The Tinto River (100 km long) is a small fluvial current that runs in a NE-SW direction through the southwest of the Iberian Peninsula. It shows the typical dynamics of Mediterranean rivers, with highly variable flows depending on the season (<0.1-100 m³ s^-1; Cánovas 2008). This river joins the Odiel River in a wide estuary at its mouth, which is characterized by the presence of two coastal sandy spits (Fig. 1B: Punta Umbría, Punta Arenillas) that protect large internal areas occupied by marshes and old barrier islands that contain cheniers and washover fans (González-Regalado et al. 2018). In this estuary, the tidal regime is mesotidal (2.1 m), with a low diurnal inequality (Borrego et al. 1993).

The interaction between tidal and fluvial inputs divides this estuary into three sedimentary sectors (Pendón 1999): a) the upper estuary, characterized by dominant tidal processes but with clearly fluvial sedimentation (mainly gravels and coarse sands); b) the middle estuary, with meandering channels separated by silty-clayey salt marsh bodies that eventually merge into one main channel, where the river waters (pH 2.7-5; conductivity 0.4-5 mS) and marine waters (pH 7-8.3; conductivity >110 mS) are mixed (Carro et al. 2006), with associated flocculation of fine sediment particles (López-González et al. 2006); and c) the marine estuary, with two main channels dominated by tidal dynamics (Fig. 1B: Padre Santo, Punta Umbría), with bioclastic gravels and sands near the mouth (Ruiz et al. 2000).

Between 1966 and 1975, two industrial concentrations were located on the estuarine border (Fig. 1B), with the world’s largest processing plant for Australian ilmenite black sand (TIOXIDE) and other alkaline, fertilizer and metallurgic plants. From then until 1990, the estuarine pollution from the mining activities was increased by acidic waste from these industrial sources and, to a lesser extent, by urban effluents. The final consequences of this increasing environmental pollution are the presence of very high concentrations of As (up to 3000 mg kg^-1), Cu (up to 4415 mg kg^-1), Pb (up to 10400 mg kg^-1) and Zn (up to 5280 mg kg^-1) in its surface sediments (Nelson and Lamothe 1993, Usero et al. 1999, Ruiz 2000). Since 1985, this zone has come under a corrective plan for the control of industrial waste disposal.

Core B (7.5 m length) was extracted in the transition between the alluvial deposits and the marshes located in the middle estuary of the Tinto River, near the town of Palos de la Frontera (Fig. 1C; 37°14´N; 6°54´W). It was obtained by usual rotary drilling techniques with a continuous recovery of sediment and a barrel diameter of 11.6 mm. Twenty-two samples were collected for a multidisciplinary analysis, according to the visual study of the sedimentary facies and their boundaries. These selected samples cover all sedimentary facies observed in core B and the main bioclastic levels.

The visual lithological description was complemented with the grain-size analysis of 18 samples, selected according to the vertical distribution of sedimentary facies. To determine grain-size distribution the coarser fractions were wet-sieved with pressurized water and floated using sieves of 63, 125, 250 500, 1000 and 2000 µm diameter. Fractions smaller than 63 µm were analysed with a Mastersizer-2000particle counter at the CIDERTA centre of the University of Huelva, Spain. Na-hexametaphosphate was used as a dispersing agent.

The trace metal content of 16 samples was determined and certified by MS Analytical, Langley (Canada). Four trace metals (As, Cu, Pb, Zn) selected as tracers of environmental pollution were obtained by ICP Atomic Emission Spectrometry, with less than 5% variation between replicate samples and a quality control based on more than 30 sample references (e.g. OREAS 904). In these samples, detection limits were 0.02 mg kg^-1 for As, 0.2 mg kg^-1 for Cu, 0.5 mg kg^-1 for Pb and 2 mg kg^-1 for Zn. The geoaccumulation index (I_geo; Muller 1969) was applied in order to define the environmental pollution in sediments (e.g. Li et al. 2016, Al Mukaimi et al. 2018, Yan et al. 2020). This index is calculated as follows:

where C_n is the measured concentration in a sample for the metal n, B_n is the background value for the metal n in the study area and factor 1.5 is used because of possible variations of the background data due to lithological variations. This index defines six pollution categories: I_geo<1 (unpolluted); 1<I_geo<2 (very low polluted); 2<I_geo<3 (low polluted); 3<I_geo<4 (moderately polluted); 4<I_geo<5 (heavily polluted); and I_geo>5 (very heavily polluted).

