Palaeo-environmental Study Area P16 East Bank of the Estuary, west coast,

PALAEO-ENVIRONMENTAL STUDY AREA P16

EAST BANK OF THE GIRONDE ESTUARY, WEST COAST, FRANCE

Plate P16 East bank of the Gironde Estuary, France. General view of the Saint- Ciers marsh

1. INTRODUCTION

Nb. The introduction, discussion and conclusions for Palaeo-environmental Study Area P16 (East Bank of the Gironde) and included in Study Area P15 (West Bank of the Gironde), sections 1, 2, 3 and 7 to 14.

The following report describes the investigations and interpretation of evidence gathered on the nature, scale and pace of coastal change on the East Bank of the Gironde Estuary, France.

2. KNOWN HISTORICAL, ARCHAEOLOGICAL AND PALAEO-ENVIRONMENTAL SETTING OF THE COASTAL COMMUNITY

The palaeoenvironmental and archaeological studies of the marshes of the East Bank of the Gironde Estuary can be used for a better understanding of how the coast evolved in response to the changing conditions through the Holocene. The quite quick sedimentation rate, and the sensitivity of this environment to coastal changes give a great interest of the sedimentary archives. These elements offer an opportunity to reconstruct the evolution of the landscapes and of the infilling of the estuarine marshes. 2.1 Sedimentological and palaeo-environmental data concerning the study area

Two different morphological types of marshes can be distinguished along the East Bank of the Gironde Estuary: an "opened" marsh (the "Saint Ciers-sur-Gironde" marsh) and a "digitated" marsh (the "Monards" marsh).

1 Palaeo-environmental Study Area P16 East Bank of the Gironde Estuary, west coast, France

The Saint Ciers-sur-Gironde marsh

The Saint Ciers-sur-Gironde marsh forms the largest marsh area on the east bank of the Gironde estuary (Figure P16.1). It is an "opened" marsh formed by the divagation of the main channel of the estuary. This marsh forms 40 000 ha of reclaimed coastal marshes extending 30 kilometers from Port-Maubert in the north to in the south.

It is bounded on its western estuarine margin by sea defences first constructed in the 17th century and on its eastern landward margin by low hills which reach a maxium elevation of 75 metres NGF. The modern marsh surface lies between +1,0 and +2,5 meters NGF. Current landuse is a mixture of pasture and arable. Satlmarshes seawards of the present sea-defences lies between 1 to 3 meters NGF.

The marsh is lined with Cretaceous limestones to the north and calcareous Eocene sands and clay to the southeast. The Holocene sequence of the marsh, composed of sands, clay and peat, onlaps Pleistocene fluvial sands and gravels accumulated on the Tertiary carbonate substratum during the Weichselian sea-level low-stand.

The rate of subsidence and the compaction of the marsh is at present unknown but is expected to be less than 0,2-0,5 millimeters per years.

The Monards marsh

The Monards Marsh is located on the northeastern bank of the Gironde Estuary, in the proximity of the mouth (Figure P16.1). It is a ‘’digitated ‘’ marsh formed by the infilling of small tributary valleys (west shore of the estuary).

On a geomorphological point of view, the reclaimed Monards marsh (Figure P16.6) is located in an ancient bay dug in the cretaceous calcareous substrate (Campanian). It shows a flat top with a mean height of 3 m NGF (Figure P16.6), covered with reed beds and grasslands and drained by natural or made-up brooks. On the estuarine side, the actual active marsh is lined by a small cliff (80 cm to 1 m in height) separates the marsh and the active schorre (mean height 2 m NGF). Further towards the estuary, a wide slikke is crossed by the made-up Monards Channel. A general profile is presented on Figure P16.7.

2.2 Archaeological data concerning the study area (Appendix P16.1, Figures P16.8 and P16.9)

25 archaeological sites were listed in our archaeological inventory along the East Bank of the Gironde Estuary: 17 are localised in the Saint Ciers-sur-Gironde marsh, 3 in the Monards marsh and 5 in other marshes or in border of the estuary.

The Saint Ciers sur Gironde marsh

Numerous archaeological sites were found in the Saint Ciers marsh, and 17 of them, particularly representative were chosen for the archaeological inventory.

Few neolithic sites are known within the marsh: "Terrier Ricard" (actually a multi-period site spanning Neolithic-Iron Age periods) and "Fréneau" are the most representative of this period.

Archaeological evidences from proto-historic periods onwards shows that the marsh was an important area for local industry and trade/sea-transport. More than 30 Iron Age salt production sites are found in the Saint Ciers marsh.

In addition to the occupied Iron Age positions a number of Gallo-Roman sites are known. Evidence of occupation became less clear after the 3rd Century A.D. There then follows a

2 Palaeo-environmental Study Area P16 East Bank of the Gironde Estuary, west coast, France

complete absence of archaeological evidence until the 11th Century A.D.

The Monards marsh

There is only few archaeological evidence within and around the Monards marsh, except for the Gallo-Roman period.

Since the "Haut Empire", the development of an antic town (Le Fâ) occured in Barzan near the Monards. The Monards marsh could be one of the possible location of the antic harbour of the Fâ (like the site of Chant-Dorat) with the discovery of: structures of jetties (?) with ceramic sherds, amphorae etc.

At this time, agricultural proprieties appear on the hills surrounding the marsh (villa of Langlade).

3. PALAEO-ENVIRONMENTAL AND ARCHAEOLOGICAL POTENTIAL: INVESTIGATIONS IN THE SAINT CIERS-SUR-GIRONDE MARSH (Plate P16.1)

This study allowed to understand the development of the Saint Ciers-sur-Gironde marsh during the late Holocene. This marsh forms the largest expanse of coastal wetlands on the East Bank of the Gironde. A geoarcheological approach was used incorporating borehole survey, sedimentological and diatom analysis, radiocarbon dating, archaeological, documentary and cartographic evidences.

3.1 Methods

Lithostratigraphic interpretation is based upon 12 boreholes which form two transects running across the marsh perpendicular to the sea-defences. Seven boreholes were made using a mechanical percussion auger which provided uncontaminated samples for laboratory sedimentological and diatom analysis. In addition, a manual gouge auger was used to produce five additional lithostratigraphic records at selected intermediate points between percussion boreholes. The upper part of the Holocene sequence was targeted for sampling as this relates to sediment accumulation since 6000 B.P. after sea-level rise slowed (Allen and Posamentier, 1993; Bertero, 1993) and also to the period during which the marshes were exploited by human activity. Where possible, boreholes (percussion and gouge augers) were placed directly over archaeological sites in order that archaeological horizons could be placed in their lithostratigraphic context. The deepest boreholes were drilled using the percussion auger and a maximum depth of 8m below the present marsh surface was obtained.

These lithostratigraphic records were supplemented by eight additional borehole logs made by the Bureau de Recherches Géologiques et Minières (B.R.G.M - French Geological Survey) which allows more widespread lithostratigraphic correlation across the marsh along additional transects. The borehole records provided by the B.R.G.M survey have been summarised by Bertero (1993) which provides the first lithostratigraphic model for the St. Ciers-sur-Gironde marsh. The spatial distribution of the boreholes is shown in Figure P16.10 and Table 1.

Continuous samples made by percussion auger from 8 boreholes were analysed in the laboratories of the Departement de Géologie et Océanographie, Université I. The sediments were first described in detail incorporating colour, compositional and structural properties. Sub-samples were taken every 10cm for particle size analysis and diatom analysis. Particle size data was produced using a Malvern 3600E laser particle size analyser. All organic material was removed using hydrogen peroxide prior to analysis. Diatom slides were prepared using standard organic digestion and centrifugation methods described by Battarbee (1986) and the resulting suspensions mounted using Naphrax. Counts of 200 valves per slide were made using phase contrast microscopy at x 400 magnification. Diatom species were identified with reference to Germain (1981) and assemblages interpreted following the Vos and De Wolf (1993) method. This method draws upon both lithostratigraphic and diatom evidence to derive

3 Palaeo-environmental Study Area P16 East Bank of the Gironde Estuary, west coast, France

an interpretation of sedimentary environment and therefore provides and excellent link between the different approaches employed in this study.

