This study is based on the analysis of core SP5 (140-cm long), collected in the western portion of the SB (UTM: 0621164/7460621; 23K, WGS84), near Madeira island (Figure 1). Core SP5 was collected by divers using the percussion probing method. After collection, this core was sealed and transferred to the Micropaleontology Laboratory of the Universidade do Estado do Rio de Janeiro (UERJ), Faculty of Geology (LabMicro/UERJ), where it was frozen. After that, the whole core was defrosted, opened, and described. Since the core had no sedimentological discontinuities, it was continuously sampled at 2- to 3-cm intervals. Sixty sediment samples were obtained and devoted to granulometric, geochemical, mineralogical, and benthic foraminiferal analyses. This core was dated with ²¹⁰Pb and ¹³⁷Cs.

For particle size analysis, about 10 g of total sediment per sample was used (although small in amount, it should be representative of the sample given its homogeneity). The sediment was washed through a 63-μm mesh sieve to separate the <63 μm and >63 μm fractions. Both fractions were collected and oven-dried at low temperature (60°C). After drying, the sediment fractions were weighed and stored. Samples from sedimentary fraction >63 μm were separated by a set of sieves with different mesh sizes (i.e., 63, 125, 250, 500, and 1,000 μm). The sediment contained in each sieve was weighed to determine the percentage of each particle size. Textural classification of these sediments was based on the classifications of Folk and Ward (1957).

Sedimentary fraction <63 μm (silt-clay fraction) was used in the mineralogical analysis. It was separated with distilled water from the coarser sediment fractions. For the mineralogical analysis by X-ray diffraction (XRD) technique, about 3 g of sediment was dried in an oven at low temperature and disaggregated in an agate mortar. XRD measurements were performed using the Philips PW1130/90 and X’Pert PW3040/60 devices at the Aveiro University (Portugal), which used Cu Kα radiation. Scans were performed between 2 and 60° 2θ (in unoriented powder assemblies). The identification and semi-quantification of minerals followed the methodology described by Martins et al. (2007).

For the ²¹⁰Pb and ¹³⁷Cs analyses, about 15 g of dry sediment was used. Twenty samples were disaggregated in an agate mortar and analyzed in an EG&G ORTEC (Hyperpure Ge, model GMX25190P) gamma spectrometer at LaQIMar (Laboratory of Marine Inorganic Chemistry, University of São Paulo, Brazil). Ages were based on a constant sedimentation model in accordance with Figueira et al. (2007) and Ferreira et al. (2014).

Samples of 10 ml of sediment were used to assess foraminiferal assemblages. The sediment samples were washed with distilled water in a 63-μm sieve. Fractions <63 μm and >63 μm were stored in beakers and oven-dried at low temperature (<60°C). Foraminifera from fraction >63 were picked from a volume of 10 ml and placed in foraminiferal slides with the aid of a mink hair brush and a Zeiss microscope, model Stemi SV11, with a maximum magnification of 200×.

For species identification, the Ellis and Messina catalog (Ellis and Messina, 1940-2015) was consulted, as well as specific references, such as Loeblich and Tappan (1987), for the identification of genera, and Boltovskoy et al. (1980), Debenay et al. (2001), Martins and Gomes (2004), and Alves Martins et al. (2019a) for identification at specific level. The online catalog available at WoRMS (World Register of Marine Species; Hayward et al., 2020) was also used to update the name of species.

Foraminiferal density (FD) was calculated as the number of tests found in 10 ml of sediment (n°/10 ml). Species richness (SR; number of species present in a sample) was calculated for all samples with foraminifera. The exp(H’_bc) and evenness (J’) indexes were estimated only for samples with a number ≥100 specimens/10 ml. According to Fatela and Taborda (2002), this is the smallest number of individuals that will allow us to characterize foraminiferal assemblage with reliability. Thus, only these parameters were determined in samples between 0 and 32 cm of core SP5. SR and evenness (J’) index were determined with the Primer 06 software.

The biotic index exp(H’_bc), based on foraminiferal diversity, was evaluated according to Bouchet et al. (2012). It was used to evaluate paleo-ecological quality status (PaleoEcoQS) [for details, see Bouchet et al. (2012) and Francescangeli et al. (2016)]. Ecological Quality Ratio (EQR) was further calculated to make a more accurate assessment of PaleoEcoQS. EQR is the ratio between the value of a biological metric [diversity in our case, i.e., exp(H’_bc)] and the expected value under reference conditions. Benthic foraminiferal assemblages from the SB, found in core SP8 (Castelo et al., 2021b; Figure 1), were used to infer pre-impacted reference conditions. Core SP8 records natural and anthropic forcing in the last ∼9.5 ka BP. The application of Foram Stress Index (FSI) and exp(H’_bc) allowed us to identify sedimentary layers that characterize a system with maximum health at ≈5 ka BP during the mid-Holocene relative sea-level highstand. The exp(H’_bc) of the assemblage associated to the maximum health of this system (recorded in core SP8 by Castelo et al., 2021b) was taken as a reference to estimate EQR. Five equal-size class boundaries were categorized as follows: 1–0.8, high; 0.8–0.6, good; 0.6–0.4, moderate; 0.4–0.2, poor; and 0.2–0, bad EcoQs. The package “entropy” (Hausser and Strimmer, 2014) in the R software (R Core Team, 2016) was used to calculate exp(H’_bc).

