According to the sedimentary record, it is possible to distinguish three different periods in both cores: the glacial period with a higher terrigenous supply, a brief transition period, and the Holocene with a lower terrigenous supply.

The FF positively correlated with the terrigenous sediment supply proxies (Figures 3A–F; Tables 3, 4). The highest percentage of the FF occurred when sea level dropped. In contrast, the carbonate content had an opposite correlation and distribution with the FF (Figures 3E–H; Tables 3, 4). Through these distributions, we can infer that the abundance of coccoliths did not cause changes in the FF. The primary process involved in these changes in the FF was the contribution of external silt–clay minerals associated with terrigenous supply, as observed by Costa et al. (2016).

The elementary ratio data show that the terrigenous sediment supply was maximal during the LGM (Figures 3A–D). This more significant terrigenous sediment supply possibly caused a dilution in the carbonate content values because they were minimal despite the higher paleoproductivity (Figures 6C–H), which was also observed by Arz et al. (1998) and Mahiques et al. (2007). After the end of the LGM, at approximately 19 kyrs BP, the terrigenous supply started to decline in the same proportion as the FF. An increase in the coarse fraction could indicate more marine influence (CaCO₃) or an increase in primary productivity because foraminiferal tests were the main component of this fraction. Both processes were observed in this period (Figures 3C–H), suggesting that deglaciation was a transition period of abrupt changes in the environment and deposition of sediments.

During the Holocene, the variability in terrigenous supply was stable compared to the differences between glacial and interglacial periods (Figures 3A–D). The stability tendency of carbonate variation (Figures 3G,H), even with variable biological productivity (Figures 6E–H), particularly in CoreGL-1109, also indicates that terrigenous sediments diluted the carbonate content.

The processes behind these observations may have been the variations in sea level and/or precipitation rates, which would have interfered with the riverine input and consequently in terrigenous supply and primary productivity, allowing a higher concentration of coarse fraction organisms. Analyzing the paleoproductivity record, we observed less productivity during this period (Figures 6E–H); therefore, we infer that a combined effect of sea level and precipitation was probably the main reason for changes in terrigenous input, FF and CaCO₃ content.

The proxies for terrigenous sediment supply may vary mainly due to three processes. The first is relative sea-level variation. Its rise would make the transport of sediments to the upper continental slope more complicated since it would increase the distance between the core area and the sediment source, covering the shelf (Mahiques et al., 2004; Nagai et al., 2010). The second might be variation in fluvial discharge. In other words, with more precipitation, a higher terrigenous sediment supply would lead to higher values of these proxies during humid periods (Behling et al., 2000; Costa et al., 2016). The third might be percentage dilution of the continental material due to mixing with biogenic marine particles/material.

The terrigenous supply had maximum values during the LGM when sea level reached its lowest position (Figures 6C,D,I). This indicates that a major continental shelf area was exposed, resulting in a more extensive area for weathering and a smaller distance between the sediment source and the study sites. Therefore, the continental slope was more likely to receive terrigenous sediments. Thus, the lower relative sea level contributed to more intense deposition of terrigenous and FF sediments (Figures 3E,F and Figure 6I). This could also have been connected to the offshore displacement of the BC to deeper regions during the LGM (Viana et al., 1998; Mahiques et al., 2007; Kowsmann et al., 2015), which would have lowered the local hydrodynamics and allowed or enhanced the deposition of FF sediments.

There was a stable sea level during the Holocene. Nevertheless, the input of terrigenous sediments showed a slight increase. This was linked to an increase in rainfall that occurred during this period (Figures 6C,D,I,J), suggesting that precipitation was directly responsible for these oscillations during this specific period.

Wang et al. (2007) analyzed precipitation during the study period based on δ¹⁸O data from speleothems. These data showed variability similar to the variation in the FF (Figures 3E,F and Figure 6J), especially in GL-824 core, probably because the primary source of fine sediments was fluvial runoff, which was higher when precipitation rates were higher. According to Behling et al. (2000), when there are periods with more precipitation, there are increases in the Fe/Ca and Ti/Ca ratio values. This was also observed in our study (Figures 6C,D,J). In other words, precipitation can be a controlling factor in the terrigenous supply. During the LGM, humidity was higher in South America than during the Holocene (Cruz et al., 2005; Wang et al., 2007; Sylvestre, 2009); therefore, fluvial runoff would have been higher thus transporting more terrigenous sediments to the continental slope. According to Jennerjahn et al. (2004) and Jaeschke et al. (2007), this increase in precipitation was related to the displacement of the ITCZ to the south and an increase in the intensity of winds from the southeast during colder periods, also increasing the humidity on the continent.

