1Deceleration of sea level rise during the mid-Holocene around 6000–4000 BC initiated delta formation all over the world (Stanley and Warne, 1994). Since that time, river deltas – which represent terrestrial outposts in a marine environment – have always been appreciated by man for fishing and settlement activities. However, continuous delta progradation and coastal changes have implied a steady struggle against siltation in order to guarantee open access to the sea.
2In the deltaic areas of the eastern Mediterranean, numerous archaeological sites testify to considerable environmental changes during the late Holocene and thus have been subject to intense geoarchaeological investigations. Most of the studies are based on the analyses of sediments from near-coastal geological archives. Besonen et al. (2003), for example, reconstructed strandline displacement of 5 km by the prograding Acheron River since 400 BC. At that time, ancient Ephyra and the nearby Nekromanteion, oracle of the dead, still lay on the marine front. Kraft et al. (1987) showed how delta progradation of the Sperchios River changed the narrow coastal pass at Thermopylae, where the Greek army under Leonidas fought their famous battle against the Persian army in 480 BC, into the wide alluvial plain of present times. The Aliakmonas, Loudias, Axios and Gallikos River deltas are responsible for the rapid siltation of the Thermaikos Gulf (Vouvalidis et al., 2005). Nearby Pella, the hometown of Alexander the Great, had already lost direct access to the gulf around 1 BC/AD (Vött and Brückner, 2006). Around ancient Troy, the wide marine embayment from the time of the so-called Trojan War around 1200 BC was silted up by the Dümrek (Simois) and Karamenderes (Scamander) Rivers (Kraft et al., 2003). Further south, the prograding Küçük Menderes (Kaystos) River had caused infilling of the Holy Harbour of Ephesus by Roman times (Brückner, 2005). Neighbouring seaports such as Priene, Myous, and Miletus were sealed off from open marine conditions by the prograding Büyük Menderes River around 300 cal BC (Müllenhoff, 2005; Brückner et al., 2006). These examples show that river deltas are good archives for reconstructing the complex interactions between sea level rise, sediment supply, and coastline displacement as well as for detecting interactions between man and the environment.
3During the past millennia, the prograding Acheloos River (Akarnania, NW Greece) has landlocked several islands of the former Echinades archipelago (Philippson, 1958; Villas, 1984; Bousquet et al., 1987; Fouache et al., 1998; Fouache, 1999; Vött et al., 2004, 2007a, 2007b; Vött and Brückner, 2006). Today, ancient Oiniadai with its famous shipsheds from the 5th to 3rd centuries BC is located some 9 km inland (fig. 1). This paper focuses on the former ports of the island. The main objectives were (i) to reconstruct palaeoenvironmental changes in the northern harbour, (ii) to clarify whether other ports existed, (iii) to find out how and when the harbours lost their function, and (iv) to document how the siltation of ancient harbours was controlled by the progradation of the Acheloos River delta.
Fig. 1 – The central Acheloos River delta, NW Greece, and the former island of Trikardo bearing the archaeological remains of ancient Oiniadai. Topographic overview and locations of the different former harbours and embayments investigated within this study.
Fig. 1 – Le delta de l’Acheloos (Grèce du NW) et l’ancienne île de Trikardo où sont situés les restes de l’ancienne cité d’Oiniadai. Esquisse topographique et localisation des anciens ports et baies successifs.

1: modern village; 2: ancient site; 3: main road; 4: vibracoring site (OIN); 5: karstic spring; 6: periodically running creek (rema); 7: paleo river channel (oxbow); 8: Holocene Acheloos alluvial plain; 9: Triassic bedrock (limestone, dolomite).
1: village actuel; 2: site antique; 3: route principale; 4: site de prélèvement de carotte; 5: source karstique; 6: ruisseau intermittent; 7: ancien chenal fluviatile (méandre); 8: plaine alluviale holocène de l’Acheloos; 9: substratum rocheux du Trias (calcaire, dolomie).
4The Acheloos River delta plain encloses several former islands of the Echinades archipelago (fig. 1). The rocky hills are made up of Triassic limestone, limestone breccia, dolomite and locally gypsum, and are thus strongly karstified (Bousquet, 1976; IGME, 1989). In general, they are bound to SSW-NNE and SE-NW-striking local tectonic fault systems that are connected to the nearby Amfilochia fault zone, an active rift running between the gulfs of Ambrakia and Patras and separating the Akarnanian mass from the central Greek mainland (Haslinger et al., 1999). Most parts of the delta are characterized by local subsidence (Vött, 2007). Ancient Oiniadai lies on top of the former Trikardo island (99 m a.s.l.) in the centre of the delta.
5Historically, Oiniadai was of great strategic importance as it is located at the entrance to the gulfs of Patras and Corinth. The Akarnanian islet was attacked repeatedly between 454 BC and the 3rd century BC by Greek powers. In 252 BC, it came under Aetolian rule (Oberhummer, 1887: 83, 101; Powell, 1904: 139; Freitag, 1994: 223). The Macedonian king Philip V conquered the city in 219 BC, renovated the shipsheds, fortified the city walls and gave Oiniadai back to the Akarnanians. In 212 BC, Trikardo was captured by the Romans and, a few years later, in 189 BC, reintegrated into the Akarnanian League. Then, historical traces quickly fade and the site sank into insignificance. In 167 BC, all Akarnania became part of the Roman Republic (Powell, 1904: 144).