The current known archaeological site distribution for the St. Ciers-sur-Gironde marsh has also been integrated with stratigraphic data to reconstruct the spatial extent of marsh development during specific time periods. The site listing from the St Ciers-sur-Gironde, Braud-et-Saint-Louis and Anglade regions has been utilised (Coquillas, 2000). A total of 96 sites are known including those both on the marsh and on its landward margins. From this number, all sites represented by only a single find, hoards and dolmens have been excluded from analysis along with all sites located at datums > 15m NGF. In this way only long-term occupation sites are considered which are more reliable for the aims of this study. The number of sites remaining after screening is also sufficient for marsh occupation to be considered in the wider context of the total settlement distribution of the region. Cartographic evidence (Migniot 1971) has been utilised to determine the nature of marsh development over the past 300 years.

A minimum occupiable datum (MOD) has been defined for successive archaeological periods to provide a preliminary indication of the maximum upper limit for the position of past sea-levels (Figure P16.11). The definition of the MOD is based upon the premise that a permanently occupied site would not have been established upon a landsurface prone to regular flooding e.g. due to the daily tidal cycle or through a fluctuating ground water table. The relationship between the MOD and contemporary tidal levels will be dependent upon the type of site from which it comes as for functional reasons sites are established either on dry land, or partly or completely submerged by the sea (Flemming, 1979-1980). Where sites are present in a distribution which are known to have had a strong maritime association, a closer association can be made between the MOD and tidal levels. All of the archaeological sites in the St. Ciers region are of a dry land setting but many demand a coastal location. Mean high water spring tide (MHWST) is therefore used as the tidal reference point for former sea-levels throughout.

Due to the general unreliability of archaeological sites as sea-level indicators (Akeroyd, 1972), the site distribution is used largely to infer patterns of vertical/horizontal marsh development. However, the nature of human occupation on the St. Ciers marsh does have significant implications for sea-level change in the Gironde Estuary. Where evidence for a fluctuation in relative sea-level rise does occur, greater emphasis is placed upon the timing of the event rather than the inferred height sea-level had attained. Thus, MOD’s are provided merely as a guide at this stage.

3.2 The Holocene sequence

The fourteen boreholes taken across the marsh show a sequence of massive clay-silts with intercalated peat units overlying sandy muds (Figures P16.12 (a), (b) and (c)). This sequence correlates well with that derived from B.R.G.M boreholes (Bertero, 1993). The Holocene units vary in depth from 3m at the landward margin to > 20m at the estuarine channel (Bertero 1993). Combining the two sets of boreholes the sequence can be described, from bottom to top, as follows:

- Sand - interbedded sands and sandy silts with occasional gravels. - Lower clay-silt - generally massive with occasional tidal bedding, frequent plant debris and in situ Scrobicularia shells in some boreholes. - Peat - well humified peat bounded by sharp contacts with organic mud intercalations; often containing woody roots. - Upper clay-silt - massive clay-silt overprinted by pedogenesis. Absent in some boreholes. - Upper clay-silt - massive clay-silt overprinted by pedogenesis. Absent in some boreholes.

Borehole evidence and local knowledge of the sequence derived through well digging (R. Goyon pers. comm.) indicates that the principle lithostratigraphic units are laterally extensive at least through the central section of the marsh.

4 Palaeo-environmental Study Area P16 East Bank of the Gironde Estuary, west coast, France

The particle size properties of each stratigraphic unit are shown in Table 2. An overlap in size parameters is shown by fine grained units in all boreholes. Little variation is shown down-core, although grading trends are apparent. The absence of sand units in the upper part of the sequence marks a significant shift in the depositional regime of the marsh. This shift can be attributed to two factors. Firstly, the shift from a transgressive phase to a regressive phase as the rate of sea-level rise slowed (Allen and Postmentier, 1993) and secondly to the decreasing competence of water to flow to transport coarse grained sediments as the infilling of the estuary progressed (Jouanneau and Latouche, 1981).

Given the homogeneity of sedimentary characteristics within the upper and lower clay-silt units, the distribution of plant rooting is one of the few criteria which can be used in stratigraphic differentiation. The modern saltmarsh sediments of St. Christoly (Figure P16.2) on the west bank of the Gironde provide a perfect modern analogue for the massive, rooted clay-silts. The high sediment load of the estuary is presumably the reason for the low organic content of the saltmarsh sediments bordering the Gironde in comparison to those of other estuaries (Mellalieu 1997). The frequent alternation of rooted and unrooted clay-silt beds is suggestive of a dynamic inter-tidal environment showing continuous shifts in the distribution of facies.

Situated in the core HOREST 9608, diatom valves were only found to be present in the lower clay-silt units (Figure P16.13). Preservation is not continuous either down through the sequence or at comparable datums across the marsh. It would appear that all assemblages have been subject to differential dissolution of diatom valves. This is suggested firstly by the low species diversity shown in each assemblage counted (12-20 species) and secondly by the high preservation potential of the species present (as recorded by Denys, 1991).

Benthic and planktonic species across a salinity spectrum ranging from marine to freshwater are present in all assemblages counted. Marine/brackish epipelic species dominated throughout (Figure P16.13), Nitzschia navicularis and Diploneis didyma being particularly abundant. Using the Vos and De Wolf (1993) method, all assemblages are interpreted as those of inter-tidal mudflats and is supported by the fine grained, and generally massive nature of the deposits to which they relate (Evans, 1965; Larsonneur, 1975; Reineck and Singh, 1980). However, given the absence of many genera, particularly Navicularis, it is possible that other inter-tidal depositional environments present in the sequence may be unrecognisable due to the effects of differential dissolution upon less robust indicator species. This is particularly the case for saltmarsh units, the presence of which is suggested by frequent beds of rooted massive clay-silts described above.

Two peat beds are present in the sequence. The first, lower, peat bed is deeply buried and its development appears to have had a restricted and irregular distribution (Figure P16.12b). This unit is only recorded in B.R.G.M. boreholes and there are no 14C assays available with which to date its formation. The second, upper, peat bed is far more extensive and has been recorded throughout the marsh interior up to the continental margin. In the interior of the marsh the upper peat bed varies between 1-3 m in thickness and occurs across a datum range of +2.0 m to -3.5 m NGF. The peat bed thins considerably towards its western estuarine margin. In some cases the reduced thickness of the unit may also be due to erosion of the upper contact. The upper clay-silt unit overlies the upper peat in a number of boreholes (Figure P16.12 a, b and c). This indicates that in marginal areas peat development was arrested by tidal inundation whereas in more sheltered/protected areas peat continued to accumulate until the marsh was reclaimed.

Seven 14C age estimates have been obtained from the upper peat unit (Table 3, Figure P16.12 a, b and c). The base of the peat at its landward margin at Anglade (HO 9611/12, HO 9613) is dated consistently at 5630 ± 70 B.P. (4605 - 4340 cal. B.C.) Two radiocarbon dates of 5030 ± 70 B.P. (3975 - 3665 cal. B.C.; HO 9607) and 4120 ± 40 B.P. (2665 - 2475 cal. B.C.; HO 9608) from the centre of the marsh indicate that the peat unit prograded towards the estuary over time. An age estimate of 2670 ± 70 B.P. (930 - 775 cal. B.C.; HO 9613) is, at present, the only assay available for the transgressive contact between peat and the overlying upper clay-silt unit. This age estimate is derived from the continental margin of the more sheltered La Vergne marsh, making it possible that the transgressive event commenced earlier in more open areas

5 Palaeo-environmental Study Area P16 East Bank of the Gironde Estuary, west coast, France

closer to the estuarine channel.