I_geo is vulnerable to distortion by grain-size effects. Consequently, a double regional background of the Tinto-Odiel estuary (Ruiz et al. 1998) was used for (i) silty-clayey sediments (As, 7.3 mg kg^-1; Cu, 13 mg kg^-1; Pb, 15.4 mg kg^-1; Zn, 39 mg kg^-1); and (ii) sandy sediments (As, 6.8 mg kg^-1; Cu, 13.1 mg kg^-1; Pb, 4.3 mg kg^-1; Zn, 15.9 mg kg^-1).

Eighteen samples were selected for microfaunistic analysis. In each sample, 20 grams of sediment were separated and levigated through a 125 µm mesh sieve. This sieve was selected for three reasons: a) it retains practically all of the malacofauna for which sieves with a larger mesh diameter are usually used (≥250 µm mesh diameter in most cases; Alexandrowicz 2017, Cristini and De Francesco 2017); b) the benthic foraminiferal assemblages picked from this fraction provide useful information on prevailing environmental conditions that is useful for an overview of paleoenvironmental changes in coastal settings (Lo Giudice Capelli and Austin 2019), two of the main objectives of this paper; and c) this sieve is suitable for screening the ostracod fauna and preventing the presence of juveniles of these microcrustaceans (stages 1-4), which are very difficult to identify (Danielopol et al. 2002).

The residue was dried in an oven at a constant temperature of 40°C for a period of no less than one day. The total populations of molluscs (bivalves and gastropods) were determined according to Gómez (2017) and the World Register of Marine Species (WoRMS). The microfaunal content (foraminifera and ostracods) was also extracted and determined. Foraminifera were classified according to Loeblich and Tappan (1987), Milker and Schmiedl (2012) and WoRMS. Ostracoda were classified according to Ruiz et al. (1997, 2000). This micropaleontological record was compared with (i) the assemblages of both groups present in the Neogene formations found in the vicinity of core B (González-Regalado and Ruiz 1996, Ruiz and González-Regalado 1996), in order to identify reworked specimens; and (ii) a database of the distribution of the two groups in the estuaries of southwestern Spain (Ruiz et al. 2000, González-Regalado et al. 2001), which was used to recognize paleoenvironmental changes or anthropogenic impacts in this core. These studies analysed the foraminiferal and ostracod assemblages of the main sedimentary facies of these estuaries (high salt marsh, low salt marsh, ebb-tide channel, main channel, tidal plain and beach).

Two dates were produced at the National Centre of Accelerators (CNA, Seville, Spain), with a subsequent calibration using CALIB version 7.1 and the reservoir effect correction (-108±31 ¹⁴C yr) calculated by Martins and Soares (2013) for this area. The final results correspond to calibrated ages using 2σ intervals. The mean calibrated ages (MCA) were used to produce a chronological framework of core B. They were completed with the comparison between the vertical geochemical variations and the main periods of mining extraction and industrial discharges (see Introduction), in order to obtain additional age references.