6Lateral and vertical variations of sedimentary facies in geological archives were determined in order to reconstruct palaeoenvironmental conditions and their changes in time and space. In the Acheloos River delta, 70 sediment cores were retrieved by means of an Atlas Copco vibracoring device (Cobra mk1) with core diameters of 6 or 5 cm. For this paper, 15 vibracores drilled in the vicinity of Oiniadai were analyzed, arranged in five transects. The maximum recovery depth of the vibracores was 15 m below surface (m b.s.). In the field, the sedimentary environment was classified based on sedimentological criteria and macrofossil remains, such as gastropods, bivalves, seeds, or plant fragments. Sampling of vibracore profiles allowed for detailed micro- and macrofaunal studies as well as for geochemical analyses of parameters such as electrical conductivity, pH-value, loss on ignition, carbonate content and concentrations of (ortho-)phosphate, (earth-)alkaline and heavy metal ions measured in a chemical pulping on the basis of concentrated hydrochloric acid. Facies determination was mainly based on ostracod species and assemblages because these are reliable ecological indicators (Frenzel and Boomer, 2004). Vertical profiles of geochemical parameters and multivariate discriminant analyses of geochemical datasets were also valuable tools for controlling and determining sedimentary facies (Vött et al., 2003). Macrofloral remains and pollen found in sediment samples gave additional information about the palaeoenvironment. A differential GPS (Leica SR 530) was used to determine the position and elevation of vibracoring sites. Earth resistivity tomography (Syscal R1 plus, Iris instruments) helped to detect subsurfaces structures and the distribution pattern of sedimentary sequences. Twenty-three 14C-AMS dates and relative age determinations of diagnostic ceramic fragments were used to establish a geochronological framework. The reservoir effect for radiocarbon-dated marine samples was corrected for an average of 402 years (Reimer and McCormac, 2002). All radiocarbon ages given in the paper were calibrated (cal BC/AD) using the calibration software Calib by Stuiver et al. (2006). Corona satellite photos (USGS, 1965), Landsat 7 ETM+ and Aster (2001) images helped to reconstruct palaeogeographical changes.
7Oiniadai’s northern harbour (figs. 1 and 2) is made up of an outer and an inner harbour section. The shipsheds lie at the eastern side of the outer embayment, which is almost 250 m wide. Five slipways were used for repairing triremes, ancient Greek warships, and for storing them during winter season (Kolonas, 1992). The inner harbour, only 80-120 m wide, is located southeast of the dockyard in a narrow indentation. In antiquity, it could be closed by a chain (Murray, 1982: 42). Transect I covers both the inner (OIN 48) and outer harbours (OIN 54). OIN 47 was drilled in front of the shipsheds.
Table 1 – Radiocarbon dating results for samples from vibracores around Trikardo, central Acheloos River delta.
Tableau 1 – Datations radiocarbone des carottes du secteur de Trikardo, centre du delta de l’Acheloos.

b.s.: below ground surface; b.s.l.: below sea level; B. latreilli: Bittium latreilli; C. rupestre: Cerithium rupestre; D. exoleta: Dosinia exoleta; P. conica: Pirenella conica; spec.: specimen(s); *: marine reservoir correction with 402 years of reservoir age; 1s max; min cal BP/BC (AD): calibrated ages, 1s-range; “;”: there are several possible age intervals because of multiple intersections with the calibration curve; Lab. No.: laboratory number; University of Utrecht (UtC), University of Erlangen-Nürnberg (Erl).
b. s.: sous la surface du sol; b. s. l.: sous le niveau de la mer; B. latreilli: Bittium latreilli; C. rupestre: Cerithium rupestre; D. exoleta: Dosinia exoleta; P. conica: Pirenella conica; spec.: specimen(s); *: correction des calibrations pour un âge réservoir de 402 ans; 1s max; min cal BP/BC (AD): datations calibrées, 1s-range; “;”: il peut y avoir plusieurs intervalles d’âge possibles du fait d’intersections multiples avec la courbe de calibration; Lab. No.: référence de l’échantillon, laboratoire de l’université d’Utrecht (UtC) ou de l’université d’Erlangen-Nürnberg (Erl).
8The facies distribution pattern shows that at OIN 54 and OIN 47, open marine conditions prevailed until circa 4400 cal BC (sample OIN 47/18 M2: 4433-4315 cal BC; fig. 3, table 1). Around that time, an Acheloos River distributary approached from the north and sealed off a large lagoon north of Trikardo. OIN 48 was never affected by open marine waters but, later, by the transgressive lagoonal unit overlying alluvial-fan palaeosol sediments. The lagoonal sediments, which were encountered at OIN 47, 48 and 54, were of an olive-green colour and showed abundant shells and shell fragments of a macrofaunal assemblage characteristic of shallow marine to brackish conditions (such as Carcinus sp., Cerastoderma glaucum, Cerithium rupestre, Cyclope neritea, Dosinia exoleta, Gibbula sp.). High content of organic material, up to almost 10%, indicates almost swampy conditions. Between 5 m and 4 m below present sea level (m b.s.l.), sedimentary conditions at OIN 47 and 48 were abruptly affected by strong freshwater input as shown, for instance, by a decrease in electrical conductivity of the sediment (fig. 4a) and by a change towards a (light) grey colour. The occurrence of fine sand indicates mid- to high-energy influence on the quiescent conditions, which until then were typical of the harbour basin. At the same time, pollen of Abies sp. increased to maximum concentrations. This proves a direct influence of Acheloos River water (Jahns, 2005, pers. comm.) as (i) the tree only exists in high mountain ecosystems such as the Pindos mountains in the catchment area of the Acheloos River, and does not occur in southern Akarnania itself and (ii) pollen from Abies sp. is not transported by wind over long distances. Acheloos river inflow occurred via a distributary which, around 1000 cal BC, branched off the deltafront east of Trikardo and flowed along the eastern and northern flank of the island (sample OIN 47/14+ M: 1151–1007 cal BC; fig. 3, see Vött et al., 2007a for further details). OIN 54 stratigraphy indicates that the river channel into the harbour was still active in the 4th century BC (sample OIN 54/12+ PR: 511–394 cal BC; fig. 3, table 1). Around 100 cal BC, the lagoonal waters of the harbour turned limnic (sample OIN 47/11 PR: 171–2 cal BC; fig. 3, table 1). Until that time, high Abies sp. pollen counts and low salinity of the sediment reflect ongoing Acheloos-borne freshwater inflow (fig. 4a). Groundwater has kept a small outflow open towards the present-day drainage channel which flows by the harbour’s entrance and is mostly fed by high-discharge karstic springs at the foot of the adjacent Lesini mountains.