The environmental conditions under which the peat units accumulated are currently unknown due to the absence of diatom data. This is disappointing in that the extent of peat development marks the major difference between lithostratigraphic evidence from the St. Ciers marsh and the Allen and Posamentier (1993) model for estuarine infilling (peat lenses only being noted as occurring in the previous model). Freshwater conditions at the start of peat development are indicated by the presence of wood fragments at the base of the unit. This evidence is supported by the presence of a submerged forest which is known to occur in the region of l'Ile Sèche at approximately 0m NGF which correlates with peat datums. The migration of channels across the marsh surface would seem to be a likely cause of erosion due to the irregular lateral distribution of the erosion surface. Minerogenic horizons within the peat unit suggest both the proximity of channels and a surface prone to flooding.

From the present lithostratigraphic data it would appear that a relative sea-level change of around 4.5m has occurred since the onset of peat formation around 5600 B.P. However, all stratigraphic contacts would have been subject to datum changes since deposition through compaction, subsidence and land shrinkage. Therefore the actual relative sea-level rise is likely to be less than that apparent. Clay-silt units and peat beds will have undergone the greatest compaction although the actual amount is difficult to establish due to the combined effects of variable lithology and variable depth of the succession (Greensmith and Tucker, 1986). However, where a peat overlies a basal sand unit, as occurs in boreholes HO9612 and HO9613 (Figure P16.12b), the lower contact is expected to be closer to its original datum. The effects of land shrinkage due to oxidation and desiccation are obvious on the present marsh surface. Topographic lows, forming a surface relief of 2m, correlate with the distribution of peat at the surface. The degree of tectonic subsidence in the Gironde region is currently unknown but is thought to be lower than subsidence rates for the area immediately to the south of the Gironde around Bay which vary between 0.0-0.7 mm/yr (Klingebiel and Gayet, 1995). These figures cannot be expected to have remained constant over the course of the Holocene as contributions from sediment loading and hydro-isostacy will have varied (Lambeck, 1997). A constant rate of 0.7mm/yr since the onset of peat formation is sufficient to account for almost all of the apparent relative sea-level rise observed using lithostratigraphic data (3.92m). Therefore considerable difference between recorded datum and datum at the time of deposition is anticipated but cannot be quantified at present. These processes will also effect the values for MOD quoted below.

3.3 Archaeology of the Saint Ciers-sur-Gironde marsh (Figure P16.8 and P16.14)

From the original archaeological distribution, a list of 63 sites remain after the screening process. Sites covering an archaeological time span from the Neolithic through to the Medieval Period are represented (8000-500 B.P). Iron Age and Gallo-Roman remains predominate but older remains are worthy of comment.

Neolithic and Bronze Age

Sites relating to Neolithic (8000 - 3750 B.P.; 6000 - 1800 cal. B.C.) and succeeding Bronze Age (3750 - 2675 B.P.; 1800 - 725 cal. B.C.) periods are distributed along the landward marsh margins and surrounding uplands, but within the marsh are rare. This is to be expected as chronological horizons relating to the Neolithic and Bronze Age periods in marsh areas will be more deeply buried within the estuary fill and therefore are generally invisible to surface survey. Neolithic sites are more numerous than those of the succeeding Bronze Age in accordance with evidence from the west bank of the estuary. The apparent occupation pattern is surprising as the low density of Bronze Age sites contrasts with archaeological evidence from the British Isles, for example, which shows that Bronze Age peoples actively exploited wetland areas (Wilkinson and Murphy, 1986), and that site numbers often exceed those from other time periods (Wilkinson, 1989).

6 Palaeo-environmental Study Area P16 East Bank of the Gironde Estuary, west coast, France

Iron Age

From the Iron Age (2675 - 2000 B.P.; 725 - 50 B.C.) onwards archaeological evidence shows that the St. Ciers region was an important area for local industry and trade/sea-transport. More than 25 Iron Age salt production sites are known within and around the St. Ciers marsh (Coquillas, 1993; Figure P16.15). The salt industry flourished in the region between 2100 - 2000 B.P. (1st to 2nd Centuries B.C.) and has parallels throughout France and western Europe. Salt production is the only evidence of Iron Age activity in the study area. Identification of such sites is indisputable through the recovery of briquetage fragments, a ceramic fabric type specific to the salt production industry. The greatest concentration of salt production sites occurs along the border of the southern sector of the marsh in the environs of Braud-et-St.-Louis, Anglade and St. Androny (Figure P16.14) although a small number of sites are known on the marsh surface itself.

Four of the cores described in this study were placed directly upon such sites: La Lombatte (HO 9703), Le Cahour (HO 9607), La Moutonne (HO 9710) and Fréneau-Aubeterre (HO 9520). The latter is the best known example in the area. It is located on the estuarine edge of the marsh and has been particularly well exposed on erosion cliffs during recent years. The outcrops reveal two phases of salt production separated by a thin mud layer. Although poorly preserved, all of the material currently found on briquetage sites has been described, along with domestic warehouse indicating a temporary settlement. The date of this occupation is La Tène III (120 - 80 cal. B.C.).

The precise relationship of salt production sites to the inter-tidal zone is often unclear as sites may be found located upon both marsh and dryland surfaces, as is also the case in the St. Ciers region. Current thinking favours a supra-tidal, rather than an inter-tidal, position for site location (e.g. Fawn et al., 1990; Perrichet-Thomas, 1986) and is also assumed here. The presence of sites on the present marsh surface suggests that at least part of the current marsh area may have been above MHWST during the Iron Age period. However, the variability in the location of salt production sites prevents a consistent relationship to contemporary tidal levels being identified. Therefore, datums derived from salt production sites only provide a maximum upper level for the position of contemporary MHWST (Prigent, 1979).

Gallo-Roman period

Whilst there is no evidence for the continuation of the salt industry in the area after 2000 B.P. (50 cal. B.C.) during the succeeding Gallo-Roman period, there is evidence to suggest that Iron Age sites continued to be occupied or were re-occupied. A number of further Gallo-Roman sites are known, many of which consist of settlements/villas and again occur on the marsh margins and further inland (indicating a wider range of activities), but more significantly, also on the outer marsh surface towards the estuary channel (Figure P16.14). This occupation pattern occurs without any indication of the construction of a drainage system or sea-defences during this period as exists, for example, in the Severn Estuary marshes, UK (Allen and Fulford, 1986). The three most important sites are clearly related to warehouse/ceramic industry and trade/sea- transport.

A road-type structure, extending 1.2 km southwestwards on the marsh surface from l'Ile Sèche to La Melonne (Figure P16.14) has been described as early as 1856. It consists of numerous rocks, tiles and warehouse/ceramic debris (over 4,000 m3) spread over the marsh surface with occasional wooden posts for stabilisation. The highly homogeneous material essentially originates from the South Saintonge production sites, located some 50 km to the northeast, and was produced there between 50 and 120 A.D.. The "road" ends abruptly in the vicinity of La Melonne, where it was sampled as a 35 cm thick layer of highly compacted ceramic fragments in core HO 9608. Apparently, the main function of this rudimentary structure was to board and export the ceramic production of the South Saintonge, and it was built with the artifacts which had been broken during the land transportation.

The second warehouse site is located in La Patte d’Oie near Anglade (Figure P16.14) which

7 Palaeo-environmental Study Area P16 East Bank of the Gironde Estuary, west coast, France

bears thick deposits of tiles and ceramics with occasional domestic waste and human bones, as well as a a large paved surface (20 m wide and over 25 m length) built on a complex piled wooden structure located close to a creek, which may have been used as a jetty for the exportation of the South Saintonge ceramic production towards the estuary.

The third site is located in Freneau-Aubeterre (Figure P16.14) where large amounts of tiles and intact ceramic were recovered during the last thirty years. Some of these were still perfectly lined up in a possible wreck. Most of the site has presently been destroyed by the erosion of the marsh edge, but a circulation surface is still observed 30 cm above the Iron age briquetage layers. This place was apparently the last stocking/transit point of the South Saintonge ceramic production before definitive exportation towards the estuary.