The lowermost 1.2 m of core B (Fig. 2; facies F1: 7.5-6.3 m depth) are composed of similar percentages of sands and silts (35-45%), with minor contributions of clays (<18%) and scarce gravels. This interval does not contain macrofauna. This absence contrasts with the abundance of mollusc fragments observed in the overlying 1.4 m (F2: 6.3-4.9 m depth), consisting of bioclastic sands and gravels with silty intercalations containing frequent vegetable remains. In this facies, a total of 162 specimens were determined belonging to four species of bivalves and four species of gastropods. Two macrofaunistic assemblages were recognized: a) the gastropods Bittium reticulatum (da Costa, 1778) and Peringia ulvae (Pennant, 1777) and the bivalves Ruditapes decussatus Linnaeus, 1758 and Cerastoderma glaucum (Brugiére, 1789), which were very abundant especially in bioclastic gravels (Fig. 2: 5.7-5.4 m depth); and b) the gastrod Bittium reticulatum and the bivalves Ostrea ostentina Payraudeau, 1826, C. glaucum and Cerastoderma edule (Linnaeus, 1758), which were well represented in the uppermost bioclastic sands of this facies (5.4-4.9 m depth).

The third sedimentary facies of core B (F3: 4.9-1.4 m depth) is represented by sandy silts (silt, 54%-63%; sand, 9%-22%) with frequent fragments of charcoal, stems and roots. Macrofauna is also absent in this facies, which is replaced by fine silty sands with frequent vegetable remains in the upper 1.4 metres of the core (F4).

The heavy metal pollution profiles of core B reveal the four sedimentary facies, with the differentiation of three pollution levels (Fig. 3). F1 is characterized by low to very low concentrations of the four measured elements, including some of the lowest values of core B (Fig. 3: As<8 mg kg^-1; Cu<22 mg kg^-1; Pb<15 mg kg^-1; Zn<35 mg kg^-1). Similar contents were also measured in the bioclastic levels of F2, which contrast with the first peak (Fig. 3: FPL) observed in the intercalated silts. In these fine sediments, As (27 mg kg^-1), Cu (39 mg kg^-1), Pb (74 mg kg^-1) and Zn (93 mg kg^-1) are considerably higher.

F3 is characterized by very low to low concentrations of these heavy metals (e.g. As<20 mg kg^-1; Cu<25 mg kg^-1). This facies includes a second relative maximum at 4.2 m depth (Fig. 3: SPL), with As (30 mg kg^-1), Cu (52 mg kg^-1) and Zn (107 mg kg^-1) as the most representative trace elements of this new peak. However, the highest contents of all these trace elements (As, 44 mg kg^-1; Cu, 284 mg kg^-1; Mn and Pb, 98 mg kg^-1; Zn, 539 mg kg^-1) were measured at 2 to 1.4 m depth, with the highest contents near the transition between F3 and F4 (Fig. 3: TPL). This sandy facies has a geochemical signal very similar to that of F1.

The geoaccumulation index clearly identifies the three pollution peaks, and especially TPL (Fig. 3). The first pollution peak (FPL) is defined by very low pollution levels for all four investigated metals (1≤I_geo≤1.68), while similar geoaccumulation indices (1.02≤I_geo≤1.83) were detected during the second pollution peak (SPL). The highest values of I_geo were obtained during the third pollution peak (TPL), with sediments moderately or heavily polluted by Zn (I_geo up to 4.5), Pb (I_geo up to 3.9) and Cu (I_geo up to 3.9). This pollution was weakest for As (1.74≤I_geo≤2.1) during this last peak. The remaining samples were unpolluted, except F4 for Pb (I_geo=1.2).

A total of 1839 shells of benthic foraminifera belonging to 11 Holocene species (1395 specimens) and 27 Neogene reworked species (444 specimens) were extracted from the 18 samples studied, with more than 99% of the individuals concentrated in F2 (Fig. 4). In this facies, both Holocene (25-35 ind. g) and Neogene (up to 16 ind. g) foraminifera are abundant.

