to relatively high sea levels at ~ 610, ~ 590, and ~ 570 ka (Fig. 8).In addition, the scores of MDS axis 2 of benthic foraminifera were more positive and increased twice at ~ 590 ka and ~ 570 ka at both ODP Site 758 and core GPC03 (Fig. 6).It is reasonable to suppose that the small fluctuations of lithogenic matter supply may have resulted in more shortterm positive scores on MDS axis 2 at both cores under the decreasing supply of lithogenic matter during MIS 15 due to the highstand.This low supply of lithogenic matter during MIS 15 may be attributed to frequent fluctuations in the episodic/more stable food supply from the surface waters to the seafloor.
The greater influence of lithogenic matter during MIS 14-13 may also explain the different variation patterns of MDS axis 2 between ODP Site 758 and core GPC03 (Fig. 8) through the efficiency of sinking POM in the water column.A shift to different variation patterns of the scores of MDS axis 2 between the two cores during MIS 14-13 implies that the food supply became more episodic (e.g., more seasonal variation) at the shallower site (ODP Site 758) than at the deeper site (core GPC03) (Fig. 8).The faunal transition from MIS 15 to MIS 14-13 may be attributed to the change from a more stable food supply to a more episodic food supply from the surface waters.
However, the relationship between PC axis 1 and %N.umbonifer at ODP Site 758 (Fig. 2) was not strong during our study period , compared with the younger interval (MIS 10-1) in core GPC04 which is located close to ODP Site 758 (Takata et al. 2024).The mode of the ballasting effect of POM by lithogenic matter in our study area seemed to differ between MIS 16-12 and the younger period (MIS 8-1).Thus, this seems to support the inference of Takata et al. (2024) that the influence of riverine lithogenic matter supply became significant for the vertical transportation of POM in our study area due to increased sea level lowering during glacial periods since MIS 8.

Paleoceanography in the northeast tropical Indian
Ocean at the onset of the mid-Brunhes dissolution interval Chen and Farrell (1991) reported planktonic foraminiferal fauna during the last ~ 800 ka at ODP Site 758 (Fig. 8).According to their faunal data, the %Globigerina bulloides declined across MIS 14, despite the possible severe carbonate dissolution to planktonic foraminiferal fauna during MIS 15-14.This species is common in the upwelling regions of low-latitude oceans (e.g., Thiede 1975).Maeda et al. (2022) conducted a sediment trap study at Station CBBT-N (southwestern Bay of Bengal) and reported that G. bulloides was common during the monsoon periods (summer and winter) owing to the enhanced surface and subsurface primary production.This ecological evidence suggests that the decline of %G.bulloides at ODP Site 758 is related to decreased primary production due to the weakening of wind-driven mixing in the surface waters of the northeast TIO across MIS 15/14 transition.In contrast, Bhadra and Saraswat (2022) reported that %G.bulloides declined during MIS 14 at IODP Site U1446 (northern Bay of Bengal; 1441 m water depth), similar to ODP Site 758 (Fig. 8), despite highamplitude short-term variations.They interpreted that the long-term declining trend of this species across MIS 14 was related to Indian summer monsoon variability.In addition, the estimated δ 18 O values of seawater increased generally from MIS 15 to MIS 13, confirming the weakening of the Indian summer monsoon (Fig. 8; Clemens et al. 2021).Thus, the long-term declines of %G.bulloides at both IODP Site 1446 and ODP Site 758 in the Bay of Bengal suggests that the weakening of the Indian summer monsoon occurred in our study area across the MIS 15/14 transition, which might have caused the long-term weakening of wind-driven mixing in the surface ocean.Across MIS 15/14 transition, changes in PC axis 1 (the proportion of lithogenic matter to biogenic carbonate) at ODP Site 758 and the depth differences in MDS axes 1 and 2 (benthic foraminiferal faunas) between ODP Site 758 and core GPC03 were observed (Fig. 8).Both changes between surface water oceanography (i.e., food production) and deep-sea biota (i.e., benthic foraminifera) were altered temporally and significantly across MIS 15/14 transition, possibly through the ballasting effect of POM by calcareous plankton skeletons and the lithogenic matter.Yu et al. (2017) reported that the "interglacial-like MIS 14" was a unique paleoceanographic setting, such as more sluggish Atlantic Meridional Ocean Circulation, and the northward shift in the ITCZ in the Northern Hemisphere.The weakening of the Indian summer monsoon across MIS 15/14 transition also seems be the paleoceanographic condition of "interglacial-like MIS 14" in the northeast TIO.The northward displacement of the ITCZ during MIS 13 (MBT-1) proposed by Ao et al. (2020) may also be related to the weakening of the Indian summer monsoon and subsequent faunal transition of deep-sea benthic foraminiferal in the northeast TIO across MIS 14.The weakening of the wind-driven mixing with the Indian summer monsoon is usually related to more southward seasonal movement of the ITCZ at the present day.On the other hand, the long-term weakening of the wind-driven mixing in the northeast TIO with the Indian summer monsoon (as shown by G. bulloides at ODP Site 758 and IODP Site 1446) across MIS 15 to 13 transition seemed to be related to the global climatic change resulting from the asymmetric climatic variations between the Northern and Southern Hemispheres across MIS 14 (e.g., Hao et al. 2015).During this period, Ao et al. (2020) reported the northward displacement of the ITCZ in the Northern Hemisphere.Yu et al. (2017) proposed the two hypotheses for the unique paleoceanographic setting of "interglacial-like MIS 14. " The first hypothesis concerns on the influence of more nutrient-rich cold water from high latitudes via the Peru-Chile Current in the eastern equatorial Pacific (EEP).In addition, the magnitude of the EEP cold tongue weakened in the EEP through the atmospheric bridge of a northward ITCZ shift and the resulting heat/moisture transport across the eastern equatorial Atlantic (EEA).Another hypothesis is that the unique orbital variations in MIS 14 that were characterized by higher obliquity fluctuations and lower eccentricity/precession, resulting in an inter-hemispheric asymmetry.Both hypotheses can be applied to the northeast TIO to change the surface water oceanography and deep-sea benthic foraminiferal fauna.Barth et al. (2018) summarized the multi-proxy data of multiple core sites in the world's ocean from the last ~ 800 ka by conducting principal component analyses.More positive excursion of δ 13 C values of benthic foraminifera was distinct during MIS 13 (their PC axis 1 in Figs. 6 and 7).Such positive δ 13 C excursion is significant in the Atlantic Ocean, but it is not obvious in the Pacific Ocean.More positive shift was not also distinct in our δ 13 C values of C. wuellerstorfi in core GPC03 (Fig. S2).More positive excursion of δ 13 C values of benthic foraminifera during MIS 13 was attributed to the unique setting of export productivity between the Antarctic zone and Subantarctic zone in the Southern Ocean (as shown in Fig. 9 of Barth et al. 2018).Barth et al. (2018) suggested that, during MIS 15-13, a build-up of Northern Hemisphere biomass as a result of accumulation of isotopically light carbon and enriched 13 C in the ocean due to Asian summer monsoon.In the Bay of Bengal, both %G.bulloides and estimated δ 18 O values of seawater at IODP Site 1446 (Bhadra and Saraswat 2022;Clemens et al. 2021) imply stronger Indian summer monsoon during MIS 15 to 13 (Fig. 8), although their interpretation was partly inconsistent to the weakening of the Indian summer monsoon across MIS 15/14 transition in our study area.Further data evaluation of the Asian summer monsoon during MIS 15-13 is necessary to confirm Barth et al. (2018), who proposed sequential steps of paleoceanographic events related to the carbon cycle during the MBDI.Finally, Takata et al. (2024) suggested that the benthic foraminiferal fauna in the northeast TIO could have been influenced by the Indian Ocean Dipole modelike long-term transition during late MIS 9-8, following the scheme of Gupta et al. (2010).Similar paleoceanographic events may have affected the long-term changes in surface-and deep-sea conditions in the northeast TIO at the onset of the MBDI.