In addition, Ammonia-Elphidium Index (AEI) was used to infer changes in oxygenation conditions of the bottom environment. Both genera are common in coastal and transitional waters; however, Ammonia dominates in low-oxic bottom waters and/or sediments (Duleba et al., 2018, 2019). This index was estimated with the equation AEI = [NA/(NA + NE)] × 100 (Sen Gupta et al., 1996), where NA is the number of Ammonia specimens, and NE is the number of Elphidium/Cribroelphidium specimens. The AEI values range from 0 (well-oxygenated) to 100% (dysoxic-anoxic).

For statistical analysis of the granulometry data, the Gradistat software (Blott and Pye, 2001) was used to obtain parameters such as sediment mean grain size (SMGS), mode, sorting, skewness, and kurtosis of the analyzed samples.

The relationship among mineralogical, geochemical, isotopic, and abundance of benthic foraminifera data was analyzed by Spearman correlations and principal component analysis (PCA) in Statistica v.13 (TIBCO Software Inc, 2018). Variables were categorized into layers and converted into years according to age estimations based on ²¹⁰Pb and ¹³⁷Cs (Supplementary Table 1).

For the PCA, variables were reduced according to the criteria: (i) element with concentration values below detection levels or sparse data (< 10 valid cases), (ii) redundancy–r > 0.9 (the variable with highest correlation with the matrix–Pearson correlation was retained), and (iii) invariance check (coefficient of variation threshold < 0.01). The data were standardized using the “ranging for variables with arbitrary zero” procedure. The goal of normalization was to change the values of numerical columns in the data set to a standard scale without distorting differences in the ranges of values (Milligan and Cooper, 1985). The analyzed components were those that exhibited eigenvalues higher than 1. Foraminiferal abundances were computed in the PCA as supplementary variables and, therefore, did not influence the other variables. The PCA plot is a representation of the correlations between active variables (environmental), supplementary variables (species) and the cases (years). Components with eigenvalues greater than 1 were retained.

In addition, a canonical correspondence analysis (CCA), which is a multivariate ordination technique, was also performed using version 7 of the PC-ORD software (McCune and Mefford, 2016), aiming to extract major gradients among biotic and abiotic variables and samples (represented by years).

Core SP5 has a light brownish gray (at the middle part) to dark gray or black color. No abrupt transitions or apparent sedimentary structures were observed. The age model of this core, based on ²¹⁰Pb data presented in Supplementary Table 1, reveals that it records the last ≈150 years, with an average accumulation rate of ≈0.83 ± 0.13 cm year^–1. The maximum fallout of ¹³⁷Cs, corresponding to 1963, was recorded at the 45-cm level.

Core SP5 is a sandy-mud and muddy-sand sedimentary sequence (Supplementary Table 2), with sediment mean grain size (SMGS) of ≈ 40 ± 15 μm. The percentage of the fine fraction (<63 μm) varies between 30.4 and 81.1% (with an average equal to 59.6%). The fine fraction is predominant both at the base and at the core top, while the sand fraction increases in the middle section of the core (Figure 2), where SMGS reaches 74.4 μm (Supplementary Table 2). At the top of the core, the sediments are bimodal, while in the intermediate portion and at the base of the core, the sediments are polymodal. The sediments are predominantly composed of 3 modes, 76.5, 152.5, and 605 μm. Sorting values (σ) reveal that the sediments are generally poorly sorted in the upper and lower sections of the core and are very poorly sorted in the intermediate zone, and have coarser granulometry.

The main mineralogical constituents are (Supplementary Table 2): phyllosilicates (48.3–87.35%, mean: 72.19 ± 7.5%), followed by quartz (7.99–31.94%, mean: 16.82 ± 4.84%), K-feldspar (0–16.45%, mean: 3.65 ± 3.26%), and plagioclase (011.39%, mean: 1.5 ± 2.51%). The accessory constituents are calcite (07.9%, mean:0.69 ± 1.49%), pyrite (0–7.13%, mean: 1.94 ± 1.21%), zeolites (0–4.84%, mean:0.27 ± 0.88%), siderite (0–4.67%, mean:0.85 ± 1.34%), anatase (0–4.39%, mean: 1.09 ± 1.41%), anhydrite (0–4.25%, mean.78 ± 1.17%), magnetite/maghemite (0–2.82%, mean:0.19 ± 0.57%), ilmenite (0–1.35%, mean:0.09 ± 0.31%), and dolomite (0–0.85%, mean:0.03 ± 0.15%). Significant changes in mineralogical composition of the sediments along the core are not observed. However, the presence of pyrite and increase of this mineral in the upper part of the core, as well as rise in feldspar/quartz ratio values should be noted. The values of this ratio also increase slightly in the middle part of the core (Figure 2A).