9Figure 4 exemplarily depicts geochemical parameters analysed for sediment samples from vibracores OIN 47, 48 and 54 clearly reflecting facies distribution patterns found for the northern harbour (fig. 3). Electrical conductivity values found for OIN 47 and 54 show a steep, partly stepwise decrease towards the top corresponding to the beginning of strong freshwater inflow to the outer harbour (fig. 4a). At OIN 54, sedimentary conditions were considerably influenced by karstic springs located in the innermost harbour area (fig. 2). Minimum contents of (ortho-)phosphate at OIN 47 and 48 are characteristic of quiescent lagoonal environments and of phases of increased freshwater input (fig. 4b). High Ca–Mg ratios indicate periods of increased supply of calcium from a marine source (fig. 4c), whereas maximum Fe–Na ratios stand for increased influence of fluvial and terrestrial processes on the ecosystem (fig. 4d).
Fig. 2 – Detailed map of Trikardo island and selected archaeological remains of ancient Oiniadai.
Fig. 2 – Carte détaillée de l’île de Trikardo et principaux vestiges archéologiques de la cité antique d’Oiniadai.

Locations of vibracoring sites and transects of earth resistivity tomography in the vicinity of the former island. Map based on Corona satellite photo (USGS, 1965) and archaeological data from Powell (1904), Kirsten (1937), and Kolonas (1992).
Localisation des différentes carottes et transects tomographiques à proximité de l’ancienne île. Carte établie d’après l’image satellitale (USGS, 1965) et les données archéologiques tirées de Powell (1904), Kirsten (1937) and Kolonas (1992).
Fig. 3 – Distribution pattern of sedimentary facies in Oiniadai’s northern harbour (transect I).
Fig. 3 – Chronostratigraphie des faciès sédimentaires du port nord d’Oiniadai (transect I).

F: fluvial (crevasse splay, flood channel); H: limnic to fluvial (flood plain or fresh marsh); T: semi-terrestrial (peat, calcareous gyttja); Ss: semi-terrestrial to limnic (swampy lake); SI: semi-terrestrial to lagoonal (swampy lagoon); R: fluviomarine (river channel); D: fluviomarine (delta); S: limnic (freshwater lake); SF: fluvial to limnic (riverwater inflow into freshwater lake); L: brackish to brackish-marine (lagoon); LF: limnic to brackish (lagoon with river water influence); P: brackish to shallow marine (prodelta); S/T: brackish to shallow marine (storm layer, tsunami deposit); B: shallow marine, littoral (sand bar, tombolo, beach); Fe: shallow marine, littoral (rocky and supra littoral zone); FI: shallow marine, sublittoral (marine embayment); F (on black): early to mid-Holocene alluvial fan (palaeosol); PS: Pleistocene-Holocene palaeosol; BR: bedrock.
F: fluviatile (dépôt de crue); H: lacustre à fluviatile (plaine d’inondation, marécage); T: dépôt organo-terrigène (tourbe, gyttja calcaire); Ss: organo-terrigène à lacustre (lac marécageux); SI: dépôt de lagune marécageuse; R: dépôt fluvio-marin (chenal fluviatile); D: dépôt fluvio-marin (delta); S: dépôt lacustre; SF: fluviatile à lacustre (dépôt d’origine fluviatile dans un lac); L: dépôt lagunaire; LF: lacustre à saumâtre (lagune influencée par des eaux fluviales); P: dépôt saumâtre à marin de faible profondeur (prodelta); S/T: saumâtre à marin de faible profondeur (dépôt de tempête, de tsunami); B: dépôt littoral (cordon sableux, tombolo, plage); Fe: dépôt supralittoral et littoral rocheux; FI: dépôt sublittoral (baie); F (sur fond noir): paléosol sur cône alluvial (Holocene ancien et moyen); PS: paléosol (Pléistocène-Holocène); BR: substrat.
Fig. 4 – Selected geochemical parameters analysed for sediment samples from vibracores OIN 47, 48 and 54 (transect I).
Fig. 4 – Analyse géochimique des échantillons des carottes OIN 47, 48 et 54 (transect I).

Vertical profiles of geochemical parameters such as electrical conductivity (a), concentration of (ortho-)phosphate (b) as well as calcium-magnesium (c) and iron-sodium ratios (d) are useful tools to determine facies distribution patterns (see Vött et al., 2003).