Comparing the precipitation oscillation to the paleoproductivity and terrigenous supply proxies, we observed that the general trend was similar between these indicators (Figures 6C–J). However, when precipitation had higher peaks without substantial sea-level variation, we did not observe significant changes in any other proxies, highlighting that rainfall may have influenced terrigenous supply and FF records, but on a smaller scale. The moisture source can also influence the δ¹⁸O speleothems record and not only the precipitation intensity (Lee et al., 2009). Furthermore, there are differences between LGM rainfall reconstructed by nonisotope proxies and interpretations of speleothem records (Sylvestre, 2009; Berman et al., 2016). Thus, we interpreted the maximum terrigenous supply to the upper slope during the glacial period to have been primarily due to the effects of lower sea level. Besides, the possible rise in precipitation could also be improving the terrigenous supply.

Several researchers have previously observed the influence of Heinrich events in the Southern Hemisphere in the tropical region of the South Atlantic (Arz et al., 1999; Vidal et al., 1999; Jennerjahn et al., 2004; Jaeschke et al., 2007), yet in our study, we did not observe a significant influence of Heinrich events or even of the Younger Dryas (Figures 6C,D). These events could have led to the movements of the ITCZ, causing anomalously higher precipitation rates in southeastern Brazil (Cruz et al., 2006; Wang et al., 2007). This should have raised fluvial runoff, transporting more terrigenous sediments and nutrients to the core region. Nevertheless, we did not observe such changes in either of these records, indicating that the main driver of most changes in the deposition of terrigenous sediments was probably relative sea-level variation (Figure 7).

Changes in sedimentary proxies can also indicate changes in the hydrodynamics of the region. A low-energy environment would allow more deposition of silt–clay minerals than a high-energy environment. This process has already been correlated with the displacement of the BC toward the coast, which resulted from the higher relative sea-level conditions and the covering of the continental shelf, as has also been described by other authors (Viana et al., 1998; Mahiques et al., 2007; Kowsmann 2015). The offshore displacement of the BC system can lead to more fine sediments depositing on the upper slope (Viana et al., 1998; Mahiques et al., 2007; Nagai et al., 2010; Kowsmann et al., 2015). This displacement would diminish the local hydrodynamics and promote the deposition of the FF. Furthermore, the “floor polisher” effect on the seabed described by Mahiques et al. (2004) would prevent mud deposition on the shelf break and upper slope depths in the periods where the BC moved toward the coast.

The XRF and FF records of the two cores have different trends. GL-824 shows a decrease over the last deglaciation, while GL-1109 remains constant. The decreasing trend in GL-1109 is observed during the early Holocene. Additionally, in GL-824 core, there was a more significant amount of FF and terrigenous sediments and lower percentages of CaCO₃ than in GL-1109 core (Figures 3E–H), highlighting differences in sedimentary processes between the two cores. This is primarily associated with their different water depths (Figure 1B). GL-824 location is more susceptible to sea-level variations because this core is located on the upper slope, an environment that during the LGM would be analog to the continental shelf. Consequently, the coastal processes influence more the GL-824 than the GL-1109. Thus, its elements and granulometry change as soon as the sea level rises. On the other hand, in the GL-1109, the trends of the terrigenous supply and FF occur later when the sea level is nearly at its maximum. GL-824 core would also have been less affected by the action of the BC compared with GL- 1109. Moreover, GL-824 core is farther north than GL-1109 core, placing it closer to sediment sources such as Rio Doce and Rio Paraíba do Sul.

The relative abundance of F. profunda was opposite to the total organic carbon (TOC) content (Lourenço et al., 2016) and the G. bulloides percentage (Pereira et al., 2018). Additionally, it was similar to higher SSTs (Lourenço et al., 2016) and the relative abundance of the oligotrophic G. sacculifer (Figures 5A–I, Pereira, et al., 2018). Therefore, we used this species to indicate oligotrophic conditions, which has also been corroborated by other studies (Beaufort et al., 2001; Hernandez-Almeida et al., 2019).