Total organic carbon (0.33–1.8%, mean:0.91 ± 0.45%) and N (0.049–0.188%, mean:0.104 ± 0.046%) contents tend to continuously increase from 80 cm to the upper part of the core. The δ¹⁵N values (between 7.775 and 12.743‰, mean: 9.752 ± 1.298 ‰) show an inverse pattern (Figure 2B and Supplementary Table 2). Only punctual increases in TOC are recorded in the lower section; the most significant is at 103 cm (Figure 2B). C/S ratio values (0.2–4.35, mean: 1.3 8 ± 1.2) rise in the ≈1927–1979 (35–75 cm) interval, where sediment mean grain size (SMGS) is relatively high (Figure 2B). Carbonate content (<16.3%, mean: 10.2 ± 4.2%) increases upward, while IR values (83.7–100%, mean: 89.7 ± 4.2%) show an inverse pattern (Figure 2A). δ¹³C values (between −26,376 and −22,008‰ mean: −23,495 ± 1.442‰) significantly reduce in the interval where S values (0.18–3.64%, mean: 1.13 ± 0.77%) are at their highest (≈110–75 cm, ≈1882–1927; Figure 2B).

The values and ranges of concentrations of the analyzed elements are included in Supplementary Table 2. In some sediment layers, elemental concentrations are above the world shale (Turekian and Wedepohl, 1961; Table 1) and local baseline (estimated by Pinto et al., 2019; Table 1), such as for Cd, Zn, Mg, and P, which increase in relation to respective baseline values in the upper 70 cm (after 1930; Figure 3). The values of Ca and Cu are below the world shale (Turekian and Wedepohl, 1961) and local baseline values (Pinto et al., 2019), but both elements have been showing an increasing trend since ≈1915 (in the upper 75/80 cm; Figure 3). It is noteworthy that Ca concentrations are quite low in the inferior part of the core (Figure 3). Instead, some elements, such as Zr, Nb, Hf, W, Th, Al, and Ta, reach higher concentrations in the period ≈1858–1927 (130–80/75 cm), but since ≈1927 (above 80/75 cm), their concentrations have been relatively low (Figure 3).

Core SP5 has generally higher Pb concentrations than the world shale (Turekian and Wedepohl, 1961) and local baseline values (Pinto et al., 2019; Figure 3). Tin (Sn) presents a distribution pattern similar to Pb; both elements decrease their concentrations in the intermediate section where the sediments are coarser (Figures 2, 3). Between ≈1882 and 1927 (≈80/75–110 cm), peaks of U, Mn, Mo, Fe, and S are observed, as well as a sharp decline in P and δ¹³C values (Figures 2, 3). In addition to these elements, As concentrations (Figure 3) also reach values higher than those of the world shale (Turekian and Wedepohl, 1961) after ≈1965 (in the first ≈40 cm) and local baseline value (Pinto et al., 2019) between ≈1920–1943 (≈80–60 cm).

Some elemental ratios were selected to represent changes in the geochemical composition of the sediments such as: Ti/Ca, which may trace changes in the lithogenic particle supply (Nace et al., 2014; Gebregiorgis et al., 2020); Y/Ni, Cr/V, La/Sc, and Th/Co, which may indicate changes in sources of lithogenic materials in the study area (Hiscott, 1984; Bhatia and Crook, 1986; Cullers, 2002; Okunlola and Idowu, 2012), and Rb/K, which is used to assess changes in water salinity (Campbell and Williams, 1965). Ti/Ca and Th/Co reach higher values in the lower part (below 85 cm), whereas Y/Ni and Cr/V display higher values in the upper 85/90 cm (Figure 4A). Rb/K values decrease sharply in the mid part of the core, between 80 and 110 cm. On the contrary, La/Sc ratio shows higher values in this interval characterized by relatively coarser sediments (Figure 4A). Changes that occurred in the values of these ratios are discussed.

Considering the classification ranges presented in Table 2, EF values are categorized as: 1 < EF < 2 for Al, As, Fe, P, and Mg; 2 < EF < 5 for Ce, Hf, Mn, Mo, Nb, Pb, S, Sn, Th, U, W, and Zr; and 5 < EF < 20 for Zn and Cd (Supplementary Table 3). Most of the elements reach maximum Igeo values < 1 (Supplementary Table 3), except for Sn and Zr (up to 1), Cd (up to 2.7), and Zn (up to 3.1). The EF and Igeo values of Sn are relatively higher in both ends of the core and decrease in the middle interval where the sediments are coarser. Some elements exhibit maximum CF values > 1, between: 1 and 2 for Al, Th, Hf, Ce, U, Nb, Zr, P, Pb, and Sn, 2 and 3 for Cd, and 3 and 4 for Zn (Supplementary Table 3). Higher PERI values (up to 329) are found in the upper core part (Supplementary Table 3 and Figure 4B). The EF and Igeo values of Cd and Zn and PERI values significantly increase in the upper 80 cm and particularly mostly in the uppermost 50 cm.

The FD ranges from 0 to 446 specimens/10 ml (Supplementary Table 4). Samples are devoid of foraminiferal tests below 66 cm (Figure 5). The highest SR is identified in layers with highest FD. Thirty-eight species are identified along the core, 33 of which have a carbonate test and 5 with an agglutinated test (Supplementary Table 4). The most abundant species are Ammonia tepida, Buliminella elegantissima, Bolivina striatula, and Cribroelphidium excavatum (Figure 5 and Supplementary Figure 1A). Ammonia rolshauseni, Pararotalia sarmientoi, and Ammonia parkinsoniana are also recognized (FD < 19 specimens/10 ml). Other bolivinids and buliminids are identified (density < 10 specimens/10 ml); the abundance of this group increased after 1998. The Ammonia/Elphidium index (AEI) shows the predominance of the first group over the second one, with a sharp peak at 30–20 cm depth (Figure 5).