The function of these sites as distribution centre implies occupation of a landsurface above MHWST located at, or at least very close to, a tidal creek system to facilitate the loading of goods onto vessels for transport. This interpretation makes warehouse sites very attractive as supportive indicators of sea-level position. However, their precision as such in the macro-tidal environment is low as their relationship to water level at various stages of the tidal cycle will change markedly and a sufficient water depth to allow constant access by boat cannot be assumed. The datum of the jetty surfac is therefore used as the maximum upper level of MHWST as with salt production sites.

Middle Ages

Evidence of occupation becomes less clear after 1750 B.P. (3rd Century A.D). Only two mainland sites show occupation until the end of the Gallo-Roman period (1474 B.P., 476 A.D. Bastisse and Picontin, 1977). There then follows a complete absence of archaeological evidence until 950 B.P. (11th Century A.D). Documentary evidence from the Abbaye de St. Etienne de Baignes testifies the presence of a farm at Vitrezay which it acquired around 950 B.P. (end of the 10th Century A.D.) indicating that some form of agriculture or stock rearing was possible on certain parts of the marsh at this time.

Implications for sea-level reconstruction

Given the nature of archaeological remains known on the St. Ciers marsh, the site distribution, particularly of the Iron Age and Gallo-Roman periods, has important implications for marsh development described below. There is no trend of decreasing site age across the marsh. This suggests the development of a broad marsh area by at least the late La Tène period of the Iron Age (2200 - 2000 B.P., 120 - 50 cal. B.C.). A complex marsh topography is also indicated by archaeological datums from which no common occupation level can be identified.

The datums of archaeological sites in the St. Ciers region used to derive figures for MOD are shown in Figure P16.11. MOD’s for the Iron Age and Gallo-Roman period are expected to be more accurate due to the maritime association of sites described above. The Iron Age MOD’s is +0.4 m NGF and is derived from the salt production site of Fréneau-Aubeterre (Figure P16.14). As mentioned above, the archaeological horizons are separated by clay-silt beds representing phases of inundation by tidal waters which suggests that contemporary MHWST could have been located as close as 50 cm below the MOD. The Gallo-Roman MOD is +0.6/+1.19 m NGF but the relationship between this level and contemporary MHWST cannot be accurately determined due to the nature of the lowest known site (L’Ile Sèche). Gallo-Roman horizons are known below this level but are considered to represent dump units rather than actual occupation levels. The metalled road which crosses the marsh between l’Ile Sèche and La Melonne provides supporting evidence for a supra-tidal marsh during the Gallo-Roman period. Whilst roads have absolutely no relationship to contemporary sea-level, its presence does suggest that at least the landward marsh area was not prone to flooding at the time of its construction (1950 - 1850 B.P, 1st Century A.D.).

3.4 Chronology of marsh development

8 Palaeo-environmental Study Area P16 East Bank of the Gironde Estuary, west coast, France

An interpretation of marsh development is described below using the current available lithostratigraphic and archaeological evidence. Documentary and cartographic sources have also been incorporated to reconstruct the final phases of development during the post-Medieval Period.

Pre-6000 B.P. (Figure P16.15)

The present study adds no additional evidence to the model of morphosedimentary change in the estuary of Allen and Posamentier (1993) during this time period. Only the uppermost units of the transgressive systems tract were sampled and often in marsh marginal situations where they occur close to the present ground surface. It is impossible to derive any palaeogeographical inferences from the data sets due to the absence of the early Holocene sediments in many boreholes and a corresponding lack of archaeological remains of equal age known within the marsh.

6000 - 2200 B.P. (Figure P16.16)

Sea-level attained a level close to present at/around 6000 B.P. (Pontee et al., 1998). Fluvial sediments began to infill the now delimited estuarine volume causing a negative sea-level tendency. A phase of prograding tidal flats and saltmarshes ensued corresponding to the highstand systems tract of Allen and Posamentier (1993) and the upper Holocene units of Bertero (1993). A restricted phase of peat development may have occurred at or before this date as indicated by deeply buried thin peat units (Figure P16.12b). A final extensive phase of peat formation then occurred over clay-silt sediments across the marsh. This period of peat formation was prolonged, spanning minimally 5600 - 2600 B.P.. From the available assays (Table 3) it appears that peat accumulation commenced at the continental marsh margins and gradually extended towards the estuary, possibly keeping pace with the progradation of mudflats and saltmarshes, and causing a negative sea-level tendency. At present there is no palaeoecological data to indicate what relative sea-level conditions promoted peat development. However, the peat does appear to have been prone to flooding and accumulated contemporarily with woodland extension along the landward margins of the marsh.

2200-1800 B.P. (Figure P16.17)

Peat development was terminated by tidal inundation and clay-silt deposition during a positive sea-level tendency at some point prior to 2200 B.P. This date is derived from a 14C assay of 2670 ± 70 B.P. from the upper contact of the peat bed in the south of the marsh near Anglade and from the Iron Age horizons associated with the overlying upper clay-silt unit. It remains possible that the timing of this event was earlier along the outer marsh margins. The transgressive phase had a complicated effect on marsh geography as a mosaic of different environments evolved. Lowering of the peat surface through tidal erosion led to the landward extension of mudflat and saltmarsh units particularly in the regions of Vitrezay and Canal St. Georges. In the central sector of the marsh, peat accumulation survived this inundation and persisted until the area was reclaimed.

The presence of Iron Age and Gallo-Roman sites on the marsh, without any evidence for the construction of sea-defences, indicates that at least part of the marsh was situated at or slightly above MHWST during this time period. This would imply that following a period of accretion of inter-tidal environments, a short phase of marsh stabilisation ensued, possibly during a negative sea-level tendency, lasting approximately 150-200 years from the current available relative dating evidence. The distribution of facies would, therefore, have remained essentially similar to that at the end of the preceding time period. However, this interpretation is derived solely from archaeological evidence due to the absence of primary lithostratigraphic and biostratigraphic data.

9 Palaeo-environmental Study Area P16 East Bank of the Gironde Estuary, west coast, France

No palaeosols have been identified in association with any archaeological site to suggest the stability of the marsh surface at this time. Soil formation may not be expected given the relatively short duration of inferred stability and the assumed hydromorphic nature of the ground conditions. If sites were situated on poorly developed soils it is possible that these horizons have been masked from detection using the analytical techniques applied by pedogenesis which has occurred since marsh reclamation; a process which has also destroyed primary lithostratigraphic evidence from the upper clay-silt. A buried landsurface has been recorded in the Anglade - St. Androny sector of the marsh (Blondy, 1987) but the lack of dating evidence does not allow it to be definately attributed to this time frame.

The archaeological sites cannot in any case be considered as being established in a completely dryland setting as the ground water table beneath the marsh at this time would have had to have been high in order to maintain peat growth. Indeed, the presence of two inundation horizons during the Iron Age occupation, and separating the Iron Age and Gallo-Roman occupations, of Fréneau-Aubeterre indicates that this site was situated in a highly marginal location. This is the only site on the marsh which shows any evidence of flooding and the majority of sites are located in more landward, and presumably more protected, locations. The distribution of archaeological sites shows a clear preference for surfaces on the upper clay-silt unit, avoiding areas of continuing peat growth. Areas of minerogenic sedimentation would probably also have been in close association with tidal channels penetrating the marsh which would certainly have been an additional factor in site selection for the foundation of warehouse sites. High level inter-creek areas and creek levees, or possibly supra-tidal coastal reedswamp areas, as defined by Shennan (1986), are possible environments for the human occupation of the marsh during this time frame.

There is no evidence in our lithostratigraphic data set for the existence of a system of barrier islands and a channel/creek between l'Ile Sèche/La Melonne and La Patte d'Oie. This pattern is solely inferred from archaeological evidence, and especially from the distribution of gallo-roman ceramic handling sites within the marsh. From available evidence, the South Saintonge productions were transported on land to l'Ile Sèche, where the road-type structure may have served as a jetty to reach a place near La Melonne where boats could access. The ceramics were then transferred to La Patte d'Oie, probably via a channel/creek running along the landward margin of the marsh, and finally reached the estuary in Freneau-Aubeterre. The existence of other creeks running through the possible system of barrier islands is very likely, especially near Vitrezay which is presently the most important harbour in the marsh; however, the choice of Freneau-Aubeterre as the main transit zone seems to indicate that the other channels were not sufficiently deep to allow boat circulation.