The Holocene record of these microorganisms starts at the uppermost sandy levels of F1, with rare specimens of Ammonia tepida (Cushman, 1926). This species is very abundant in the upper bioclastic gravels and sands of F2 (994 ind./60 g), together with Ammonia beccarii (Linnaeus, 1758), Haynesina depressula (Walker and Jacob, 1798) and Haynesina germanica (Ehrenberg, 1840). These species (Fig. 5) were found together with the scarce ostracods extracted in this core [Fig. 4: Leptocythere fabaeformis (Mueller, 1894) and Loxoconcha rhomboidea (Fisher, 1855)]. Few specimens of A. tepida and A. beccarii characterize the silty levels of this facies. This facies also contains a considerable number of Neogene species (Fig. 5), the most representative ones being Nonion commune (d’Orbigny, 1846), Bulimina elongata (d’Orbigny, 1826), Fursenkoina acuta (d’Orbigny, 1846), Globobulimina affinis (d’Orbigny, 1839), Melonis pompilioides (Fichtel and Moll, 1798) and Marginulinopsis costata (Bastch, 1791). Echinoderm spines are also very abundant in F2.

Three foraminiferal zones are differentiated within F3 (Fig. 4): a) the basal zone (F3a), characterized by the disappearance of all the previous groups coinciding with the second pollution peak; b) the intermediate zone (F3b), with only specimens of Trochammina inflata (Montagu, 1808) (<5 ind./20 g), an agglutinated-walled foraminifera; and c) the upper zone (F3c), represented by the exclusive presence of Entzia macrescens (Brady, 1870), another agglutinated-walled foraminifera (<5 ind./20 g). Charcoal and small plant remains (stems, roots) are abundant in the two upper zones. Foraminifera, ostracods, other faunal groups and plants were not found from the transition between F3 and F4 (e.g. TPL) and the top of core B.

Two dates were obtained from two levels, at depths of 6 m (sample CNA-4274; organic matter; 5892-5711 cal yr BP; MCA~5.8 cal kyr BP) and 5.7 m (sample CNA-4272; shell; C. edule; 5483-5368 cal yr BP; MCA~5.4 cal kyr BP).

The silty-sandy sediments of F1 are characterized by the very low concentrations of all the elements analysed, as well as the absence of fauna and plant remains. This facies is interpreted as alluvial deposits, which probably originated from erosive action of the Tinto River drainage network. Of these all deposits, the Cabañas creek is still preserved in the studied sector (Fig. 1C). Its geochemical signal is similar to that measured in nearby Pleistocene to Holocene fluvial sediments of the Tinto River (Cáceres et al. 2013).

The lower and middle bioclastic gravels and sands of F2 (Fig. 4: 6.3-5.4 m depth) include an assemblage of Holocene molluscs (Bittium, Peringia and Ruditapes), benthic foraminifera (mainly Ammonia and Haynesina genera), ostracods (Leptocythere and Loxoconcha genera) and fragments of echinoderms typical of shallow subtidal ebb-tide channels or sandy tidal flats with variable salinity and partially vegetated bottoms (Ruiz et al. 2000, González-Regalado et al. 2001).

In the uppermost sandy layers of this facies (5.4-4.9 m depth), the increasing presence of cockles (C. glaucum, C. edule) and the decreasing density and diversity of foraminifera (Fig. 4) point to a partial restriction of tidal inputs and a decrease in water depth, with a transition to an intertidal, very shallow channel with brackish waters. The silty intercalations probably correspond to a transition to the border of this channel (Gómez 2017, Ordiales et al. 2017). The presence of abundant Neogene foraminifera involves the simultaneous erosion of the adjacent Mio-Pliocene formations (González-Regalado and Ruiz 1996, Ruiz and González-Regalado 1996). Some of these Neogene foraminifera currently live in the adjacent Gulf of Cádiz, mainly at depths of 30 to 100 m (e.g. B. elongata; Mendes et al. 2012), but they have not been found in the middle zones of the adjacent estuaries (González-Regalado et al. 2001).

The few species of benthic foraminifera found in F3 are characteristic of marshes (see Murray 2006 for a review). T. inflata, the most representative species of F3b, is more common in low salt marshes, while E. macrescens (F3c) is usually more abundant in high salt marshes (González-Regalado et al. 2001, Jones 2014). No foraminifera were found in F3a and F4. This facies corresponds to the anthropic filling of the area during the last third of the 20th century.