In the 0- to 32-cm interval, FD is > 148 specimens/10 ml, and SR varies between 9 and 22 (Figure 5). Exp(H’bc) and J’ indices ranged between 2.68 and 8.38 and 0.46–0.56, respectively (Supplementary Table 4). The most abundant species is A. tepida; its frequency is >50% from 1974 and up to 78% (Supplementary Figure 1B). This species is followed by B. elegantissima (up to 32%) and P. sarmientoi, B. striatula, and C. excavatum. Their abundances do not exceed 6% (Supplementary Figure 1B).

The EQR values vary between 0.21 (poor EcoQS) and 0.67 (good EcoQs), with highest values in upper layers of the core (Figure 6 and Supplementary Table 5).

The first two PCA components explain about 83% of the total variance (Figure 7A). The first component (PC1; ca. 66% of variance) is positively related to C/S, δ¹³C, coarser grain size (medium + coarse sand), δ¹⁵N, Mg, and Fe (Table 3). These variables have the highest correlation with the layers deposited around 1943, 1947, 1938, 1953, and 1958, in this order of importance (Table 4). Foraminiferal species do not show a positive correlation with these years. PC1 is also negatively related to fine fraction, TOC, and most of the heavy metals (Figure 5). Years with highest correlation with these variables are 1974, 1980, 1989, 1998, 2001, 2004, and 2015. Foraminiferal species with highest negative correlation with PC1 are A. tepida, P. sarmientoi, B. elegantissima, A. parkinsoniana, C. excavatum, B. striatula, and A. rolshauseni. The second component (PC2; ca. 17% of variance) is positively related to K-feldspar and Hf but exhibits a negative correlation with Al (Table 3). Bolivina striatula and C. excavatum are correlated with the period around 2004. Foraminiferal species do not have a significant correlation with the negative axis of PC2.

The CCA results for axes one and two (explaining 56.5 and 11.3%, respectively, of data variability) represent the main relationship of species with selected sedimentological variables and samples represented by decades (Figure 7B). The positive side of axis 1 is mostly related to PERI, La/Sc, fine fraction, TOC, C/N, and the species P. sarmientoi, A. tepida, and A. rolshauseni and the decades 1980s, 1990s, and 2000s, while its negative side contains the species A. parkinsoniana, B. elegantissima, C. excavatum, and B. striatula, the sedimentological variables Ti/Ca, δ¹³C, δ¹⁵N, and C/S, and the decades 1940s, 1950s, 1960s, 1970s, and 2010s. Axis two separates the species A. parkinsoniana, B. elegantissima, and P. sarmientoi (positive side) from C. excavatum, B. elegantissima, and A. rolshauseni (negative side), and associates the first group with most of the sedimentological variables and the second group with the 1940s, 1950s, 2000s, 2010s, and the gradients of TOC and C/N ratio.

The mineralogical and geochemical results reveal that the sediments of core SP5 are essentially siliciclastic. Chromium (<110 ppm) and Ni (<65 ppm) concentrations (Supplementary Table 2) are below the North American shale composite values (NASC; Garver et al., 1996) and indicate that the main source of the sediments are felsic rocks (Hossain et al., 2017). The La/Sc vs. Th/Co diagram (Figure 8A) based on Cullers (2002) and Okunlola and Idowu (2012), the Zr-Th-Sc ternary diagram (Figure 8B) based on Bhatia and Crook (1986) and Okunlola and Idowu (2012), and the Hiscott diagram (Hiscott, 1984) of Cr/V vs. Y/Ni (Figure 8C) also point out a metamorphic felsic origin as main source of the sediments. This is in accordance with the geological evolution and lithological composition of this region (Heilbron et al., 2020).

The variation of concentrations of some chemical elements is positively correlated with the fine fraction of the sediments (such as Al, Bi, In, Nb, Ta, Ti, V, and U; Supplementary Table 6), indicating variations in the fluvial contribution of eroded materials from the drainage basin to the study area, as also observed by Gonçalves et al. (2020). The increase in concentrations of these chemical elements may reflect calmer hydrodynamic conditions favorable for fine sediment accumulation. However, some chemical elements such as Cd, Cr, Cu, Ni, P, and Zn show non-significant correlations with fine sediments (Supplementary Table 6) and may be associated with pollutant sources related to anthropic activities developed in the region (Magalhães et al., 2001; Mounier et al., 2001; Molisani et al., 2004; Araújo et al., 2017a,b). Other elements, such as As, Be, Ca, Ce, Co, Fe, Mg, Mn, Na, Rb, Sr, U, and Y, are significantly and positively correlated with sand fraction (Supplementary Table 6). These results suggest that natural/anthropogenic processes may influence the compositional characteristics of sediments in the study area.