L’évolution verticale des paramètres géochimiques tels que la conductivité électrique (a), la concentration en (ortho-)phosphate (b), les rapports calcium-magnésium (c) et fer-sodium est utile pour déterminer la chronostratigraphie des faciès sédimentaires (cf. Vött et al., 2003)
10Additionally, earth resistivity tomography was carried out along two transects in the outer harbour (oin-g-10 and oin-g-11, fig. 5). Inverse model resistivity sections show bedrock units below 18 m b.s. They are covered by marine sediments with extremely low resistivity values. Increasing resistivity towards the top indicates the change towards lagoonal and later limnic deposits. Figure 5 also illustrates the river channel that entered the harbour from the north.
Fig. 5 – Simplified inverse model resistivity sections for the northern harbour and the southern embayment based on earth resistivity measurements.
Fig. 5 – Profils géoélectriques du port nord d’Oiniadai et de la baie Sud de Trikardo.

Logs represent simplified facies profiles of vibracores.
Les logs représentent les faciès simplifiés des carottes.
11In summary, the northern flank of Trikardo was exposed to a lagoon since around 4400 cal BC when a large Acheloos distributary, prograding southward, closed the area off from open marine conditions. From Mycenaean to early Roman times, the harbour bay was connected to the sea via a lagoon which gradually silted up through the present. Its remains can still be seen north of Kounovina. Our data provide evidence of freshwater inflow by an Acheloos River channel into the northern harbour between approximately 1000 and 100 cal BC. However, it is still unclear (i) whether this river channel was of natural or anthropogenic origin and (ii) whether it was used as a waterway to reach the shipsheds from the east. Around 100 cal BC, the northern harbour turned into a limnic environment and silted up soon afterwards.
12The southern embayment, scarcely 200 m long and 100 m wide, is located at the southern flank of Trikardo just opposite to the northern harbour and exposed towards the present course of the Acheloos River (figs. 1 and 2). Oiniadai’s fortification wall crosses the swampy ground and divides the indentation into two parts. The wall itself has a gap, several meters wide, which might be interpreted as the entrance to the inner section of a port. Transect II runs from the inner part of the bay (OIN 49), north of the wall, to the (central) outer part of it (OIN 63), south of the wall. OIN 41 was drilled at the entrance of the embayment (fig. 2).
13Facies distribution patterns prove that OIN 49 underwent an evolution almost completely detached from the outer bay (fig. 6). The profile is made up of peat and lake deposits. At 4.64 m b.s.l., an intermediate lens of medium sand, 11 cm thick, as well as fine sand components encountered in the overlying limnic sediments (4.53-3.87 m b.s.l.) reflect a sudden and temporary fluvial impact. In contrast, sedimentary sequences at OIN 41 and 63 start with littoral deposits. At OIN 63, well sorted medium to coarse sand with abundant fragments of marine macrofauna such as Acanthocardia sp., Arca noae, Conus sp., Ostrea sp., Echinoidea and corals reflects a beach environment. At OIN 41, silty fine sand with fragments of marine shells and Crustacea represents the corresponding (sub-)littoral shoreface unit. Subsequent laminated prodeltaic deposits found in vibracore OIN 41 show that, later, a distributary of the Acheloos River approached the site. Soon afterwards, the distributary reached the southern embayment, partly eroded an older peat layer at OIN 63 and flushed into the inner section of the bay as indicated by the sandy intercalations encountered at OIN 49. Radiocarbon dating of this event yielded an age of circa 3100 cal BC (sample OIN 49/15+ PR: 3261–2925 cal BC; fig. 6, table 1). Muddy lagoonal deposits encountered on top of the deltaic sand units outside the wall bear signs of freshwater input such as fragments of a brackish macrofauna, a light sediment colour and high amounts of fine sand of fluvial origin. Strong fluvial influence to the outer section of the bay persisted at least until the end of the 1st millennium BC (fig. 6). The subsequent lagoonal environment existed considerably longer compared to the northern harbour. A layer of shell debris, deposited at OIN 41 possibly due to high-energy storm or tsunami wave action, dates to the 2nd century AD (sample OIN 41/14+ M: 94–196 cal AD; fig. 6, table 1). During late Byzantine times, the lagoon was filled up by alluvial sediments (fig. 6).
Fig. 6 – Distribution pattern of sedimentary facies in the southern embayment of Trikardo (transect II).
Fig. 6 – Chronostratigraphie des faciès sédimentaires de la baie méridionale de Trikardo (transect II).


14Earth resistivity measurements along a N–S trending transect (oin-g-12, fig. 5) revealed that in the midst of the gap in the wall, immediately south of OIN 49, high resistivity values reach up to 3.75 m b.s. (about 3.10 m b.s.l.) and are bound to almost rectangular contours. It is concluded that the wall was originally erected on top of the bedrock. Later, when the fortification system was out of use, it was partly torn down. Even if the uprising structure was made out of bedrock, the inner section could never have been used as a harbour because the relative sea level was at approximately 3.25-3.10 m b.s.l. during Classical-Hellenistic times (Vött and Brückner, 2006; Vött et al., 2007a; Vött, 2007), i.e. at the upper edge of the low resistivity structure. Figure 5 also shows that the fluviodeltaic and fluviolagoonal sequences found at OIN 63 correspond to a river bend which tangentially flowed into the outer section of the bay.
15The swamp around OIN 49 is fed by karstic springs north of the vibracoring site (fig. 2). Not far to the northwest of the indentation, a sinkhole, 60 m wide and approximately 20 m deep documents strong karstification. The inner section of the southern bay may therefore correspond to a kettle-like karstic hollow incorporated into the fortification system to ensure freshwater supply for the city or simply to shorten the length of the wall.
16Seven vibracorings were carried out around a circular-shaped bedrock outcrop lying around 250 m east of the southeastern fringe of Trikardo (fig. 1). Transect III runs from north to south (OIN 44, 64, 65 and 69), and transect IV (OIN 66, 67 and 68) from east to west (figs. 2, 7a, 7b and 8).