The trend of sea-level change during this period is difficult to elucidate. The difference between MOD’s for the Iron Age and Gallo-Roman periods (+0.4m and +0.6m NGF respectively) does not reliably indicate a relative sea-level rise. Significant overlap occurs between the datum distribution of both periods (Figure P16.11) and both MOD’s merely represent the maximum upper limit for the position of contemporary MHWST rather than a fixed datum. Evidence of flooding at Fréneau-Aubeterre cannot be taken to indicate relative sea-level rise as the site is situated in a very exposed position directly on the banks of the estuary and shows persistent re- occupation. Further, the abandonment of marsh sites following 1900 B.P. after only a short period of occupation (around 60 years) cannot be attributed to rising relative sea-level. Archaeological sites across the region, including those located on dryland in the continental interior, were also abandoned at the same time. This indicates that a cultural factor or some other environmental variable was affecting human activity in the region and resulting in depopulation. A static sea-level or negative sea-level tendency is inferred only through the absence of definitive evidence for a positive tendency. The duration of these assumed conditions is difficult to determine as there is little archaeological dating evidence after 1750 B.P. However, the MOD’s do show that there has been a relative sea-level change of around 2.0m since the latter date (using MHWST as the tidal reference point).

1800 B.P. to Present (Figure P16.18)

10 Palaeo-environmental Study Area P16 East Bank of the Gironde Estuary, west coast, France

The burial of archaeological horizons by overlying clay-silt units presumably reflects continued inter-tidal mudflat and saltmarsh development from 1750 B.P. onwards or possibly as a result of a continuing positive sea-level tendency since around 2500 B.P. The datum of the archaeological horizon at La Melonne (-0.2m NGF) shows that up to 2m of stratigraphy has accumulated since this time although comparison of archaeological datums shows that this rate of accumulation is not continuous across the marsh e.g. the Gallo-Roman horizon at Ile Sèche is buried beneath 0.7m of clay-silt; at Le Cahour and La Lombatte archaeological material occurs at the present ground surface. Infilling of a complex marsh topography to a common datum is therefore inferred.

Evidence from 1000 B.P. to 350 B.P. (10th to 16th Centuries A.D.) is complex. Documentary evidence recording the presence of a farmstead at Vitrezay around 950 B.P. suggests that at least part of the marsh surface was stable during the Medieval Period. A negative sea-level tendency may be inferred although this interpretation is based upon a single piece of evidence and should be treated with caution, particularly as the activities on the farmstead are unknown. It is possible that the settlement was used purely for summer grazing on the saltmarsh surface rather than year round stock rearing and/or arable farming which would require land less prone to tidal inundation. All sites of similar age are found above the present marsh surface and therefore cannot be used to support a negative tendency. In the absence of archaeological and litho-/biostratigraphic data, the available documentary evidence is not sufficient to resolve whether a positive sea-level tendency post-1800 B.P. was interupted by a negative tendency around 950 B.P.

Documentary evidence covering the period from 750 B.P. to 350 B.P. (13th to 16th Centuries A.D.) suggests that a positive sea-level tendency ensued. At this time the marsh is described as being covered by reed beds and fens inaccessible for the greater part of the year due to tidal inundation (Blondy, 1987). Documents relating to 550 B.P. to 450 B.P. (15th Century A.D.) also describe deep tidal channels extending from the estuary to ports at Braud-et-St. Louis and Anglade, showing that the communication at Freneau-Aubeterre was still sufficiently deep to

allow navigation. A palaeochannel system is still visible today on the marsh surface at St. Thomas de Conac (Figure P16.1) (Blondy, 1987). Extensive tidal inundation and continued minerogenic deposition is therefore indicated for this time period.

Cartographic evidence provides little additional information in the developement of the marsh. The earliest available maps date to the 16th Century (Migniot, 1971) and are not of sufficient scale, and provide insufficient anchor points, to reliably compare the shape and form of the relevant sections of the coastline. The next major phase in the history of the marsh was the reclamation of the area between Blaye and St. Bonnet in 1654 A.D., and of the remaining northern marsh sector a little later (Blondy, 1987). A map of the reclaimed marshes from 1677 A.D. clearly shows the main channels of the drainage system employed which can still be traced today (Migniot, 1971). There have been no subsequent marsh innings since this date. Inter-tidal mudflats and saltmarshes continue to accumulate on the seaward side of the sea defences to a datum slightly higher than the reclaimed marsh surface.

3.5 Discussion

The St. Ciers marsh, widely opened on the estuary, differs from many other marsh systems in the Gironde in terms of its size.. In contrast, to the narrow embayments formed between the well developed Pleistocene fluvial terraces and tributary systems of the Medoc peninsula which can be expected to be infilled relatively rapidly promoting the advanced formation of peat units (Diot and Tastet, 1995; McMillen, 1996; Pontee et al., 1997), the St. Ciers marsh system forms a much deeper and larger embayment. This bay would have been open and susceptible to estuarine influences for a longer period of time and thus present a more complex Holocene succession. Differences in sea-level tendencies are therefore to be expected (Long, 1992) and points of comparison are at this stage limited.

11 Palaeo-environmental Study Area P16 East Bank of the Gironde Estuary, west coast, France

Peat formation post-6000 B.P. is the only estuary wide event which can be recognised on both sides of the estuary thus far (no peat units are known in the estuary with which to correlate the older more deeply buried peats of the St. Ciers marsh). West bank peat beds occur at similar datums to the primary upper peat unit of the St. Ciers marsh (-1m to +2m NGF) and age estimates indicate that peat formation commenced slightly earlier around 6000 B.P. (Pontee et al., 1998). The conditions in the estuary at this time were optimum for the formation of peat. Relative sea-level rise had slowed but not actually stopped maintaining high water tables in marginal zones. Sedimentation rate exceeded the rate of sea-level rise allowing the rapid accretion and progradation of inter-tidal environments behind which would have followed peat accumulation in backwaters and supratidal areas where minerogenic sediment delivery was restricted.

Peat accumulation in the St. Ciers marsh may have been terminated through either autocompaction and/or an increasing rate of relative sea-level rise. Supporting evidence for this rise in relative sea-level, which commenced around 2600 B.P., can be found in the form of various local scale events recorded at different points in the estuary. At Monards Marsh, Barzan, on the north east bank of the estuary a sea-level index point for a positive sea-level tendency is recorded at a comparable datum and dated to 2840 ± 70 B.P. (Massé et al., submitted). On the west bank of the estuary, the timing of this event correlates to the formation of a chenier ridge known as the Cordon de Richard. Two radiocarbon age estimations have been obtained from the chenier of 2975 ± 120 B.P.(corr) and 2021 ± 162 B.P(corr). although it must be stated that rising relative sea-level is not the only possible process leading to its formation (Pontee et al, 1998). Following the formation of this ridge sediment supply was still sufficiently high to allow marsh progradation to continue (Pontee et al, 1998) as inferred for the same time period in the St. Ciers marsh up until its reclamation.

3.6 Conclusions

A geoarchaeological approach has been applied in the examination of the development of the St. Ciers-sur-Gironde marsh, the first such examination of the east Gironde marshes. The model has been produced using lithostratigraphic borehole records in combination with diatom, archaeological, documentary and cartographic sources. The combination of data sets has proved advantageous, especially over timescales covering the past 2000 years, where litho- and biostratigraphic evidence is lacking.