Thus, in order to be able to characterize possible changes in sediment characteristics and associate them with possible causes, four main intervals were considered along core SP5:

The geomorphology of the bottom of the SB is complex and has changed considerably over time, as shown in studies of Friederichs et al. (2013) and Reis et al. (2020). The compositional characteristics of the sediments of Ints. 1 and 2 may have resulted from Guandu River delta processes and/or changes in geomorphological characteristics of the study area, as suggested by its sedimentological characteristics. Araripe et al. (2011) suggested that the elements Hf and U are associated with the mineral zircon, which is a possible product of erosion of igneous rocks located in surroundings of the bay. Thus, the increased concentration of these elements in Int. 2 indicates that the supply of materials is mainly from continental provenance. In fact, the high values of Ti/Ca ratio (Figure 4) coupled with relatively high concentrations of Zr, Nb, Hf, W, Th, Al, Ta, and IR observed in Ints. 1 and 2 (Figure 3 and Supplementary Table 2) also indicate the accumulation of essentially lithogenic particles (Nace et al., 2014; Gebregiorgis et al., 2020) and presence of a quite low biogenic component, as supported by low carbonate contents, and absence of mineralized foraminiferal tests (Supplementary Table 4), ostracods, and mollusks. The poor preservation of biogenic carbonates should be caused by low pH conditions. It should be noted that foraminiferal tests start dissolving when the pH decreases to values below 7.6 (Prazeres et al., 2015). The pH decrease may have resulted from organic matter degradation processes and from acid percolation through the sedimentary column. However, carbonates were still identified in these intervals (Figure 2); since not only CaCO₃ but also other types of carbonates with differentiated resistance to dissolution by acids should be present.

These geochemical data indicate that the sediments are mostly supplied by the erosion of local and regional rocks. However, the reduction in La/Sc and Th/Co values and higher values of Y/Ni and Cr/V ratios (Hiscott, 1984) in Ints. 3 and 4 (Figure 4), reveal a greater occurrence of mafic and ultramafic rocks in the composition of the sediments, which may be related to changes in the characteristics of sedimentary dynamics in the region. Despite the expressive record of lithogenic supply in core SP5, the mineralogical results also suggest changes in proximity to the source area of the sediments or changes in reworking processes, mostly in Int. 4. The higher values of feldspars/quartz ratio in Int. 4 (Figure 2) may indicate higher sedimentary input from the drainage network adjacent to the study area and presence of less weathered sediments, since minerals from the feldspar group are less resistant to weathering than quartz (Lira and Neves, 2013). This change may be due to the natural evolution of the SB, but it may also have been influenced by anthropic actions.

Population increase (and related activities) in the river margins (Barcellos et al., 1997; Borges and Nittrouer, 2016a,b) may have facilitated the transport of less reworked sediments to the study area. Exploration of sand in the river margins have promoted disfigurement of the river channels and collapse of the margins, which generated holes and lateral creeks (SEMADS, 2001) and enhanced the greater contribution of sediments of continental origin to the study area (Barcellos et al., 1997). On the other hand, sediments from the lower part of the core (Int. 1 to Int. 3) have relatively low feldspars/quartz ratios, indicating that the sediments have undergone greater reworking than those from Int. 4.

Transitional waters are normally exposed to both continental drainage and marine action (Perillo, 1995). Changes in continental/oceanic influence were inferred from Rb/K ratio. This ratio has been used to interpret varying levels of salinity (Liu and Cao, 1984; Wang et al., 2016; Zou et al., 2021) as the following: (1) Rb/K ratio of ≤0.004 indicates freshwater conditions; (2) > 0.004 to ≤ 0.006 suggests influence of fresh to brackish water, and; (3) > 0.006 fully marine conditions (Campbell and Williams, 1965). The use of this ratio along core SP5 is based on the assumption that fine marine sediments contain higher Rb contents because of increased concentrations of this element in seawater (0.12 ppm) compared to freshwater environments (0.0013 ppm). Potassium concentrations in core SP5 are positively correlated with sandy sediments (with higher proportion of fine, medium, and coarse sand; Supplementary Table 6), as their highest concentrations were reached in coarser sediments (Figure 3), as well as Rb concentrations (Figure 3). K concentrations can be associated with several mineral types such as K-feldspar (KAlSi₃O₈), mica (Kal₂Si₃AlO10(OH)₂), and illite (K,H₃O)(Al, Mg, Fe)2(Si, Al)4O10[(OH)2,(H₂O)]. The data obtained by Pinto et al. (2019), along core SP2 and collected near the mouth of Guandu river (22°55′0.00′′S; 43°45′60.00′′W), reveal that the main clay mineral is illite (45–74%). Thus, K concentrations may represent a lithic component consisting of K-feldspar and phyllosilicates. Thus, the reduction of Rb/K ratio values in Int. 2 (Figure 4) likely suggests the influence of less saline water. This inference is also supported by the stable isotope results of organic matter, as explained below.

The plot of δ¹³C vs. C/N (Figure 9A) and δ¹³C vs. δ¹⁵N (Figure 9B) and the values and ranges suggested based on Deines (1980), Lamb et al. (2006), Martinelli et al. (2009), Barros et al. (2010), and Bueno et al. (2019) allow us to infer that the study area received a contribution of organic matter from several sources, including terrestrial C3 plants, freshwater algae, oceanic productivity, and sewage discharge (Figure 9). The most negative δ¹³C values (Meyers, 1997) may be related to greater supply of an organic material of fresh water algae origin. This corroborates the inference regarding lower Rb/K values, which indicate higher freshwater influence in the study area (Figure 4). However, the fact that these more negative values are simultaneously recorded with higher levels of S and slight increase of pyrite (Figures 2A,B) allows us to deduce that they may have also been influenced by diagenetic changes in the sediment, which led to the production of sulfides in anoxic sedimentary sub-environments. On the other hand, according to Stein (1991), δ¹³C values around -20‰ may be related to marine plankton. Thus, the less negative values of δ¹³C recorded between ≈1930 and 1955 associated with lower values of C/N ratio may represent higher contribution of organic matter from marine productivity and between ≈1960 and 2015 associated with relatively high values of C/N ratio, which should indicate greater contribution of sewage to the study area. This is also evident in increased TOC values since ≈1963 (45–0 cm; Figure 2).