17The base of OIN 44 shows lagoonal sediments which consist of silty clay and include abundant fragments of a brackish macrofauna. The sediments were accumulated in a shallow-marine to brackish environment on top of the bedrock. The lagoonal transgression dates to the 6th millennium BC (sample OIN 44/28 PR: 5465–5322 cal BC; fig. 7a, table 1). A subsequent peat layer (4.87–4.00 m b.s.l.) contained several ceramic fragments which were dated to late Helladic or even earlier times (Lang, 2005, pers. comm.). The pit of a fig (4.33 m b.s.l.) and a piece of bone, 5 cm long (4.10 m b.s.l.), as well as abundant fragments of Hexaplex trunculus, up to 5 cm large, were also found within the peat unit. Obviously, the swampy site was used as waste dump and, at the same time, lay close to a wharf used by fishermen. Like Bolinus brandaris, Hexaplex trunculus, an edible species, was used to produce dye for colouring textiles. The species is well adapted to polluted harbour areas (Poppe and Goto, 1991: 136). Two 14C-AMS dates from the lower (4.83 m b.s.l.) and the upper (4.10 m b.s.l.) part of the peat layer yielded ages of 3088–2926 cal BC (sample OIN 44/20+ PR) and 2902-2713 cal BC (sample OIN 44/17 PR; fig. 7a, table 1), respectively. These results prove an early Helladic age of the ceramic fragments. OIN 44 thus revealed, for the first time, early Bronze Age colonization of Trikardo. The upper part of the peat unit and the subsequent lagoonal deposits contained several stones, up to 6 cm large, which seem to be filled in by man in response to rising sea level. Upward increasing contents of fine sand indicate strong fluvial input to the brackish water body, before it was silted up by alluvial sediments. This is confirmed by the distribution pattern of electrical conductivity values of sediment samples taken from vibracore OIN 44. Electrical conductivity decreases abruptly from 1.5-2.5 mS/cm for samples between 6 and 4 m b.s.l. from lagoonal environments to values around 0.5 mS/cm for samples taken above 4 m b.s.l., which were strongly affected by fluvial influence.
Fig. 7a – Distribution pattern of sedimentary facies in Trikardo’s southeastern harbour (transect III).
Fig. 7a – Chronostratigraphie des faciès sédimentaires du port sud-est de Trikardo (transect III).
Fig. 7b – Distribution pattern of sedimentary facies in Trikardo southeastern harbour (transect IV).
Fig. 7b – Chronostratigraphie des faciès sédimentaires du port sud-est de Trikardo (transect IV).


18The base of OIN 65 shows a rocky (supra-)littoral facies, followed by lagoonal and then limnic deposits both of which were influenced by river water inflow (fig. 7a). The limnic sediments are covered by marsh deposits and a subsequent palaeosol both of which contained numerous ceramic fragments. However, most of the potsherds were weathered and could not be used for age determination. At OIN 64, diagnostic sherds were found in the freshwater marsh unit (0.93–1.33 m above present sea level a.s.l.) and the palaeosol (1.33–1.52 m a.s.l.) and were dated to Roman to Byzantine times (Lang, 2005, pers. comm.). As the OIN 65 ceramic findings belong to the same facies units it is assumed that they are also of Roman to Byzantine age. Bedrock encountered at OIN 64 was not affected by marine bioerosion and a lagoonal unit is missing. The sedimentary sequence of OIN 69, drilled some 50 m north of the bedrock outcrop, is similar to OIN 65, but shows a thick stratum of well sorted fluvial sand. As the profile is void of ceramics, anthropogenic activity seems to have been concentrated on the outcrop itself.
1914C-AMS dating of a rock boring mussel from the (supra-) littoral zone at OIN 69 (2.92 m b.s.l.) showed that bio-erosion took place until 3060–2915 cal BC (sample OIN 69/13 ST/M/BE; fig. 7a, table 1). At the same time, peat started growing at OIN 44 (sample OIN 44/20+ PR: 3088–2926 cal BC, 4.83 m b.s.l.; fig. 7a, table 1). The change from bio-erosion to peat formation was caused by a major shift in coastal dynamics. It is known that an Acheloos River distributary reached OIN 49 (transect II) shortly after 3261–2925 cal BC (sample OIN 49/15+ PR; fig. 6, table 1). This distributary had already passed by south of the bedrock outcrop close to OIN 44 and OIN 69 (transect III) shortly before 3088–2926 cal BC (sample OIN 44/20+ PR; fig. 7a, table 1), i.e. almost at the same time. The distributary, rapidly prograding westward, sealed off a part of the lagoon southeast of Trikardo and induced quiescent hydrodynamic conditions favourable to peat formation at a site where, shortly before, rocks in supra-littoral position were affected by marine bio-erosion. The radiocarbon dated wood fragment from OIN 69 (sample OIN 69/14 HR: 2892–2702 cal BC, 3.21 m b.s.l.; fig. 7a, table 1) is assumed to have been washed in after bio-erosion had already stopped.
20It is suggested that bio-erosion at OIN 65 occurred at the same time as found for OIN 69. The 14C-AMS dated juvenile specimens of brackish gastropod species encountered at OIN 65 (sample OIN 65/14 M2: 1375–1247 cal BC; fig. 7a, table 1) originate from quiescent lagoonal waters and seem to have possibly been thrown on top of the rocky shore around 1300 cal BC by wave action. Earth resistivity tomography gives a clear picture of the bedrock outcrop showing that its northern and southern flanks are steeply inclined (oin-g-5, fig. 11). A large river channel to the north of the outcrop is assumed to be responsible for the deposition of fluvial sediments at OIN 69.