The data available at present suggests that by around 5000 B.P. a broad area of saltmarshes and inter-tidal mudflats had accumulated to around -2.5m NGF in the St. Ciers region. Upon this surface a peat unit developed. The peat surface was subsequently modified by erosion, presumably through channel migration. Clay-silt deposition later continued and upon this surface human occupation occurred during the Iron Age to Gallo-Roman periods (2100-1750 B.P.), possibly due to a reduced rate of sea-level rise. A maximum height for MHWST during the Gallo-Roman period has been estimated from archaeological datums at +0.6m NGF. Clay- silt deposition recommenced (if not previously interrupted) in the region during the 3rd Century A.D. (1750 B.P.) at the earliest and continued until the reclamation of the marsh in the 17th Century A.D. (200 B.P.) by which time the elevation of the marshes had attained their present datum. All inferred datums are subject to error through a combination of compaction and subsidence, the effects of both being currently unknown in this area. Further modifications to this model may be anticipated in future as new borehole and archaeological evidence comes to light.

This study develops the previous model of estuarine infilling and sea-level change (Allen and Posamentier, 1993). The identification of a thick and laterally extensive peat unit between +2.0m to -3.5m NGF introduces a significant additional facies type to the Holocene sequence. A gradual and continual rise in relative sea-level is described. There is insufficient resolution in

12 Palaeo-environmental Study Area P16 East Bank of the Gironde Estuary, west coast, France

the data set at present to identify any fluctuations with certainty although two or possibly three points in the combined Holocene record may represent deviations about the overall trend which were not taken into account in the previous model. The first is characterised by the upper peat unit, dated to 5600 to 2600 B.P., which represents a negative tendency during a period when sediment accumulation rate exceeded the rate of relative sea-level rise. The second occurs during the Gallo-Roman period (2000-1750 B.P.) during which time human occupation of the marsh surface suggests, at best, a slowing in the rate of sea-level rise. A third potential negative tendency may have occurred during the Medieval period around 950 B.P., although evidence for this is almost completely lacking. Pollen analysis may help elucidate these palaeoenvironmental changes in the marsh in future to a greater degree than the diatom evidence presented here.

4. PALAEO-ENVIRONMENTAL AND ARCHAEOLOGICAL POTENTIAL: INVESTIGATIONS IN THE MONARDS MARSH (Plate P16a)

4.1 Sedimentological study of The "Monards" marsh

We studied the evolution of sedimentary environments and processes, sea-level and climate during the last 6,000 years, in the sediment of the "The Monards" estuarine marsh (in a macrotidal context).

Data and methods

Four cores (Table 4, Figure P16.6) were collected with a vibration corer, and three additional drillings were made with an auger. All of these failed to reach the cretaceous substrate.

The cores were opened, described and photographed. Samples were taken at regular 10 cm intervals for laboratory analyses (water and carbonate contents, wet sieving and grain size measurements). Six samples of plant or shell debris were selected for AMS 14C datation. Radiographs were realised on 1 cm thick slabs in cores HOREST 9417 and 9418 for analyzing the sediment structures.

Benthic foraminifera were determined under optical and SE microscopes and counted in a constant volume (20 cm3) of > 100 µm sieving residue in 112 samples in cores 9417 and 9418 and 8 samples in auger T3. The mineralogical composition of the > 100 µm residue was semi- quantitatively estimated in the same way.

Grain-size measurements were obtained with a 2600E Malvern laser-sizer.

Results

Lithology

Four main sediment facies were distinguished from bottom to top:

Grey laminated silty-sandy mud

This first facies is observed below 5 m depth in the central part of the marsh (cores 9417 and 9418, Figure P16.19). It becomes more sandy towards the base. Escape burrows, microfaults and slumpings occur in some levels. The deposits are rich in micas and silt aggregates, and show some calcareous debris (sponges) derived from the surrounding cretaceous substrate. Plant debris, wood fragments and vivianite are present. Carbonate contents are slightly higher than in the rest of the deposits. This laminated mud is missing in the western area of the marsh (core 9502, Figure P16.20), where it is replaced by laminations of plant debris interbedded in a dark grey homogeneous silty mud.

Homogeneous dark grey silty mud

The thickness of this second facies is over 2 m. It is rich in organic matter, fecal pellets,

13 Palaeo-environmental Study Area P16 East Bank of the Gironde Estuary, west coast, France

roots and plant debris more or less epigenised by iron monosulfides and pyrite. Some vivianite appears as black, rounded or mamillate micro-concretions, 1 or 2 mm in diameter. These concretions are composed of tetrahedral lamellar microcrystals. The formation of vivianite usually indicates a rapid burying of organic matter (Pye, 1981). Rounded eolian quartz grains are frequent. Calcareous sponge debris are absent. The decrease in carbonate content may be due to a low pH of the subsurface sediment resulting from the temporary oxidization of organic matter. This phenomenon is common in salt marshes (Zwolsman et al., 1993). In this facies, organic debris are generally pyritized, which determines strongly reducing conditions. Fecal pellets are extremely abundant. Foraminifera, diatoms and saline water ostracoda occur in some levels.

Compact silty mud

The thickness of this third facies varies between 1.7 and 1.8 m. It consists of a brown to grey plastic and compact mud with a low water content. It contains numerous roots and some reed debris forming brownish or blackish spots. It is generally homogeneous, and pyritized elements begin to show traces of oxidization. This process results in a partial decarbonatation of the sediment. As in other european marshes (Verger, 1968), grey- green, irregularly shaped calcareous concretions are frequent. They are 1 to 4 cm in size. Some fecal pellets and eolian quartz grains are present. Calcareous sponge debris reworked from the substrate reappear.

Cultivated clayey soil

This facies appears as a 13 to 17 cm thick surficial layer. It is rich in organic matter, roots and plant debris, and contains minor amounts of fecal pellets, eolian quartz grains and sponge debris. Grain-size distribution

Numerous authors addressed the grain-size distribution of intertidal sediments. Muddy and sandy tidal flats can be divided in three zones: upper flat, middle flat and lower flat (Klein 1985; Frey et al., 1989). The surface sediment becomes coarser from the shoreline towards the boundary between intertidal and infratidal zones (Davis, 1983; Frey et al., 1989; Alexander et al., 1991). On a sandflat, the following distribution is described: (1) homogeneous mud with some sand interbedded on the upper flat; (2) interbedded sands and muds on the middle flat; (3) dominant sands on the lower flat (Reineck and Singh, 1980). Tendencies are far less clearly identified on mudflats. Alexander et al. (1991) describe an example of intertidal muddy sedimentation from southwestern Korea, where spatial trends in grain size are similar to those found on sandflats. Median grain increases from 8.8 F (2 µm) on the upper flat to 4.6 F (40 µm) on the lower flat, and then decreases to 5 F (30 µm) in the adjacent infratidal zone. Sorting is generally poor, with maximum values on the middle flat where the silt percentage is higher.

Sediments from the Monards Marsh are strongly homogeneous. They are composed of very fine sands (1.6 to 30 %), silts (46.1 to 74.9 %) and clay (19.1 to 46.8 %). The mean diameter ranges from 7.3 to 27.3 µm, and sorting is poor (standard deviation between 1.25 and 1.85). The down-core evolution of the median grain shows the existence of two distinct units in cores 9417 and 9418 (fig. 4). The lower unit is the laminated mud, where the median is higher and variable. The upper unit gathers the three other facies, showing a lower and more constant median grain, except in some levels where calcareous concretions occur. This evolution indicates a more energetic depositional environment for the laminated mud. Fine layers result from the settling of suspensions during high water slack, whereas coarse layers result from a gradual discharge from ebb and flow currents on the intertidal flat.