In fact, highest TOC contents reaching ca. 1.8% are found in Int. 4, associated with fine grained sediments (Figure 2 and Supplementary Table 2). However, as TOC concentrations can reach up to 12.5% in coastal areas influenced by human activities, such as domestic or industrial sewage discharges (Kennish, 1997), this enrichment can be considered as moderate. TOC contents are, in general, <0.8% before ≈1930, which may indicate low organic matter preservation and/or high heterotrophic activity and/or remobilization after deposition (Souza et al., 2017). However, the presence of pyrite in Int. 1 and Int. 2 and along the entire core (Figure 2 and Supplementary Table 2) indicates low oxic conditions of the sediments, even in the sandiest layers. This mineral is formed in anoxic sediments, and its preservation also occurs in oxygen-depleted environments (Wang and Morse, 1996).

The prevailing low-oxygenated conditions are also corroborated by the low values of C/S ratio. According to Lyons and Berner (1992), C/S ratios below 2.8 ± 0.8 indicate sediments deposited under reducing conditions. Most part of this core has C/S values <2.8. However, a peak of C/S ratio occurs in the middle part of the core (Int. 3), which is associated with coarser grained sediments (Figure 2) indicating the presence of more “oxic” sediments when compared to another interval, namely, the Int. 1. However, the presence of oxic boundaries is already noticed in the previous interval (Int. 2) through the increase in redox-sensitive elements, such as U, Mn, Mo, S, and Fe (e.g., Urban et al., 1999; Brown et al., 2000; Zheng et al., 2000, 2002; Adelson et al., 2001; Thomson et al., 2001). The high enrichment of these metals is usually attributed to changes in redox conditions of the sediment. Thus, the increase in redox-sensitive metals, especially U and Mn, depletion of Ti/Ca values associated to coarser sediments, and grain size indicate strengthening of hydrodynamics in Int. 2, probably because of higher connection of the study area with oceanic waters. Increased marine influence may have been favorable to the development of carbonate-bearing organisms, especially foraminifera, which led to progressive enrichment of biogenic carbonates from Int. 3 onward.

In summary, the results presented reveal that:

On the basis of the EF values and classification criteria of Sutherland (2000), Int. 4 has sediments moderately to significantly enriched in Cd, Zn, and Sn. Similarly, the Igeo values and Müller’s (1986) classification criteria suggest moderately to heavily polluted conditions by these PTEs in this interval. Contamination with the Cd and Zn metals is widely documented in the study area (e.g., Lacerda et al., 2001; Mounier et al., 2001; Martins et al., 2019; Castelo et al., 2021a,b). Companhia Ingá Mercantil (or CIA Ingá, currently inactive) may have significantly contributed to this situation. This company, founded in the 1950s in Madeira Island (Figure 1), was responsible for the production of high-purity zinc (Magalhães et al., 2001; Mounier et al., 2001; Molisani et al., 2004; Araújo et al., 2017a,b). The liability generated by CIA Ingá was the target of several tailing leaks, such as what happened in February 1996 where 50 million L of contaminated water and mud were dumped into the SB (Melo, 1996). However, Cd and Zn concentrations tended to decline at the top of core SP5 after the inactivity of CIA Ingá in 1997 (Fundação Estadual de Engenharia do Meio Ambiente [FEEMA], 1997; Molisani et al., 2004).

Araújo et al. (2017a,b) found higher concentrations of Zn and Cd than this study in areas close to Cia Ingá. This suggested the presence of internal vertical flows that promote remobilization of fine particles, enhancing the dispersion of metals and resuspension and transport of sediments contaminated along the bay. It is possible that the presence of these contaminants in core SP5 has resulted from the dispersion of pollutants from the source areas. Sediment destabilization through ship traffic, heavy swell, and dredging operations in navigable channels to give access to the Sepetiba/Itaguaí port may have released contaminants to the water column. These contaminants were transported by currents to the study area and several other areas.

