Holocene coastal evolution and environmental changes in the lower Río Guadiaro valley, with particular focus on the Bronze to Iron Age harbour ‘Montilla’ of Los Castillejos de Alcorrín (Málaga, Andalusia, Spain)

Phoenicians were the first to systematically develop the area surrounding the Strait of Gibraltar at the end of the 9th century B.C. Following pioneering studies in the Río Guadiaro estuary (Málaga/Cádiz) in the 1980s, a German‐Spanish cooperation project focussed on the role of indigenous people in the Phoenician colonisation trading networks at Los Castillejos de Alcorrín (Manilva, Málaga), one of the most important Early Iron Age settlements in southwestern Iberia. In the recent past, combined with systematic archaeological surveys, geoarchaeological research embedded in the interdisciplinary project ‘Archeostraits’ aimed at (i) deciphering palaeoenvironmental and coastal changes in the surroundings of Los Castillejos de Alcorrín throughout the mid‐ to late Holocene; (ii) constraining palaeoenvironmental conditions during early Phoenician colonisation; and (iii) better understanding human–environment interactions during the Final Bronze and Early Iron Age (i.e., end of 9th and 8th centuries B.C.). Coring transects along the Río Guadiaro allowed for differentiating successive palaeoenvironments and for establishing a chrono‐stratigraphy for the Holocene sedimentary infill of the valley. Based on these results, the deposition of shallow marine sands, overlying deltaic deposits of alternating sand and mud, and the subsequent development of lagoonal conditions in the lower Guadiaro valley took place before the Phoenicians established the first settlements along the coast.


| INTRODUCTION
For a number of reasons, the narrow Strait of Gibraltar is an area of outstanding geological, prehistorical and historical significance and a place of far-reaching cultural contacts. Phoenicians were the first to systematically develop this area, which is of considerable geostrategic significance. Starting mainly from Tyros in today's Lebanon, they extended their colonial expansion across the Mediterranean beyond the Strait of Gibraltar at the end of the 9th century B.C. They laid the foundations of a pan-Mediterranean oikoumene (i.e., living space and settlement area of humans in the Mediterranean area) that included the coastlines of the southern Iberian Peninsula and northwestern Africa (Aubet, 2006(Aubet, , 2009(Aubet, , 2016Marzoli, 2012Marzoli, , 2015Marzoli, , 2018Marzoli, , 2020Pappa, 2013). The Strait of Gibraltar at the western end of the old world was thus transferred into a crucial point of different networks connecting three continents.
From an archaeological point of view, and particularly in the context of Phoenician research, the region has attracted remarkable attention since the 1980s: the studies of the Madrid Department of the German Archaeological Institute including coastal research in the Río Guadiaro estuary (Hoffmann, 1988a,b;Schubart, , 1988 (Marzoli, 2012;Marzoli et al., 2009Marzoli et al., , 2010Marzoli et al., , 2020aSuárez Padilla & Marzoli, 2013).
Excavations unearthed an indigenous settlement whose construction at the end of the 9th century B.C. was related to the Phoenician expansion. Its size, monumental fortification and peculiarity of the interior construction, structure and design are unique in the region and are among the earliest evidence of iron metallurgy in southwest Europe. The reasons for the abandonment of the site at the beginning of the 7th century B.C. are still unclear. Evidence of destruction is lacking. The findings rather point to an orderly abandonment that coincided with a restructuring of the indigenous settlement of the coastal regions of southern Spain and the expansion of the Phoenician port cities (Marzoli et al., 2020a,c).
Within the framework of the interdisciplinary research project 'Archeostraits' systematic archaeological surveys in the territory of Los Castillejos de Alcorrín produced new evidence of the territory's use from the Palaeolithic to the late Middle Ages (Marzoli, et al., 2020c); three settlements were discovered between Los Castillejos de Alcorrín and the Guadiaro river, with surface finds (i.e., handmade pottery of the local Late Bronze Age tradition and Western Phoenician wheelware) pointing to a close connection with Los Castillejos de Alcorrín. Apparently, the central site and the satellite settlements were founded and abandoned at the same time, likely documenting a centrally controlled settlement policy reacting to the Phoenician colonisation efforts on the part of powerful indigenous inhabitants (Marzoli et al., 2020a). The occupation of the territory was obviously connected to the Phoenician exploration of the strait, which from the end of the 9th century B.C. onwards led to the emergence of new markets where the Phoenician maritime and indigenous continental economic areas met. The Guadiaro and favourable harbour locations near its mouth gained geopolitical importance (Marzoli, 2020;Marzoli et al., 2014).