Fig. 8 – Bedrock outcrop southeast of Trikardo.
Fig. 8 – Affleurement du substrat au sud-est de Trikardo.

Remains of a wall with abundant ceramic fragments dating to Roman to Byzantine times (small photo to the left, white arrows), as well as with a flint of probably prehistoric age (small photo to the right). Ceramic findings in vibracores revealed that the bedrock was used as a quay on a lagoonal shore during early Helladic times (OIN 44) and at the riverside of a large Acheloos meander during Roman to Byzantine times (OIN 64, 65). Nowadays, the site is almost completely covered by alluvial deposits. The left background shows Oiniadai’s acropolis, the right middleground the entrance to the necropolis bay. View to the northwest (photos taken by Vött, 2005).
Les vestiges d’un mur et de nombreux fragments de céramique témoignent d’une occupation humaine aux périodes romaine et byzantine (petite photo à gauche, flèches blanches). Un éclat de silex (petite photo à droite) date vraisemblablement des temps préhistoriques. La présence de céramiques indique que l’affleurement rocheux servait de quai sur la berge d’une lagune, à l’époque Helladique Ancien (OIN 44), et sur la berge d’un vaste méandre de l’Acheloos aux périodes romaine et byzantine (OIN 64, 65). De nos jours, le site est complètement colmaté par des alluvions. La photographie d’ensemble représente l’acropole d’Oinaidai. Vue vers le nord ouest (clichés Vött, 2005).
21Cores OIN 66 and 67 were drilled between the outcrop and Trikardo (figs. 1, 2 and 7b). Bedrock was encountered in both profiles. This proves that, in former times, a W–E trending rocky promontory existed which is now partly covered by alluvial deposits. At OIN 68, river channel sediments include numerous weathered ceramic fragments which seem to be associated to the Roman to Byzantine sherds found in adjacent overbank marsh deposits at OIN 64 and 65. Geoelectric measurements along a W-E transect (oin-g-6, fig. 11) show that the fluvial sediments of OIN 68 were deposited by a large meandering Acheloos River channel east of the bedrock outcrop. It is still unclear, however, whether fluvial deposition was linked to natural crevasse dynamics or to an artificial canal. The fact that soil formation occurred on the top of the marsh deposits at OIN 64, 65 and 69 indicates that sediment supply by the Acheloos River stopped some time after Roman to Byzantine times.
22To summarize, our results show (i) that the southeastern part of Trikardo lay on the rocky shore of a large lagoon until about 3000 cal yr BC, (ii) that an Acheloos distributary passed by shortly afterwards in a westward direction and induced quiescent hydrodynamic and locally swampy conditions and (iii) that an Acheloos meander existed near the rocky promontory southeast of Trikardo during Roman to Byzantine times. From an archaeological point of view, it appears (i) that Trikardo was colonized in early Helladic times, (ii) that, then, a harbour existed on the shore of a swampy lagoon at the southeastern fringe of the island and (iii) that in the same area, people were living on the banks of a large Acheloos meander during Roman to Byzantine times.
23The necropolis bay, located east of the acropolis, extends almost 250 m both in N–S and in E–W direction (figs. 1 and 2). The ancient main road to Oiniadai runs above the northern shore of the bay where recent archaeological excavations have revealed the remains of the necropolis. The road leads towards the main gate of the polis (Powell, 1904: 157).
24Transect V consists of two vibracores, one in the northern part of the indentation (OIN 70), the other some 650 m further east (OIN 10) in the open alluvial plain. The base of OIN 10 shows sand bar deposits of a NE–SW running tombolo system between the Lesini mountains and Trikardo which was formed during an early phase of westward delta progradation from the area around Katochi (Vött et al., 2007a). An intermediate phase of lagoonal conditions is documented by both macro- and microfaunal indicators (figs. 9, 10a and 10b). Subsequently, rising sea level and decreased sediment supply during the 6th and 5th millennia BC when the main distributary of the Acheloos River delta approached Trikardo from a northern direction (see above, Vött et al., 2007a) allowed the tombolo to become partially submerged and favoured predominantly shallow marine conditions. Increased saltwater influence resulted in a high biodiversity (figs. 10a and 10b). The Acheloos River distributary which approached Trikardo from the north caused a prodeltaic facies and later induced lagoonal conditions at OIN 10. The lagoonal environment east of the necropolis bay lasted at least until the end of the 4th millennium BC (sample OIN 10/19 M: 3482–3372 cal BC; fig. 9, table 1). A subsequent change in the flow direction of the Acheloos distributary led to the partial erosion of lagoonal deposits and the accumulation of deltaic sand, several metres thick.
Fig. 9 – Distribution pattern of sedimentary facies in the necropolis bay of Oiniadai (transect V).
Fig. 9 – Chronostratigraphie des faciès sédimentaires de la baie de la nécropole d’Oiniadai (transect V).


25Shallow marine conditions at OIN 70, located in a sheltered position, were not affected by prodelta dynamics. The subsequent lagoonal phase set in around 4306–4164 cal yr BC (sample OIN 70/15 M: 4306–4164 cal BC; fig. 9, table 1) which shows good correspondence with the results obtained for OIN 47 (see fig. 3). Sedimentary facies changed when the site came under strong fluvial influence some time after 1419–1314 cal BC (sample OIN 70/10+ PR; fig. 9, table 1), and this is documented by fine sandy intercalations, reduced salinity and a lighter colour of the lagoonal deposits. At OIN 70, lagoonal deposits reach up to 2.69 m b.s.l. Compared to the relative sea level evolution near Trikardo (Vött and Brückner, 2006; Vött et al., 2007a; Vött, 2007), it is concluded that the lagoon persisted until circa 250 cal AD. Then, conditions turned limnic and the bay was filled up with alluvial sediments.