The two dates obtained in F2 indicate the deposit of 0.3 m of bioclastic sediments in about 400 years, with a mean sedimentation rate of 0.75 mm yr^-1. If this average rate is extrapolated to the entirety of F2, the base of this sedimentary facies was deposited around 6.4 cal kyr BP (Fig. 6). The age of its lower limit coincides with the flooding of this estuary during the beginning of the Holocene transgression maximum (~6.5 cal kyr BP; Zazo et al. 1994), clearly contrasted by the determined fauna (see the previous section). This would indicate that a) the deposit of basal alluvial sediments (F1) would have occurred before ~6.4 cal kyr BP; b) the transit from subtidal to intertidal paleoenviroments took place around 5.3 cal kyr BP ago within F2; and c) the maximum of the Holocene transgression caused the flooding of this area and the first episod of natural pollution (FPL), clearly reflected within the finest sediments of this facies (Fig. 3). This peak is one of the oldest episodes of natural pollution in southwestern Europe and has been detected in other cores of the Tinto River estuary (Arroyo et al. 2021). This small peak may be due to the coincidence of the Holocene flooding and the weathering/oxidation of the mineral deposits concentrated in the Iberian Pyrite Belt, a process that began in the Oligocene (Essalhi et al. 2011).

F3 includes a moderate pollution peak at 4.2 m depth (SPL) and the highest concentrations of trace metals at 1.4 m depth (TPL). According to the previous datings and the sedimentation rates obtained by Lario et al. (2002) in this estuary, the first of them is attributed to the first mining activities in this area (4.7-4.5 cal kyr BP; Rothenberg and Blanco Freijero 1980, Olías and Nieto 2015).

The uppermost peak is due to the conjunction of intensive mining and industrial discharges during the 19th and 20th centuries (Ruiz 2000, Ruiz et al. 2009). Both episodes of pollution coincide with the disappearance of all the faunal groups studied (Fig. 4). This negative impact of pollution on the assemblages of molluscs, foraminifera and ostracods has been contrasted in different littoral environments (Armenteros et al. 2016, Reddy et al. 2016) and may even cause the disappearance of these groups (Ruiz et al. 2000, González-Regalado et al. 2001).

The time interval between these two episodes witnessed a period of silting, with the transition from a low salt marsh to a high salt marsh in recent centuries (Fig. 6). During the intermediate stage with little mining and industrial activity, the foraminiferal record of this period is very similar to that observed in undisturbed salt marshes (Jones 2014, Soualili 2018, Haller et al. 2019). No evidence of Roman mining pollution (~2.1-1.5 cal kyr BP) has been found in this core, as in other cores of this estuary (e.g. Leblanc et al. 2000). This absence may be due to a) an erosion of the Roman deposits in this fluvial context; or b) the location and paleoenvironmental evolution of these cores, which may include sediments deposited in emerged areas during Roman times. Finally, the upper sandy layer (F4) comes from the anthropic filling of this sector in the last 30 years.

The correlation with other cores collected in the middle estuary of the Tinto River estuary (Ruiz et al. 2020, Arroyo et al. 2021) can be used to obtain a synthetic paleoenvironmental evolution of this sector since the Upper Pleistocene, with four main phases: a) the Upper Pleistocene, when the fluvial network eroded the Neogene marine formations and created several coves on a Miocene substrate; b) the Lower Holocene (ca. 10-6.5 kyr BP), with the deposits of alluvial, sandy sediments: c) the Holocene transgression (ca. 6.5 kyr-5.4 kyr BP), with the flooding of this area and the transport of marine bioclastic sands towards the innermost zones of the estuary; and d) the Late Holocene (ca. 5.4 kyr BP), with the development of wide salt marshes and the fit of the main channels.