The harbour settlement of Montilla, discovered in the course of the coastal research in 1987 (Hoffmann, 1988a,b), is only known fragmentarily. But based on the succession of pottery finds in three shallow excavations, it was possible to prove that it was an indigenous foundation that had close links with a Phoenician settlement probably located nearby (Schubart, 1988). Moreover, its connection with Los Castillejos de Alcorrín is evident. While the geoarchaeological research during the mid-1980s suggested a coastal embayment and direct marine access to Montilla (Hoffmann, 1988a,b;Schubart, 1988), these studies lacked detail, particularly in terms of the chronological resolution and the spatio-temporal evolution of palaeoenvironments in the surroundings of the site. In addition, as confirmed by a number of publications since then (e.g., Brisset et al., 2018;May et al., 2021;Vacchi et al., 2016), the decelerating rates of the relative sea-level (RSL) rise since 7000 B.P. 2 | PHYSICAL SETTING

| Geological, geomorphological and climatic setting
The study area is located in the active convergence zone between the African and Eurasian plates and belongs to the southwestern part of the Betic Cordillera, which stretches from the southeastern Iberian Peninsula to Morocco (Rif Cordilleras; Gibbons & Moreno, 2002;González-Castillo et al., 2015). The Betic Cordillera has evolved in the context of the Alpine orogeny from the Late Cretaceous onwards (Gibbons & Moreno, 2002;Grützner et al., 2012;Vergés & Fernàndez, 2012).
The study area belongs to the Flysch zone, which comprises a series of isolated tectonic units between the internal and external zones of the Betic Cordillera as well as an extensive area west of Estepona (Martín-Chivelet et al., 2002;Reicherter & Peters, 2005).
As part of this Gibraltar Flysch Zone (Reicherter & Peters, 2005), the lower reaches of the Río Guadiaro are surrounded by Paleogene to Neogene sand-, clay-, marl-and limestones (Aljibe and Algeciras Formations), which are partly covered by diverse post-orogenic Neogene to Quaternary deposits (e.g., alluvial and colluvial sediments, dunes), particularly in river valleys, local topographic depressions and coastal lowlands.
Displacements of the MIS 5e and 5c marine terraces suggest negligible rates of tectonic uplift or subsidence in the study area during the Late Quaternary (Zazo et al., 1999(Zazo et al., , 2008. In contrast, coastal uplift is highest at the Strait of Gibraltar (between Algeciras and Tarifa), some 30-40 km to the SW (Zazo et al., 1999). As subsidence is inferred for the Mediterranean coastal areas east of Estepona (Zazo et al., 1999), the study area is situated in a transitional zone with elevations of~6 and <2 m above the present mean sea level (asl) for the 5e and 5c terraces, respectively.
In the study area, the typical subtropical-Mediterranean Csa (Köppen & Geiger, 1968) climate with dry, hot summers and rather humid, mild winters (annual mean temp.~18°C, annual mean precipitation 700 mm, mainly winter rain; Gutierrez de Ravae Aguera et al., 1986) is related to large fluctuations in water discharge of the majority of streams. Smaller arroyos are indicated by torrential runoff, while the larger perennial rivers (such as the Río Guadiaro) mostly originate in the higher parts of the Betic Cordillera and show considerable seasonal differences in discharge volumes and sediment transport. The average tidal range amounts to~2 m in the Gulf of Cádiz and decreases to <40 cm along the adjacent Mediterranean coast, that is, in the study area. Superficial Atlantic waters enter the Mediterranean through the Strait of Gibraltar as fast-moving currents with seasonal fluctuations, resulting in anticyclonic eddies and a F I G U R E 1 Distribution of Phoenician settlements of the 8th and 7th centuries B.C. in the Western Mediterranean region and along the Atlantic coasts (green circles, compiled by D. Marzoli and E. Puch). The locations of La Silla del Papa and Los Castillejos de Alcorrín (red squares), both important Iron Age settlements of the southern Iberian Peninsula (e.g., Marzoli, 2012;Marzoli et al., 2010Marzoli et al., , 2020aMoret et al., 2008Moret et al., , 2017 and in the focus of the Archeostraits project, as well as the Río Guadiaro catchment (red polygon) are depicted as well.
seasonal rise in the summer mean sea level, and influencing coastal processes and oceanographic conditions in the Alborán Sea (Zazo et al., 1994).