Sediment structures

In the intertidal zone, sediment structures primarily depend on the type of sediment deposited. On a sand dominated flat, homogeneous muds with interbedded sand layers are found on the upper flat, interbedded muds and sands on the middle flat, and cross-bedded sands on the lower flat (Reineck and Singh, 1980). On the middle flat, when the amount of mud decreases,

14 Palaeo-environmental Study Area P16 East Bank of the Gironde Estuary, west coast, France the bedding evolves from lenticular or undulating type towards flaser-type. In contrast, these tendencies are not clear on mudflats, because the low abundance of silts hinders the readability of sediment structures (Alexander et al., 1991). On a mudflat in Korea, Frey et al. (1987) observe: (1) sandy muds with intensely bioturbated undulated bedding on the upper flat, (2) sandy-clayey silts with predominant undulated bedding and occasional parallel laminations and ridges on the middle flat, and (3) sandy silts with laminations, ridges and intense bioturbation on the lower flat. On mudflats, the primary structures are better preserved in the lower intertidal and subtidal zones than in the upper intertidal and supratidal zones. Typical structures are parallel or slightly undulating millimetric to submillimetric laminations, becoming thicker (millimetric to centimetric) and commonly cross-bedded in the subtidal zone (Alexander et al., 1991). According to Bouma (1963), marsh deposits are finer and less sorted than other intertidal sediments. Sediment structures are characterized by the extreme abundance of roots and nodulous or irregular laminations (Reineck and Singh, 1980). Tooley (1992) also described clayey or sandy silts with horizontal laminations and iron stains around roots. In the inner part of the Gironde estuary, the slikke consists of muds showing very regular laminations, locally interrupted by sandy lenses or layers. The schorre consists of clay with plant debris, and primary bedding is erased by the roots of aquatic plants. In the outer estuary, shells and shell fragments occur (especially Scrobicularia), along with an intense bioturbation and some sandy lenses (Allen, 1972). The primary sediment structures of the marsh deposits were described in the intertidal zone near Mortagne (Frégard, 1994). The shore is characterized by a reworked sandy mud showing no specific structures but abundant roots. The slikke is characterized by (1) a sandy mud with inclined and slightly disturbed laminations or (2) muddy and sandy beds, often deeply reworked by bioturbation processes (especially burrows) and the deformation resulting from slumping (Frégard, 1994).

Grey laminated silty-sandy mud

The term "tidal bedding" describes a parallel bedding due to the alternation of muds and sands (Reineck and Singh, 1980). This typical structure is found in the cores below 480 cm (9418) and 508 cm (9417). It shows a rhythmic alternation of sandy and muddy laminations. They may be parallel, slightly undulating, horizontal, subhorizontal or strongly oblique where slumpings occurred during deposition. The rhythmicity depends on the tidal cycle, and is marked by variations in grain size and thickness of the layers. During a spring/neap cycle, the silty layers are relatively coarser and thicker (0.5 - 2 mm) during springtide, whereas the muddy layers are thicker during neaptide. Each couple of layers corresponds to a tidal cycle. 28 tidal cycles per 14 days may be recorded in the subtidal zone, whereas only 14 remain in the middle of the intertidal zone which is higher in the tidal frame. The number of recorded tidal cycles varies between 14 and 28 in between, and is lower than 14 towards the upper intertidal zone (Tessier, 1993; Shi, 1993). In cores 9417 and 9418, the number of tidal cycles varies between 9 and 15, indicating a high position in the intertidal zone during deposition, and the thickness of springtide/neaptide cycles varies between 0.8 and 2 cm. Dominant bioturbation traces are vertical burrows, sometimes masking the primary structures. Wood fragments are rare, and roots and shell fragments are absent. There are numerous microfaults associated with slumpings.

Homogeneous dark grey silty mud

The deposits are homogeneous, and show numerous roots and plant debris mineralised by iron monosulfides, sometimes forming dispersed aggregates. Some levels display microfaults associated with the slumpings. Other characters appear in the lower part of core 9418: (a) a gritty structure with polygonal cracks interpreted as compaction features (White, 1961) or salinity variations in the muddy deposits (Reineck and Singh, 1980); (b) mineralised microfilaments (probably iron monosulfides) forming a dense network in a silty mud displaying blurred parallel, subhorizontal laminations, and (c) vivianite microconcretions (1 or 2 mm in diameter).

Compact silty mud

15 Palaeo-environmental Study Area P16 East Bank of the Gironde Estuary, west coast, France

No layering is observed. The presence of stains is due to the occurrence of iron-rich concretions and traces of roots. Polygonal or pris matic cracks indicate dessiccation phases. Sparse calcareous concretions of varied size appear in the sediment. Submillimetric parallel laminations appear in the lower part of this facies.

Cultivated clayey soil

The facies is chaotic, without anay organisation, and show numerous roots and plant debris.

Benthic foraminifera

Lowman (1949) showed that foraminifera associations may be used as environment indicators in marshes. Phleger (1954; 1966) and Lankford (1959) identified the major species and associations in the marshes of the Mississippi and Gulf of Mexico. Scott and Medioli (1980) showed that Trochammina inflata and Miliammina fusca frequently occur in the median and upper parts of marshes in numerous worldwide locations. Scott and Leckie (1990) and Williams (1994) examined the distribution of present-day foraminifera in marshes along the atlantic coast of America. An extensive study for the Gironde estuary was made by Pujos (1976).

Foraminifera groups (Figure P16.21, P16.22 and P16.23; also Plates P16b, P16c Na P16d).

26 benthic foraminifera species were identified in cores 9417 and 9418. They were gathered in 6 groups according to their ecological characters.

1. Trochammina inflata (Montagu) is well adapted to low salinity (Haynes, 1973) and characterizes the internal schorre (Scott et al., 1991). 2. Haynesina germanica (Ehrenberg) is one of the most frequent species found in brackish water environments occurring in the muddy mediolittoral and infralittoral parts of the Gironde estuary, especially on the slikkes. 3. Ammonia tepida (Cushman) is adapted to very low salinities, and is widely distributed in the Gironde estuary. It is present mainly in schorre environments and associated channels, but occurs on the slikke as well. 4. The genus Elphidium, with the species E. excavatum (Terquem), E. gerthi (Van Voorthuysen), E. Williamsoni (Haynes n. sp.), E. magellanicum (Heron-Allen and Earland) and E. incertum (Williamson) occurs mainly in the muddy infra- and mediolittoral zones, as well as in vegetated submerged areas. 5. The Asterigerinata mamilla (Willamson) group, including Discorbis nitida (Williamson), Planorbulina mediterranensis d'Orbigny and Quinqueloculina sp., mainly occurs in the circa- and infralittoral zones, where salinity is high. 6. The Cibicides lobatulus (Walker and Jacob) group, including Gavelinopsis praegeri (Heron-Allen and Earland), Fussurina orbignyana (Seguenza), Fussurina laevigata (Reuss), Bulimina marginata d'Orbigny, Blphidium aculeata d'Orbigny, Miliolinella subrotunda (Montagu), Cassidulina crassa d'Orbigny, Cassidulina leavigata d'Orbigny, Lagena apiopleura (Loeblich and Tappan), Lagena striata d'Orbigny, Lagena sp., Bolivina sp. and Trifarina angulosa (Williamson), characterizes the continental shelf.

In this study, groups 5 and 6 indicate a marine influence and were gathered under the "marine group" label.

Some planktonic foraminifera species were observed where the marine influence is

16 Palaeo-environmental Study Area P16 East Bank of the Gironde Estuary, west coast, France

strong. They consist of species living between the surface and 20 m depth in temperate marine waters: Globigerina pachyderma f. dextre, G. bulloides, G. quinqueloba and Hastigerina pelagica.

Foraminifera biofacies

The distribution of the groups allows the definition of several biofacies:

Biofacies A1 is dominated by the "marine group" (groups 6/5 - 2 - 4 or 6/5 - 3 - 2). It shows a high diversity (up to 26 species) and a high number of individuals (up to 211 per cm3), with the occurrence of some planktonic forms. It can be associated with an external slikke environment, close to the subtidal zone of a marine estuary (0.4 to 1.6 m height above spring-tide low-water level).

Biofacies A2 is dominated by H. germanica, with a subordinate "marine group" (groups 2 - 5/6). It can be associated with the transition zone between external and internal slikke (1.6 to 2.8 m height).