Tin concentrations are relatively higher at the top and the lower part of core SP5 (Figure 3) but decrease in the middle zone of the core (in coarser sediments; Figure 2). However, this change does not fully accompany changes in sediment grain size and may be related to lithogenic sources of the metals. In addition, the increase of Sn concentration in the upper part of core SP5 may also be a consequence of anthropogenic contributions. Tin contamination in the upper part of this core may be also associated with two main sources: use of biocides such as tributyltin (TBT) and ore deposits of cassiterite. Ban on the use of paints with antifouling properties based on TBT started in Europe and Asia during the 1970s and 1980s (Alzieu, 2000). However, restriction in the use of TBT only occurred in Brazil in 2007 (NORMAM-23/DPC, 2007). The overall decline of this biocide toward the top of the core may probably be attributed to this restriction. The main source of Sn in Brazil is cassiterite. There are important reserves of cassiterite in Rondônia, northern Mato Grosso, southern Amazonas, and southern Pará where exploitation of this element is important (White, 1974). Minor deposits or occurrences of cassiterite are located in Pará, Amapa, Paraiba, Rio Grande do Norte, Ceará, Bahia, Minas Gerais, Goiás, São Paulo, and Rio Grande do Sul (White, 1974). Ore materials transported from other regions of Brazil are refined and metallurgically treated at Companhia Estanífera do Brasil near a steel plant, Brazilian government’s National Steel Mill, in Volta Redonda (Rio de Janeiro State), about 60 km from Sepetiba, which has been in operation since the 1950s (Rogers, 1968; Warhurst, 1999). This activity should be also a source of this chemical element for SB. Studies have revealed that dispersion of Sn from this alloy industry through the atmosphere (wind transport of particulates) has been important in the region and is causing serious public health problems (Azevedo et al., 2019). In addition, there are lithologies containing stanniferous deposits or occurrences associated with, for instance, Funil Granit, located in the Resende Basin, Rio de Janeiro (Pereira et al., 2003), which is about 85 km from the SB. Therefore, the presence of Sn-enriched lithologies in the region should also be a source of this chemical element.

Phosphorus contents increased in Int. 3 and remained relatively high (generally above the local and world shale baseline values) in Int. 4 (Figure 3). Phosphorous has significant positive correlations with TOC, carbonates, and several metals, such as As, Ca, Cd, and Zn, but not with pyrite (Supplementary Table 6). This is due to the fact that the relative abundance of this mineral is generally lower in the upper section of the core, since it is the least depleted in oxygen. This suggests that the increase in P in Int. 3 and Int. 4 may have been influenced by municipal sewage, diffuse (fertilizers) runoff from agricultural soils, and its increase should be also related to sources of pollution. Phosphorus is a chemical element that acts as a nutrient for phytoplankton and facilitates the reproduction of algae and vascular plants, contributing to the eutrophication process (Schauser et al., 2004), that is, high supply of organic matter to the bottom. Castelo et al. (2021a) suggested that the supply of organic matter favored sub-oxic conditions in the region, and that the occurrence of pyrite was the result of the anoxic condition in the bay’s microenvironments contributing to retention of PTEs.

High concentrations of As in the vicinity of Madeira Island have been previously reported by Magalhães and Pfeiffer (1995) (Figure 1). The relatively high concentrations of As recorded in core SP5 in Ints. 3 and 4 (Figure 3) can be associated with high-purity Zn and Cd treatment process by metallurgical companies in the region (Barcellos et al., 1991). However, the highest enrichment of As recorded in core SP5 occurred in Int. 2 (Figure 3) before the aperture of CIA Ingá. Therefore, this enrichment is also probably associated with lithological sources and to the weathering of the rocks of the region (as also observed by Pinto et al., 2019).

Arsenic is rarely found as a native element because of its binding affinity for other elements and mineral species except hydrothermal ores (Reimann et al., 2009). It can form As³⁺ oxides, arsenolite, and claudetite (both As₂O₃), polymorphs with similar thermodynamic stability (Nordstrom and Archer, 2003). These minerals can form products of arsenic sulfides (AsFeS) occurring primarily in hydrothermal and magmatic ore deposits (O’Day, 2006). Relatively high concentrations of As have been found in beach sand from Espírito Santo and Rio de Janeiro states (Mirlean and Baisch, 2016). The presence of relatively high concentrations of As in these coastal regions is still very poorly understood but has been attributed to various geochemical processes such as early diagenetic processes (Mirlean and Baisch, 2016) and hydrothermal vents associated with past magmatic activity (Heilbron et al., 2020). In this region, As, on average, is about 30 times higher than the background value for sandy sediments and 10 times higher than the background value in near shore mud (5 mg/kg) (Mirlean and Baisch, 2016). High As concentrations were also found in shallow groundwater in the Paraíba do Sul delta (N of the study area) probably as a result of release of As bound by authigenic sulfides (Mirlean and Baisch, 2016). The rate of oxidation of sulfides depends on several factors, including amounts of arsenopyrite (e.g., AsFeS), particle size distribution, temperature, and time of exposure of the material to atmospheric water and oxygen (O’Day, 2006).

The data of core SP5 suggest that many natural and anthropogenic factors may have caused the enrichment of some chemical elements. However, the sediment enrichment in PTEs in the inner SB area since the 1950s, but especially since the 1970s, may have been associated primarily with anthropogenic forcing.

The absence of foraminifera between ≈1858 and 1925 suggests that the natural environmental conditions in the study area were not suitable for the presence of these organisms during Int. 1 and Int. 2. The sediments deposited in this period were also characterized by low Ca concentrations, which corroborates the absence of foraminifera and mollusk shells. Note that the presence of carbonates (more regularly in Int. 2) may indicate favorable conditions for the formation of these minerals by abiogenic processes and/or their transport and accumulation in the study area. According to Silva et al. (2000), precipitation of evaporitic minerals occurs in a predictable order, as salt water evaporates, with calcite being the first mineral to be deposited. In fact, during the ≈1882–1925 interval, this area of the bay should have been associated with processes of the deltaic complex of Guandu River. The site where core SP5 was collected was in an elevated region and, therefore, must have gone through phases of subaerial exposure, periods of greater freshwater influence, alternating with other phases in which the area was invaded by marine waters. These phases favored the precipitation of inorganic carbonates. However, such conditions were not favorable to the establishment of benthic foraminiferal communities.