| Holocene sea-level evolution and palaeoenvironments in southern Iberia
Following the last glacial maximum (LGM), rapid eustatically induced RSL rise with~8 mm/year in the Western Mediterranean is inferred until 7500 B.P., when RSL reached~7 m below the present mean sea level (bsl) (Lambeck et al., 2002;Vacchi et al., 2016). At that time, the maximum marine transgression is inferred at numerous places, for example the Gulf of Valencia at~2 km inland of the present coastline (Brisset et al., 2018), and Mediterranean deltas such as the Ebro delta (Cearreta et al., 2016) developed due to rates of fluvial sediment input exceeding those of the decelerated RSL rise (0.6 mm/year). This period is also referred to as the highstand system track (HST; cf. Brisset et al., 2018) and was related to the progradation of prodeltaic deposits of the Spanish continental shelf (e.g., Río Guadalhorce; Fernández-Salas et al., 2003) or of beach-ridge systems such as documented from the gulfs of Almería and Valencia (Brisset et al., 2018;Goy et al., 2003). Further deceleration of RSL rise is inferred since 4000 B.P. (Vacchi et al., 2016); particularly during this period, glacial isostatic adjustment or active (neo)tectonics, paired with temporarily high sediment supply from the hinterland (Anthony et al., 2014), is considered the main driver for coastal changes, resulting in the continued progradation of coastlines.
Palaeoenvironmental changes in southern Spain are documented by changing vegetation patterns or fluctuating soil erosion, as typical for the entire Mediterranean. Landscape degradation is generally related to the successive aridification towards the semiarid conditions during the mid-Holocene (Jalut et al., 2009;Petit-Maire, 1990) and the human impact on vegetation and soil cover by deforestation and agriculture (Jalut et al., 2009). Most of these changes occurred contemporary to the presence of humans, and ambiguities remain about whether the key controls for the changing palaeoenvironments were the climate, human impact, or a combination of both (Bellin et al., 2013;Roberts et al., 2004;Roberts, 1990;Van Andel & Zangger, 1990;Wainwright, 1994;Wolf & Faust, 2015).
In SE Spain, the opening of the landscape may have started after c. 4400 B.P. (~2400 B.C.) during the late Chalcolithic. Successively increasing dryness, likely combined with intensified burning, pas-toralism, and forest destruction (Carrión et al., 2010) resulted in a change towards sclerophyllous vegetation. Anthropogenic influence (and disturbance) such as agriculture, mining, deforestation, and pastoralism is assumed to have peaked during and after the Roman occupation (approx. 1st century B.C.-3rd century A.D.). In addition, human-induced landscape degradation and desertification in SE Spain (Río Aguas) between 6000 and 500 B.P., that is, between the Late Neolithic and Medieval times, is inferred by Castro et al. (1998). Comparable findings are also documented from Cádiz Bay (Lagoon of El Gallo, SW Spain) (Arteaga et al., 2004(Arteaga et al., , 2008López Sáez et al., 2002). Recent investigations on Lake Medina, situated some 80 km to the NE of the study area, brought evidence of an arid and warm climate during the early Holocene, which was followed by more humid conditions with high lake levels during the Holocene Climatic Optimum (7800-5500 B.P.) (Schröder et al., 2018). Since then, a progressive aridification (cf. Schirrmacher et al., 2020) and the establishment of typical Mediterranean-type vegetation took place, whereas increasing anthropogenic impact was detected only during the last~2000 years. Further studies on varved lake sediments in Zoñar Lake, Andalusia, suggest more humid conditions during the Roman Classical period (Iberian-Roman humid period,~2600-1600 B.P.) and drier conditions from 2000 B.P. onwards (Martín-Puertas et al., 2009). This agrees with other studies reporting more humid conditions in the Mediterranean area at that time (Luterbacher et al., 2006;Roberts et al., 2004;Zanchetta et al., 2007). In the Lake Medina record, however, increased xerophilous and anthropogenic nitrophilous taxa indicate drier, warmer conditions and elevated human influence at that time (Schröder et al., 2018), which is in agreement with other marine and lake records (Corella et al., 2011;Moreno et al., 2012). T2 and T3 are of major interest for this study. In T2 (Figure 12), that is, in the lower slope section, the archaeological strata directly cover the sandy Neogene bedrock and include basal cultural layers with settlement structures such as in-situ stone pavements, pits and fire places, covered by sediments and a palaeosol with abundant ceramic remains. Particularly based on this trench, Schubart ( , 1988 concluded that Montilla