Fig. 10a – Bivalves and gastropods encountered in sediment samples of vibracore OIN 10 (transect V), compared to grain size distribution.
Fig. 10a – Granulométrie et contenu malacologique des échantillons de la carotte OIN 10 (transect V).
Fig. 10b – Species and associations of ostracods and foraminifers encountered in sediment samples of vibracore OIN 10 (transect V), compared to grain size distribution.
Fig. 10b – Analyse des ostracodes et des foraminifères identifiés au sein de la carotte OIN 10 (transect V).
Fig. 11 – Simplified inverse model resistivity sections for the southeastern harbour and the necropolis bay based on earth resistivity measurements. Logs represent simplified facies profiles of vibracores.
Fig. 11 – Profils géoélectriques du port sud-est et de la baie de la nécropole. Les logs représentent schématiquement les faciès des carottes.


26Earth resistivity tomography along a SE–NW transect east of the indentation revealed a wide zone of low conductivity values, almost 350 m wide, extending from around 6 m b.s.l. to the terrain surface and corresponding to the deltaic sediments encountered at OIN 10 (oin-g-9, fig. 11). Similar findings were made in geoelectrical profiles oin-g-5 and oin-g-6 (fig. 11). The distributary reached OIN 10 and prograded towards the west where it came into contact with the Trikardo hill. River water flowed into the necropolis bay and deposited fluvial sediments at the entrance to the northern harbour (OIN 54, OIN 47). Fluvial influence in the eastern (OIN 70) and northern (OIN 47) embayments of Trikardo started quasi contemporaneously between 1300 and 1000 cal BC (samples OIN 47/14+ M and OIN 70/10+ PR; figs. 3 and 9, table 1). From that time on, lagoonal conditions in these bays were characterized by ongoing river water input. Later, when a large river meander was formed east of Trikardo, a channel branched off near OIN 70 and followed the eastern and northern flanks of the hill. This is shown by Corona satellite photos (USGS, 1965) as well as by younger generation river channels detected by earth resistivity tomography (fig. 11). The 14C-AMS date obtained for the uppermost sample of OIN 70 (sample OIN 70/8+ M: 1363–1230 cal BC; fig. 9, table 1) turns out to be unreliable when compared to relative sea level evolution.
27In summary, the sedimentary sequences encountered in the northern harbour (OIN 47) and in the necropolis bay (OIN 70) are almost identical. They show thick lagoonal units strongly influenced by fluvial input that started around 1300–1000 cal BC. The lagoonal environment, strongly affected by Acheloos River water inflow, persisted until about 250 cal AD. Although there is no information on harbour installations in the necropolis bay, natural conditions were highly appropriate for a port in Antiquity. The site was protected by the acropolis and lay close to the main road towards the city. It thus represented an ideal position for a commercial harbour. In this regard, further archaeological and geomorphological research is needed to confirm this hypothesis.
28Many ancient accounts on Oiniadai unanimously describe the site as well protected by the surrounding swampy lowlands and mention that it is located at the riverside of the Acheloos (Freitag, 1994; Vött et al., 2007b). Strabo (63 BC–26 AD), for instance, noted that Trikardo lies 70 stades (about 13 km) distant from the river mouth at his time (Strabo 10, 2, 21). However, there is no specific information on the harbours of the polis. Only Polybius (210–127 BC) reported that, in 219 BC, Philip V intended to incorporate the harbour and the dockyard into the existing fortification system and shipped building material to Oiniadai via the Acheloos River (Polybius 4, 65, 2–9).
29Leake (1835, vol. III: 564ff.) and Sears (1904: 227, 235) speculated that a canal independent from the Acheloos River guaranteed the communication of the northern harbour with the sea. Heuzey (1860: 447f.), Lolling (1876/77: 282) and Philippson (1958: 402ff.) were convinced that this canal was connected to the main stream. Lang (1905: 29ff.) and Lehmann-Hartleben (1923: 80) suggested that the northern port opened to a marine embayment or to a lagoon. So did Kirsten (1937: 2207). He also thought that a bayou of the Acheloos River supplied additional inflow to the harbour. Freitag (1994: 232) and Fouache et al. (2005) concluded that the city lay at the marine front. Villas (1984: 115), however, had already vibracored in the harbour basin and found lagoonal deposits for the time when the shipsheds were in use (Murray, 1982: 40ff.).
30Our data are in good agreement with the accounts of the ancients and confirm Villas’ (1984) results: in antiquity, the northern harbour communicated with the Ionian Sea via a lagoon, the Lagoon of Oiniadai, into which an Acheloos River channel discharged freshwater (figs. 1, 2, 3 and 5). The channel partly entered the northern harbour and thus decelerated its silting up (Vött et al., 2004). Following Kirsten (1937: 2207) and Freitag (1994: 232), the Lagoon of Oiniadai is not the Lake Melite of the ancients. The latter should rather be located between Trikardo and Skoupas (Vött et al., 2007a).