Emiliania huxleyi and F. profunda showed a similar general fluctuation (Figures 5A–D). E. huxleyi is known to tolerate a wide range of ecological conditions (Okada and Honjo, 1973; Okada and McIntyre, 1979; Boeckel and Baumann, 2008) and can dominate in both oligotrophic (Okada and Honjo, 1975; Kleijne, 1993; Haidar and Thierstein, 2001; Tyrrell and Merico, 2004) and eutrophic environments (Brand, 1994; Young, 1994; Boeckel et al., 2006). In contrast, Gephyrocapsa spp. exhibits an inverse distribution compared to F. profunda (Figures 5A–F; Tables 3, 4), indicating the different environmental preferences of these species. In the Iberian margin upwelling region, E. huxleyi and Gephyrocapsa spp (mainly small Gephyrocapsa and G. oceanica) were both indicators of upwelling periods (Ausín et al., 2018). However, they were separated into distinct ecological preferences, with E. huxleyi related to a more stable, warmer, and nutrient-poor water column associated with the upwelling relaxation stage and Gephyrocapsa spp. to colder water and higher nutrient availability associated with the early stages of the upwelling event (Ausín et al., 2018; Jin et al. (2019) in the South China Sea also found that Gephyrocapsa spp. were dominant in coastal water associated with higher diatom production and increased silicon content, while E. huxleyi was associated with a regular nutrient regime with lower amounts of silicon.

Comparing our record to SSTs estimated by Lourenço et al. (2016), we analyzed the same process described by Jin et al. (2019), with E. huxleyi dominating the period with warm and stable water and Gephyrocapsa spp. with colder nutrient-rich (silicon) water of the CW (Figures 5C–F and Figures 6B–F). Additionally, E. huxleyi had higher abundances in the periods with lower riverine influence, more negligible rainfall, and higher sea level, interpreted here as a low nutrient period, which was similar to F. profunda and the subtropical species that are associated with high temperature and oligotrophic environments (Boeckel et al., 2006). In this work, we interpreted the variations in E. huxleyi with a possible relation to more oligotrophic environments and as a response to evolutionary factors. This species first appeared and rapidly grew in the late/middle Quaternary, replacing the former more abundant Gephyrocapsa spp. (Toledo et al., 2016), making it a species with a signal that is difficult to deduce. Furthermore, as a cosmopolitan species, it may not indicate upwelling processes but is more typical of a stable regime, as observed in other studies (Andruleit and Rogalla, 2002; Jin et al., 2019).

Since Gephyrocapsa spp. and E. huxleyi seem to represent different ecological aspects, the N ratio (Flores et al., 2000) is not the best proxy to estimate paleoproductivity in this region. Therefore, the leading proxy used in this study to infer changes in primary productivity in surface water was the contrast between eutrophic Gephyrocapsa spp. and F. profunda, which represents low nutrients in surface water.

Variations in the relative abundances of F. profunda and Gephyrocapsa spp., particularly in GL-1109, were very similar to TOC content and G. bulloides relative abundance (Figures 5A–H), showing higher nutrients/productivity during the glacial period and lower nutrients/productivity during the Holocene, with a transition period of ∼19 kyrs BP to ∼14 kyrs BP. These observed variations likely reflect the opposite trend in the relative abundances of F. profunda and Gephyrocapsa spp., which reveals a change in the upper water column during these periods, with the LGM period characterized by a shallower nutricline and high productivity and the interglacial period characterized by a deeper nutricline and lower productivity. This is also supported by the UPZ species in the subtropical group, an indicator of oligotrophic surface environments, being more abundant in the Holocene than in the LGM, suggesting a warmer and poor-nutrient surface ocean during the Holocene. Nagai et al. (2010), mainly applying benthic foraminifera, and Mahiques et al. (2007), using sedimentary data, also inferred high productivity during the LGM on the southeastern Brazilian upper slope. Toledo et al. (2008) reported evidence of decreased productivity over the Holocene compared to the LGM along the Brazilian continental margin, which is consistent with our observations.

The terrigenous supply and FF content oscillated similarly to the Gephyrocapsa spp. record, in contrast to the subtropical group (Tables 3, 4). Three processes could lead to this configuration, or a combination of all of them: 1) there was high availability of nutrients in the UPZ during the low stands of sea level, caused by a higher riverine input transporting more nutrients along with terrigenous material. 2) Higher influence of coastal nutrient-rich water due to the offshore displacement of the BC during low stands of sea level, leading to withdrawal of the surface warm oligotrophic TW (Figure 7). 3) High turbidity in the UPZ due to higher FF content and terrigenous sediments in the water column decreasing the light availability to the LPZ, limiting the growth of F. profunda.