was more and more influenced by the Phoenicians during the 8th century B.C., which is expressed in successively increasing proportions of wheel-thrown early Phoenician pottery in the archaeological strata. Similarities in pottery findings also point to a close relationship between Los Castillejos de Alcorrín to Montilla. In T3, that is, in the easternmost part of the floodplain, Neogene sandstone builds the base of the sedimentary succession as well, followed by a unit of grey loam with abundant ceramic remains at 0.5-1.0 m asl. Then follows a 40-50-cm-thick, light brown sand layer, with several stones and an amphora handle at its upper boundary, which may have been intentionally put in place (Schubart, 1988). Above, a grey-brown loamy unit with abundant ceramic remains and stones was found. In this trench, West Phoenician wheel-thrown pottery dominates throughout the entire profile.  Hoffmann (1988a) found that the boundary to the Neogene basement seems to slope towards the centre of the floodplain in the area of T3: in his core 39, located a few metres to the SW of T3 This layer was interpreted to be of marginal marine origin (Hoffmann, 1988a,b), related to a brackish environment; remains of wheel-thrown Phoenician pottery are reported to have been found in this unit as well. Hoffmann (1988a,b) thus assumed that (i) the coastline was located in direct adjacency to Montilla during Phoenician times, and moorage of

| Field work
Field work was carried out between 2015 and 2017, including geomorphological and geophysical surveys in the form of electrical F I G U R E 3 ERT and coring transects A (a, b) and B (c, d). In general, the highest resistivities are related to coarse-grained channel deposits of the palaeo-Guadiaro (Facies E), as particularly found along transect A but also in the SW part of transect B (d). The lagoonal deposits (Facies C), found, for example, in M2 and M3, at the base of G5, and below the channel deposits in G2 and G6, are indicated by the lowest resistivities. Floodplain deposits (Facies F), the artefact-bearing layers in M1 (Facies O) and the shallow marine to deltaic sediments (Facies B) show low to medium resistivities. As indicated by the low resistivities along the NE section of transect B (c) above the pre-Holocene basement, no river channel deposits were found close to Montilla. ERT, electrical resistivity tomography.
F I G U R E 4 Stratigraphy of cores M2 and G2, which contain the most important stratigraphical units described in this study. Granulometric (grain size distribution and mean grain size) and selected geochemical results (C or TOC, Ca, Fe, Si and Sr contents) as well as radiocarbon ages are shown as well. The blue wavy line represents the position of the present mean sea level.

| Sedimentological and geochemical analyses
Sedimentological and geochemical analyses (e.g., Figure 4) were carried out in the laboratories of the Institute of Geography, University of Cologne. For granulometric and geochemical analyses, the oven-dried (40°C) and hand-pestled sample material was sieved to separate fractions ≤2 mm (fine fraction) and >2 mm (coarse fraction). The fine fraction was taken for further analysis. For granulometric analyses, samples were pretreated with H 2 O 2 (30%) and 0.5 N Na 4 P 2 O 7 (46 g/l) to remove organic carbon and for aggregate dispersion. The grain-size distribution of each sample was measured threefold using a Beckman Coulter LS 13320 Laser Particle Analyzer with 116-grain-size classes from 0.04 to 2000 μm. Grainsize parameters were calculated using GRADISTAT software (Blott & Pye, 2001) and followed the nomenclature of Folk and Ward (1957).
For further analyses, the fine-grained fraction was ground using a planetary ball mill (Retsch MM 400). Total carbon (C) and nitrogen (N) contents were analysed by combusting the ground sample material (~25 mg) at 950°C using folded tin containers and a vario EL cube element analyzer (Elementar). For the determination of total organic carbon (TOC) contents, 20 mg of the sample material was filled in silver containers and measured after digesting with HCl (10%). The total inorganic content (TIC) was calculated by subtraction of TOC from C. The results were corrected by a daily factor based on the measurement of a known standard (Acetanilide).  (Hammer et al., 2001). Parameters were selected based on a prior inspection of the data set. Elements with data discontinuities as well as low concentrations and large errors were excluded, and Spearman's rank correlation coefficient was used to avoid autocorrelations. Variables TOC, Sr, Fe, Ca, Si, mean grain size, and sorting were used for PCA (n = 400).  (2000), Karanovic (2012) and Pint et al., (2012Pint et al., ( , 2015 for ostracods ( Figure 7).