31Murray (1982: 43) suggested that the southern embayment may have been used as harbour. He interpreted an earthen dam amidst the marshy ground as a pier. According to OIN 49 and earth resistivity data, the inner section of the indentation was instead probably used to retain freshwater from the adjacent karstic springs, i.e. as a reservoir for drinking water (see above, figs. 2, 5 and 6). The dam was probably constructed by peasants after the city was abandoned in order to cross the swamp on dry ground. If there was a harbour at all during ancient times, it must have been outside the wall. Landing would have been possible at the shore of the fluviomarine channel encountered at OIN 63 as well as on the adjacent lagoonal shores.
32Based on Polybius’s accounts, Leake (1835, vol. III: 168) and Heuzey (1860: 455) were convinced that, during Classical-Hellenistic times, there was another harbour at the riverside southeast of the island. They argued that Philip V would not have had to incorporate the shipsheds into the fortification system as they were already part of it (contra: Murray, 1982: 36ff). Bursian (1862: 122), Oberhummer (1887: 245), Weil (1903: 344) and Philippson (1958: 403) adopted their view whereas Lehmann-Hartleben (1923: 110f) and Kirsten (1937: 2207) rejected it.
33Our studies revealed an early Helladic harbour southeast of Oiniadai along the swampy shore of a lagoon (figs. 7a and 7b). In addition to the findings of a submerged early Helladic settlement in the Bay of Platiyali near Astakos (Delaporta and Spondylis, 1990; Delaporta et al., 1990), this is another proof of early Bronze Age human activity in Akarnania. The lagoon east of Oiniadai was strongly affected by river water inflow since 1300–1000 cal BC and existed until 250 cal AD (see above). Earth resistivity tomography revealed a former promontory of Trikardo, now partly covered by alluvial deposits, which may have been used as a quay during Classical-Hellenistic and Roman times. The bedrock surface at OIN 64, OIN 68 and OIN 69 reaches up to 2.78–2.42 m b.s.l. According to the local sea level curve (Vött and Brückner, 2006; Vött et al., 2007a; Vött, 2007) it should have been possible to use the promontory as a quay until around 250 cal AD (figs. 7a, 7b and 11). Saltwater influence at OIN 65, OIN 68 and OIN 69, however, persisted up to around 2.10 m b.s.l. Presumably, this is due to landward saltwater inflow via the Acheloos River meander which flowed by close to the promontory during Roman to Byzantine times and which was used as a river harbour. However, further research will be necessary to achieve a better temporal resolution for the palaeogeographical situation around the promontory during antiquity. It seems further probable that, during Classical-Hellenistic and Roman times, a commercial harbour existed at the necropolis bay which – similar to the conditions found for the northern harbour – was affected by considerable freshwater input by an Acheloos River channel running along the eastern and northern flanks of Trikardo (figs. 9 and 11).
34Murray (1985) localized ancient Nasos on the former island of Skoupas and suggested that it was founded between 252–219 BC when manoeuvering to Oinidai’s harbours had become difficult due to ongoing siltation. Cargo and passengers could have been shipped to Trikardo on flat-bottomed river boats (Murray, 1985: 106). Vött et al. (2004, 2007a) showed that Skoupas was at the seafront until late Byzantine times and that during Classical-Hellenistic to Roman times, the main mouth of the Acheloos River delta lay southeast of the island representing an ideal waterway in a shallow water lagoonal system. Although it has to be expected that the people of Oiniadai were aware of the silting up of their harbours, sedimentological evidence of dredging was not encountered (Marriner and Morhange, 2006).
35Based on sedimentological, geochemical, macro- and microfaunal and palynological analyses of sediment samples from 15 vibracores, on the relative age determination of ceramic fragments as well as on 22 14C-AMS dates, the following conclusions can be made.
36(i) Around 3000 cal yr BC, in early Helladic times, a westward prograding distributary of the Acheloos River delta induced swampy conditions in the lagoon at the southeastern fringe of Trikardo. The quiescent environment was used as waste dump and port by early Helladic fishermen. A rocky promontory, nowadays covered by alluvial deposits, served as natural quay (figs. 1, 2, 7a, 7b, 8 and 11).
37(ii) Around 1300–1000 cal yr BC, in late Helladic times, an Acheloos River distributary approached Trikardo and flowed along its eastern and northern flanks. Lagoonal environments in the necropolis bay as well as in the northern embayment thus were strongly affected by freshwater inflow (figs. 1, 2, 3, 5, 9 and 11). Both bays represented ideal harbour sites. The natural quay in the southeast was possibly still in use. Further archaeological studies, however, are needed to clarify if Trikardo was populated at all at that time.
38(iii) Until Classical-Hellenistic to early Roman times, the northern harbour basin still experienced considerable river water inflow. The shipsheds were connected to the Ionian Sea via a lagoonal system intruding from western direction. The harbour finally silted up around 1 BC/AD (figs. 1, 2, 3 and 5). The necropolis bay offered similar anchoring conditions (figs. 9 and 11). The southeastern promontory was probably still used as a wharf on a lagoonal shore. Later, until Roman to Byzantine times, a harbour existed along the riverbank of a nearby Acheloos meander (fig. 7a and 7b).
39(iv) The inner section of the southern embayment never served as a harbour. Simple landing sites may have existed outside the city wall (figs. 1, 2, 5 and 6).
40It could be shown that Trikardo was of great importance as a seaport during 4000 or so years of human activity on the former island. However, the people of Trikardo never had a harbour on the sea frond. Instead, they preferred quiescent lagoonal environments and used the Acheloos River channels as waterways and to protect their harbour basins from rapid siltation. Thus, Oiniadai’s northern harbour, clearly documented by the archaeological remains of the shipsheds, simply represents an outstanding example of a history for intense and changind interactions between mankind and the sea.