With the establishment of oceanic circulation similar to the present one, the study area became progressively more suitable for the development of benthic foraminifera. Starting from ≈1938 (Int. 3 to Int. 4), foraminifera appeared in the sedimentary record (Supplementary Table 4 and Figure 5). The large variability of the physicochemical parameters and the instability of the substrate prevented the significant development of a stable community in the study area in Int. 3 (≈1935–1974). In this interval, the assemblages were affected by the extreme complexity of the hydrodynamics, with large asymmetries in the distribution of ebb-flood currents caused by the semi-diurnal variation of tides and friction with bed forms (Cunha et al., 2006). These conditions could have generated the polymodal and very poorly sorted sediments of Int. 2 and Int. 3.

The prevalence of calm hydrodynamic conditions in Int. 4 after ≈1974 (Figure 5), combined with abundant food supply, allowed for increase in benthic foraminifera abundance. However, only poorly diversified assemblages were recognized. Mainly species with opportunistic behavior, such as A. tepida, B. elegantissima, B. striatula, and C. excavatum, common in anthropized coastal environments, as has been noticed in previous studies (e.g., Duleba and Debenay, 2003; Vilela et al., 2004; Alves Martins et al., 2019b; Castelo et al., 2021a).

Ammonia tepida is a common species in coastal environments because of its tolerance to physicochemical variations such as in salinity, temperature, and nutrient availability (Zaninetti et al., 1977; Bouchet et al., 2007; Frontalini et al., 2009). The relatively high values of Ammonia/Elphidium ratio in Int. 4 indicates a shallow confined transitional environment. According to Duleba et al. (2018, 2019), the highest proportion of Ammonia spp., in relation to Elphidium spp., suggests the presence of stressful coastal environmental conditions due to bottom water and/or sediment pore water hypoxia. Thus, the highest values of the AEI in Int. 4 indicate a progressive trend toward the occurrence of eutrophication and consequent reduction in oxygen associated with the presence of sediments with moderate to considerable ecological risk, according to PERI values.

The enrichment of PTEs (namely, Cd, Zn, and Sn) seems to have been a conditioning factor for the establishment of diversified assemblages and the presence of species sensitive to these adverse environmental conditions. Along the core, the most abundant species (A. tepida, B. elegantissima, B. striatula, and C. excavatum) are known to be tolerant to high instability of transitional waters and presence of high TOC, low oxygen contents (Brönnimann et al., 1981; Murray, 1991, 2006; Moodley and Hess, 1992; Alve and Murray, 1999; Frontalini et al., 2009), and even metal-induced pollution (Vilela et al., 2004).

The PCA results and isotope data (Figures 7A, 9) show that the increased density of most frequent species in foraminiferal assemblages found in the upper part of core SP5 is associated with increase in organic matter of mixed sources, including those of marine and land origin materials, in addition to calm hydrodynamic conditions. Although these assemblages were found in sediments moderately to strongly polluted mainly by Cd and Zn, factor 2 of the PCA (Figure 7A) shows that the higher abundance of these organisms is reached in less polluted sediments by PTEs. However, it is not the abundance of foraminifera that shows the quality of the environment but the composition and diversity of their assemblages, which are related to the response of the species to environmental stress (Alves Martins et al., 2020). The high metal content and amount of organic matter, and the redox state of the sediments are conditioning factors of the type of foraminiferal assemblages found in the study area. However, the quality of the organic matter also seems to be an important factor. Some species, such as A. parkinsoniana, B. elegantissima, C. excavatum, and B. striatula, are mostly related to better quality of organic matter (as indicated by the higher δ¹³C and δ¹⁵N values) related to marine productivity.

PaleoEcoQS was poor in 1975, highlighting the impact of anthropogenic activities. They worsened after the metal spill event from CIA Ingá that occurred in 1996 (Melo, 1996; Magalhães et al., 2001; Mounier et al., 2001; Molisani et al., 2004). Since then, environmental quality progressively improved, and PaleoEcoQS became good in 2015. Hence, our results indicate a possible trend for recovery of the study area after more pronounced environmental degradation. A similar situation was also reported by Rodrigues et al. (2020) by comparing the distribution of Cd and Zn in surface sediments from periods 1990–2000 and 2000–2010, and by Molisani et al. (2004) and Araújo et al. (2017a) in sediment cores.

Benthic foraminifera are confirmed to be a resilient group of microorganisms and able to recover after severe stressful conditions (Bouchet et al., 2007; Oron et al., 2014) caused both by natural and anthropogenic processes. The biotic index exp(H’_bc) applied on the fossil record appeared to be accurate in reflecting the environmental conditions occurring in the bay, confirming previous findings (see review in O’Brien et al., 2021). This index further complements the assessment of the environmental quality obtained from the chemical composition of the sediment. Direct (i.e., PTEs) and indirect (i.e., benthic foraminifera) indicators gave a complementary evaluation of the ecological quality. The former helps in identifying and quantifying the source of pollution, and the latter further illustrates the adverse effects of PTEs on living organisms. The results of this study also shed light on the potential of fossil foraminifera to serve as proxies for identification of in situ reference conditions (Alve et al., 2009; Dolven et al., 2013; Polovodova Asteman et al., 2021). Identification of local reference is crucial to reliably monitor the health of marine and transitional waters.