| Dating techniques
The chronological framework is based on altogether 36 14 C-AMS ages, which were derived from samples of cores B, G and M. 14 C-AMS analysis was performed at the 14 CHRONO Centre for Climate, the Environment and Chronology, Queen's University Belfast (UK).
Each sample was calibrated using the INTCAL20 or MARINE20 calibration datasets and Calib 8.2 software  Reimer et al., 2020); 14 C-dated marine carbonates were corrected for a local marine reservoir effect of ΔR = −160 ± 25, the weighted mean of 2 regional datapoints from the Marine Reservoir Correction Database Siani et al., 2000) (Table 1).
Two optically stimulated luminescence (OSL) samples were taken from closed core M6 (parallel core to M1; samples M6 OSL-1, M6 OSL-2) to verify/falsify a Holocene origin of the sterile sand sheet in M6 below the occupation layer/ground floor. Sample preparation and OSL measurements were performed under red-light conditions at the Cologne Luminescence Lab (CLL; Institute of Geography, University of Cologne). Analyses included dose rate estimation by means of high-resolution gamma spectrometry, measurement of in-situ water contents, and calculation of cosmic dose rates following Prescott and Hutton (1994). Palaeodose determination was based on sand-sized (100-200 µm) quartz grains separated by a combination of dry sieving, HCl (10%) and H 2 O 2 (10%) treatment, density separation (2.62 g/cm 3 < quartz < 2.68 g/cm 3 ) and HF etching (40% for 40 min).
Small aliquots (1 mm) per sample were measured on a Risø TL/OSL reader equipped with blue LEDs and a U340 filter following the SAR protocol (Murray & Wintle, 2003). Protocol performance was evaluated with preheat plateau tests (preheat plateau between F I G U R E 5 Stratigraphy, sampling strategy (sample depth and sample numbers; n = 41), distribution of sediment samples based on the PCA results (PCA based on TOC, Sr, Fe, Ca, Si, mean grain size, sorting), and, ultimately, classification into different facies, exemplified by sediment core G6. PCA results for all samples in this study (n = 400) are shown in Figure 6. In general, samples from similar facies cluster based on their PCA values (i.e., based on the sedimentological and geochemical fingerprints). Facies A represents littoral to sublittoral; Facies B represents deltaic/fluvio-marine; Facies C represents lagoonal; Facies D represents coastal lake/wetlands (freshwater-dominated); Facies E represents fluvial (channel deposits) and Facies F represents alluvial (overbank fines/floodplain). PCA, principal component analysis.
Dose rates and luminescence ages were calculated with the DRAC software (Durcan et al., 2015).  F I G U R E 7 Results of the micropalaeontological studies (Foraminifera, Ostracoda) and grain size data for cores M2 and M3. 14 C-AMS dating results (cf.  As illustrated by sediment cores G2, 5 and 6 (cf. Figures 4, 5 and 10), areas of low resistivities in the ERT transect are related to finegrained, silt-dominated sediments. Characteristics of these sediments with low resistivities, however, seem to vary considerably. At the base of G6, silty sand to sandy silt with numerous macrofaunal remains comprises the lower part of the core (16-12 m b.s.; Figure 5).
These sediments are summarised in Facies A, which is interpreted to have a shallow marine, sublittoral to littoral origin (cf. Supporting Information: Table S1). In G6, Facies A is replaced by alternating layers of clay, silt and fine sand (12-10 m b.s.), which points to a fluctuating input of sand-sized clastic materials, separated by periods of still-water conditions. These deposits are summarised in Facies B (Supporting Information: Table S1), which is interpreted to reflect deltaic deposition into a shallow marine environment.
Subsequently, Facies B is covered by a unit of well-sorted,  Table S1).
In contrast to these rather fine-grained core sections with low resistivities, coarse-grained deposits of well-stratified sand and gravel were found at depths where higher resistivities of 30-90 Ωm  Note: All ages were calibrated using CALIB 8.2 software and the data sets of Reimer et al. (2020).
| 141 on top of the low-inclined floodplain of the lower Río Guadiaro (Supporting Information: Table S1). In G5, the transition from facies C/D to Facies F matches the transition of lowest (<7 Ωm) to moderate (7-25 Ωm) ERT values at~7 m b.s. (Figure 3).