Conclusions
This study investigated fossil benthic foraminifera and sediment geochemistry at ODP Site 758 and core GPC03 in the northeast tropical Indian Ocean during MIS 16-12 and led to the following conclusions: 1. MDS axis 1 is related to the specific depth habitats of benthic foraminiferal fauna, possibly to the trophic condition in the TIO during MIS 16-12.The difference in MDS axis 1 between the two sites was smaller during MIS 15, but it was larger during MIS 14-13 because of the depth gradient in the distributions of benthic foraminiferal faunas.MDS axis 2 may be related to the low food supply with episodic food pulses/relatively stable low food flux.The variation pattern of this axis was generally similar during MIS 15 between the two cores but different during MIS 14-13.This indicates that the depth gradient between these two cores was intensified during MIS 14-13, which may be attributed to the shoaling boundary between the Indian Deep Water and Indian Bottom Water.Such a paleoceanographic event might be related to deep-water circulation changes in the Atlantic Ocean during the same period.2. During the highstand across MIS 15/14 transition, changes in the proportion of lithogenic matter to biogenic carbonate represented by PC axis 1 at ODP Site 758 and the depth gradient in the distributions of benthic foraminiferal fauna between ODP Site 758 and core GPC03 were discernible.During such paleoceanographic events, the linkage between the surface and deep oceans was altered temporally and significantly, particularly across MIS 15/14 transition, mainly due to the ballasting effect of POM by calcareous plankton skeletons in addition to lithogenic matter.3. The decrease of %Globigerina bulloides across MIS 15/14 transition represents the reduced primary production due to the weakening of wind-driven mixing in the surface ocean in our study area.Such paleoceanographic condition has been attributed to the long-term weakening of the wind-driven mixing in the northeast TIO with the Indian summer monsoon, possibly due to the northward displacement of the InterTropical Convergence Zone across MIS15 to 13.The long-term weakening of the Indian summer monsoon across MIS 15/14 enhanced the depth gradient in the distribution of benthic foraminifera in the northeast TIO.

Fig. 1
Fig. 1 Locality maps of our study area with ODP Site 758 and core GPC03 ([a] and its close-up [b])

Fig. 2
Fig. 2 Age-depth plots at ODP Site 758 (a) and core GPC03 (b).Dashed lines represent both the upper and lower 95% confidence ranges

Fig. 3
Fig.3Occurrences of benthic foraminifera.The abundance of benthic foraminifera per unit weight (BFN) and relative abundances of selected species of benthic foraminifera at ODP Site 758 (orange line) and core GPC03 (blue line) were shown

Table 1
Loadings of ten elements for PC axis 1 at ODP site 758 (See figure on next page.)Fig. 8 Stratigraphic variations of the several indices in this study.Mass accumulation rate (MAR) of CaO and Al 2 O 3 and benthic foraminiferal accumulation rate (BFAR), sea-level change