Gephyrocapsa spp. and EPP indicators reflect the same three distinct periods noted in the geochemical and sedimentary record. This suggests that a more substantial continental influence or the BC offshore displacement during the glacial period would have enhanced nutrient availability in surface water. In the same way, during the LGM, a weaker BC would have had more eddies and meanders because stronger currents were closer to linear fluxes. These meanders could have transported more nutrients to the photic zone from greater depths if they were anticyclones or by mixing ocean water with shelf water, which contains higher nutrient concentrations. The Plata River plume could also have reached the study area more frequently during winter-like conditions, promoting higher nutrients and terrigenous input.

Guerreiro et al. (2013) observed a more significant presence of G. oceanica in the water column associated with higher productivity related to fluvial discharge on the central Portuguese margin. Mahiques et al. (2007) and Nagai et al. (2010) suggested an increase in water column temperature and more intense action of the BC during periods of higher sea level, indicating a displacement of the warm water of the Brazil Current toward the coast, which would have prevented any increase in water productivity or the deposition of organic matter.

The correlation between the terrigenous supply, productivity, and nutrient content in surface water, represented by Gephyrocapsa spp. (Table 2), can be explained by two factors: 1) there was probably high nutrient content in the water column transported by enhanced continental runoff during lowstands of sea level and more winter-like conditions increasing the influence of the Plata River plume (Portilho-Ramos et al., 2019); and 2) the displacement of the BC toward the coast during higher sea level enhanced the influence of the TW, transporting more oligotrophic water, deepening the nutricline and preventing any increase in nutrient content (Figure 7). This shows that in the study region, Gephyrocapsa spp. is a more opportunistic species than E. huxleyi, especially when there is a more substantial influence of coastal water because G. oceanica tends to dominate such regions (Guerreiro et al., 2013; Ausín et al., 2018). Furthermore, all processes could act simultaneously, increasing continental runoff related to enhanced rainfall (Wang et al., 2007) and favorable conditions for the northward penetration of the Plata River plume, as also suggested by Portilho-Ramos et al. (2019).

During periods of lower sea level, the input of nutrients and terrigenous materials favored the productivity and deposition of organic carbon (Figures 6A–I), which was observed by Lourenço et al. (2016). Lower CaCO₃ content corresponded to the period of higher SSTs, productivity, and terrigenous supply (Figures 3G,H and Figures 6A–H), which demonstrates an inverse correlation between the productivity record and carbonate content (Tables 3, 4), suggesting a possible dilution effect of the carbonate content by terrigenous input. Along with this dilution during periods of higher terrigenous input, there was an increase in FF content, increasing turbidity and lowering light penetration in the LPZ, limiting the growth of F. profunda and enhancing the relative abundance of the UPZ dweller Gephyrocapsa spp.

The deglacial transitional period was characterized by a decrease in paleoproductivity and terrigenous supply (Figures 6C–H). This was probably related to the processes discussed above (i.e., sea level rise and diminished precipitation), both limiting the nutrient input to the study area from both sources: continental input (regional input or lower Plata River plume influence) and the closer BC system increasing the oligotrophic TW influence. In addition, sea level rise may have modified the region’s geomorphology, limiting shelf-break upwelling to areas closer to the coast.

As stated before, the Holocene was a period with the lowest productivity. However, the transition between the Middle to Late Holocene had low percentages of F. profunda (EPP, Supplementary Material) in the GL-1109 core (Figure 6H). This was not related to precipitation or sea-level oscillation because these parameters were relatively stable during this period. The low F. profunda relative abundance was caused by the higher relative abundances of E. huxleyi, representing approximately 70% of the assemblage during this period (Figure 5D). Most likely, related to changes in sea surface nutrients, also in the same interval, we observed high percentages of the eutrophic indicator Helicosphaera spp. (Boeckel et al., 2006) and lower rates of oligotrophic Umbellosphaera spp. (Boeckel et al., 2006; Saavedra-Pelitero et al., 2011) and subtropical groups (Supplementary Figure S2). Austral summer-like conditions observed in the Middle-Late Holocene could have promoted increased shelf-break upwelling, which would have been enhanced due to higher BC meander-driven upwelling, causing higher rates of primary productivity in more offshore positions affecting the GL-1109 region, but not the shallower GL-824 core, which would have been influenced by colder coastal water and did not show such peaks in E. huxleyi relative abundance.