Finally, the westernmost core B1 slightly differs from the aforementioned cores and is generally indicated by quiescent depositional conditions in a marginal delta setting (Figures 9 and 11). The moderately sorted sand at the base of B1 possibly represents littoral deposits, which are followed by lagoonal mud with mollusks, some even  4.3 | Stratigraphy, facies interpretation and chronological results in the landward part of the study area (cores G8 and G9) G9 and G8 represent the northernmost cores in this study (Figures 9   and 10). Since the sediments show overall comparable characteristics to those in the seaward cores of transect A, we only briefly document the facies succession of cores G8 and G9 (cf. Supporting Information: Table S1). At G9, the sedimentary succession starts with deltaic

| Stratigraphy, facies interpretation and chronological results along transect B
ERT and coring transect B (Figures 2 and 3c,d) stretch between the foot slope directly below the former excavation area of the Montilla archaeological site and the Guadiaro River. Comparable with the depositional sequence of G6, core M2 (Figures 4 and 7) starts with F I G U R E 9 Oblique view of the facies distribution along coring transects A and B, as well as between transect A and the most landward cores G8 and G9; 14 C-AMS ages are depicted (all ages given in cal B.C./A.D.). As documented by G8, G2 and G6, palaeo-channels of the Río Guadiaro, dated between 1800 and~85 cal B.C., were located closer to Montilla than the present channel. Direct river access to Montilla, however, could not be confirmed; no river channel deposits were found in the directly adjacent cores (background: satellite image from Google Earth/Maxar Technologies, 2011). In the northeastern part of ERT transect B, cores M1 and M7-M9 were drilled at the foot slope directly below the former excavation area of the Montilla archaeological site (Figure 12). At the base of M1, the partly cemented and intensely weathered yellow sand is associated with resistivities of >30 Ωm and interpreted as the weathered sedimentary rock of Neogene age (cf. Hoffmann, 1988b) (Facies N), which was found at the base of M3 as well; its upper boundary rises towards the NE (Figure 3c). Facies N is topped by a thin layer of gravel in a red, clayey matrix that can be interpreted as a palaeo-surface, analogous to previous investigations in the surroundings of Montilla (Hoffmann, 1988b) (Facies P). This palaeo-surface is covered by a sequence of sterile brown medium sand, which in turn is overlain by a thin, brown, clay-and silt-containing stratum at 2.20 m b.s. (Facies P). Above, brown to grey, poorly sorted and rather heterogeneous deposits were found, which are dominated by silt and fine sand, contain abundant ceramic fragments, and show even higher resistivities of >50 Ωm (uppermost 1-2 m, Figure 3c). Most of the pottery was classified as ceramics from indigenous production.
These layers are interpreted as artefact-containing colluvial sediments, related to the nearby settlement of Montilla (Facies O; cf. Supporting Information: Table S1).
While a similar sedimentary sequence exists at M9, a slightly different sequence was found at M7/8, located a few decametres from T3 described in Hoffmann (1988b) and Schubart (1988) ( Figure 12). F I G U R E 12 Eastward section of coring transect B, that is, cores in the surroundings of Montilla, with facies distribution and 14 C-AMS age estimates. The map in the lower left (satellite image from Google Earth/Maxar Technologies, 2011) shows the location of cores carried out in this study (M1-3, 7-9) and in the study of Hoffmann (1988a,b;numbers 50-54 and 36-40, 55-56). T1-T3: archaeological excavations/trenches as described in Schubart (1988); profile T3 reproduced from Schubart (1988); the elevation of trench T3 was adjusted to the cores of this study due to apparent differences in measurements. is covered by floodplain strata (Facies F).
At G7, comparable to G11, units of sandy gravel (Facies E) and interdigitated mud (Facies C) alternate between 8 and 5 m b.s. at G7, that is, in the area of the southwestern bank of the modern Río Guadiaro (Figures 2 and 11). This sequence is followed by a rather thick unit of sandy gravel ( | 147 delta-type deposits (facies A and B) are older than 6000 B.C., except for the basal layers of core G7 (Figures 10 and 11). Delta-type deposits were found at the base of the most landward core G9, giving evidence of a more landward position of the Río Guadiaro mouth before~6000 B.C., that is, related to a landward shift of the coastline during the rapid eustatic sea-level rise until~6000 B.C. (~8000 B.P.).
This agrees with existing sea-level data from the area (Lario et al., 2002;Fernández-Salas et al., 2003;Zazo et al., 2008) and suggests that the shallow marine to littoral deposits below the delta-type facies date back to the transition from early to mid-Holocene (7th-6th millennium B.C.), that is, the end of the Epi-Palaeolithic period (before 6500 B.C.) (Figure 13a). However, no human occupation is documented in the directly adjacent area at that time, although evidence of Palaeolithic occupation is present from Gorham's Cave at Gibraltar (e.g., Pettitt & Bailey, 2000).
The sedimentological, geochemical and microfaunal results of M2 and M3 suggest a transition from shallow marine (below approx. 8.00 m b.s.) to open lagoonal and finally closed lagoonal conditions (above~5.50 m b.s.). 14 C-AMS ages from Facies C in G9, G6 and G2 as well as M2 and M3 (cf. Table 1, Figure 10) suggest that this transition occurred already during the mid-Holocene, that is, around or shortly after 6000 B.C., during the Neolithic period (~5500 B.C., Figure 13b). This may be explained by the evolution of a barrier in the seaward sections of the former embayment (e.g., at G7; Figures 10 and 11) at a time when local RSL rise decelerated (Vacchi et al., 2016) and littoral processes dominated coastal evolution. In any case, shallow marine deposits with fluctuating depositional conditions at G7 suggest that the sandy barrier and/or probably also delta-type conditions shifted seawards, resulting in open lagoonal environments in large parts of the former embayment. Sandy barrier to sublittoral deposits were also documented at G11 sometime before 4300 cal B.C.
In the lagoonal deposits (Facies C), autochthonous species (i.e., lagoonal species) coexist with allochthonous species of shallow marine origin; however, most of these allochthonous species are of pre-Holocene or even pre-Quaternary origin, which indicates an alluvial (Río Guadiaro) or even colluvial (adjacent slopes, Pliocene sedimentary rocks) sediment input. Variations in the content of these allochthonous species may be used as a proxy for either increased fluvial input and proximity of the Río Guadiaro or slope activity and erosion (i.e, colluvial sediment input): periods with river proximity and/or higher slope activity and erosion are separated by periods when lagoonal conditions with predominantly autochthonous, finegrained sedimentation prevailed.
From around 5000 B.C., rates of local RSL rise are assumed to have further decreased in southern Spain (cf. Vacchi et al., 2016;Zazo et al., 2008). The lagoonal conditions seem to have prevailed in most of the study area during the Neolithic, but a fluvial input in the seaward cores (e.g., G11) is suggested by high sand contents in mud layers and/or sand-dominated layers dating to ca. 4300 B.C.
Unfortunately, no chronological information is available in the landward cores G8 and G9 from these periods. However, the chronological information from Facies C suggests that lagoonal However, no evidence of channel deposits was found in core B1: here, standing water bodies may have prevailed during the Iron Age times, likely related to freshwater conditions. Potential Early Iron Age to Roman river channels may be found south of core B1, an area for which work/coring permits were unfortunately not granted.
Subsequently, the Roman period is characterised by increased numbers of settlements/occupation sites in and adjacent to the floodplain. Likely, the Río Guadiaro served as a major route to access the hinterland, which was controlled by towers at the narrow valley section close to G9 (Figure 13). Finally, the siltation of the Guadiaro alluvial plain and the shift towards terrestrial conditions in the southeast of Montilla seems to have taken place only within the past 1000 or so years. G3 (Figure 11) indicates high sedimentation rates in the most recent centuries, which could be connected, for example, to an eastward shift of the main channel of the Guadiaro estuary, as can also be assumed due to the two tower systems of modern age (late 15th and 16th centuries)

| Stratigraphical correlation with previous results
Based on the study presented here, and particularly based on the findings of transect B and the results from cores M1, M3 and M7, the sedimentary succession in the surroundings of Montilla and its palaeoenvironmental interpretation is in general agreement with the findings of Hoffmann (1988a,b) and Schubart (1988). Core M1 (e.g., Figure 9) with a~2 m thick layer with abundant artefacts (Facies O) reflects the findings along the NW sections of coring profiles I and II in Hoffmann (1988a) (cf. Figure 12). The sterile light brown sand at the base was dated to a pre-Holocene age using OSL (minimum age of 190 ka); it may tentatively be interpreted as terrace deposits of a Pleistocene Facies P). This is in perfect agreement with the Phoenician pottery remains in the layers of Facies O above, dated to the 8th century B.C.
As the upper boundary of the pre-Holocene surface at M7 is approx. 2 m lower than in the other surrounding cores M1 and M9 as well as in the excavation T3 described in Schubart (1988) (Figure 12), the palaeo-topography, related to the palaeo-surface associated with the Phoenician occupation layer/ground floor found in M1, seems to indicate a topographical low at M7, which is stretching a few metres towards the former settlement of Montilla. This situation may explain the existence of the stone layer (pavement?) in M7, which may tentatively be regarded as a countermeasure against swampy conditions, and it also explains the different sedimentary succession at M7, starting with (peat-like) Facies C directly above the Neogene Facies N.
In this area, Phoenician ceramic fragments are reported to have been found in a grey mud-dominated unit (i.e., a unit similar to Facies C in M7) reaching down to 1.50 m bsl (5.70 m b.s.), directly covering the Neogene basement in core 39 of Hoffmann (1988a). This layer was interpreted to be of marginal marine origin by Hoffmann (1988a,b), implying that the coastline was located in direct adjacency of Montilla during the Phoenician presence (Schubart, 1988). In contrast to these studies, however, ceramic fragments were only found until~0 bsl (~2.80 m b.s.) in the alluvial Facies F of M7: in the underlying Facies C (0-1.20 m bsl), which comprised abundant plant/ root remains and a peat-like organic-rich layer at the base and which may be considered similar to the basal deposits described in core 39 of Hoffmann (1988a,b), not a single ceramic fragment was documented. According to the 14 C-AMS dating results of root remains from this layer, these sediments accumulated sometime  Hoffmann (1988a,b), which may have affected coring location 39 but not M7 of this study. However, these questions cannot be solved conclusively based on the data at hand.
As discussed before, lagoonal environments ceased during the 3rd millennium B.C. in the area around Montilla. In agreement with the evolution of other coastal lowlands in southern Spain (e.g., Brisset et al., 2018;May et al., 2021), delta progradation, floodplain evolution and the seaward migration of the coastline to approximately its present location had already taken place before the Phoenicians arrived in MAY ET AL.
southern Iberia, resulting in a river-dominated landscape in the lower Guadiaro valley at that time. Lagoonal conditions at Montilla had thus likely already disappeared during Phoenician times, and a direct access by boat (e.g., in the form of a lagoonal coastline) to the settlement, postulated in previous research (Hoffmann, 1988a,b;Schubart, 1988), could not be confirmed. At that time, the deposition of overbank fines related to the evolution of the alluvial plain already took place in that area, as discussed in the previous section. Since no Holocene fluvial deposits were found in the direct vicinity of Montilla (transect B, cores M1-3, 7-9), a direct connection of Montilla to the open sea via the Río Guadiaro, that is, in the form of a river harbour directly at the settlement site, must also be considered unlikely. However, our results from transect A suggest that at around or after~1800 cal. In contrast to the previous investigations, our results suggest that the Phoenicians encountered a river-dominated landscape when establishing first trading contacts in Andalusia in the 8th century B.C.
Access to Montilla by boat was most likely via an old branch of the Río Guadiaro, which was, although not providing direct access to the site, located closer to Montilla than the present channel. Subsequent channel shifts/dynamics likely resulted in the abandonment of the site before 700 B.C., and the (assumed) appearance of Barbesula on the opposite (SW) side of the valley during the 7th century B.C. may point to a river channel shift towards the SW valley sections.
Finally, it may tentatively be assumed that these palaeoenvironmental changes had considerable influence on local occupation patterns. One of the reasons, if not the most important one, for the rather rapid (and non-destructive) abandonment of Los Castillejos de Alcorrín and its related settlements including Montilla may be seen in the dynamics of the estuary channels. More explicitly, the contemporaneous/simultaneous abandonment of Montilla and Alcorrín after the 8th century B.C. may be explained with a shift of the river channel towards the SW Río Guadiaro valley section, which may have resulted in a decreasing importance of the trading pathways along this trajectory.