Biofacies B is characterized by a weaker marine influence (groups 2 - 3 - 5/6 or 2 - 4 - 3). It can be associated with the upper slikke (2.8 to 4.5 m height). The occurrence of T. inflata indicates the proximity of the schorre.

Biofacies C1 (groups 2 - 3 - 1 or 2 - 1 - 3) shows more or less abundant T. inflata and characterizes the external schorre (4.5 to 5 m height).

Biofacies C2 (groups 1 - 2 - 3 or 1 - 3 - 2) shows dominant T. inflata associated with two other species and corresponds to a well developed internal schorre (5.0 to 5.2 m height).

Biofacies C3 (group 1) is dominated by one species (T. inflata) and indicates the most internal part of the schorre, partly emerged (5.2 to 5.5 m height).

Biofacies C4 (groups 3 - 1) is dominated by A. tepida, associated with some T. inflata and corresponds to a channel running across the schorre.

Biofacies D and E respectively correspond to samples showing some T. inflata individuals and barren zones. They respectively correspond to the transition zone towards the reclaimed marsh and the reclaimed marsh s.s. (5.5 to 6.0 m height).

The vertical distribution of foraminifera biofacies combined with 14C dates (Table G5) allows to reconstruct the succession of palaeoenvironments in the Monards marsh (Figure P16.24).

Discussion and conclusions

Although none of the cores reached the cretaceous substrate, it is likely that the sedimentation of the ancient bay initiated with a phase of fluvial infill during the early Holocene (Figure P16.24). This first phase is overlain by a transgressive marine sequence with basal schorre facies which were sampled at the base of core 9417. Then a 1 m thick internal slikke facies is deposited, and the maximum of the transgression coincides with the most external slikke facies (biofacies A1), appearing between 8 and 10 m depth in cores 9418 and 9417. Two samples were dated in this interval. 14C dates give a similar age of 5,600 to 5,400 yrs B.P. This transgressive sequence may be associated to the last phase of the rapid sea-level raise which occurred at the beginning of the Holocene until 6,000 years BP (Morzadec-Kerfourn, 1974; Kidson, 1986). At that time, the Monards Marsh was largely bathed by estuarine influences, and the sites of cores 9417 and 9418 were in the upper part of the intertidal zone, as suggested by the reduced number of individual tidal cycles (9 to 15) recorded during a spring/neap tide cycle. The exceptional preservation of the primary sediment structures suggests that wave penetration in the bay was almost inexistant, however numerous faults and

17 Palaeo-environmental Study Area P16 East Bank of the Gironde Estuary, west coast, France

slumpings indicate an instability of the deposits. After the maximum of the transgression, the central part of the marsh (core 9418) shows two oscillations from an external slikke to an internal slikke environment, with two levels showing maximum representations of the "marine group" (7.2 m and 6 m).

The overlying deposits show a reversed tendency, with a regressive sequence of deposits prograding towards the estuary. This corresponds to the phase of sea-level stabilization after 6,000 years BP and is marked by the appearance of schorre and brackish water marsh environments, corresponding to the deposition of the homogeneous dark grey silty muds. A basal unit of external and internal schorre facies is developed, between 6.8 and 5.2 m in core 9417, and between 5 and 5.4 m in the centre of the marsh (core 9418). One dated sample in this interval gives an age of 6130 +/- 130 yrs BP, which is inconsistent with the age of the underlying slikke deposits. This is probably due to a reworking of the plant debris which were sampled there as they do not come from a true organic horizon. The overlying unit is a brackish water marsh facies indicated by the alternance of biofacies E and D up to 2 m depth. By this time, the site of all the cores had reached an equilibrium height close to high spring tide level, and were almost permanently emerged. However, there is a brief reoccurrence of schorre facies between 2.4 and 2 m in the central part of the marsh (core 9418), indicating a last marine incursion, or at least a positive tendency in sea-level. A sample located just below this datum (2.55 m depth) gives an age of 2840 +/- 70 yrs BP. Two other dates were obtained in the brackish water marsh facies. The first sample, at 3.17 m depth in core 9417 gives a quite consistent age of 4640 +/- 100 yrs BP. The second sample, at the base of core 9504, gives an age of 9180 +/- 260 yrs BP. This is much too old compared with the other data and may be explained by a reworking of the plant debris sampled there.

The top part of all the sequences, corresponding to the compact silty mud facies, are totally azoic (biofacies E) and correspond to the terminal continental marsh phase.

The sections recovered with a hand auger near the landward western edge of the marsh show that internal schorre facies (biofacies C3) are developed above 9.8 m depth. They are replaced by a brackish water marsh facies around 5 m depth and by a continental marsh around 2.2 m depth. This simpler sequence than in the centre of the marsh confirms that the internal schorre progrades from the edges towards the center of the bay.

Very few curves of holocene sea-level variations on the French Atlantic coast have been proposed (Pirazzoli, 1993; Lambeck, 1997). These curves document a general common trend, with a rapid raise from 18 to 7 or 6 kyr B.P., but the very sparse data do not allow to determine if the present-day level was reached either with or without a phase above it (emerging and submerging coastlines respectively). Although the general trend outlined in this study is consistent with the current knowledge on holocene sea-level variations, the dates are too sparse and to questionable to allow further assessments. Due to the lack of true organic horizons interbedded in the deposits, the dated samples are mainly plant debris which may be partly reworked from older deposits.

However, on a chronological point of view, the results are consistent with what is known elsewhere in the Gironde estuary. While clearly younger than the infill of the median/internal parts of the estuary (Mellalieu et al., 2000), the deposits of the Monards Marsh appear to be contemporaneous to those of the wide North-Medoc marshes on the opposite bank of the estuary (Pontee et al., 1998). In particular, the age of 2800 B.P. for the slight positive sea-level tendency documented in this study is in good agreement with the dates obtained for the "Cordon de Richard" chenier-type feature (2800 to 2500 B.P.) developed on the opposite bank along the Pleistocene gravel terraces.

4.2 Archaeological synthesis of the "Monards" marsh (Figure P16.9).

Cette courte synthèse sur le marais des Monards a pour but de dresser un état de nos connaissances sur la présence de l'homme dans ce contexte et d'évaluer ses relations ou son adaptation aux conditions particulière de ce milieu.

18 Palaeo-environmental Study Area P16 East Bank of the Gironde Estuary, west coast, France

Comparé aux autres aires d'études de l'estuaire : marais de La Perge, de Reysson en Médoc ou de Saint-Ciers-sur-Gironde sur la rive droite de la Gironde, le marais des Monards fait figure de minuscule entité : deux km au maximum de pénétration dans les terres pour une largeur n'excédant pas 1 km. Il appartient en fait à cet ensemble de petites dépressions marécageuses échelonnées tout au long de la côte. Il s'insère dans le secteur des falaises et des hauts plateaux argilo-calcaires de la Saintonge maritime qui ont laissé peu de place à l'établissement de marais côtiers. Seuls de petits affluents de l'estuaire ont réussi à franchir cet obstacle par le jeu de petites fractures élargies et approfondies lors des phases de régression, puis recomblées lors des transgressions marines. Le marais des Monards sert de confluence à deux petits cours d'eau: le Moque-Souris et le Désir, avant de les évacuer vers l'estuaire par un goulet d'étranglement, à l'endroit même du port actuel des Monards.

Le marais des Monards est surtout situé au centre d'un haut lieu de l'occupation humaine en Saintonge depuis la fin de la Préhistoire. Les communes environnantes telles que Barzan, Arces, Épargnes, Meschers, Mortagne, Chénac et Saint-Seurin-d'Uzet, regorgent de sites archéologiques de toute époque. L'exemple de la commune de Barzan sur la marge occidental du marais est à lui seul édifiant. Et pourtant avec une telle densité de découvertes archéologiques, il est paradoxal que le marais des Monards n'ait pas cristallisé de façon évidente l'occupation humaine comme cela a pu être observé par ailleurs autour de l'estuaire.

Nous n'avons aucun site néolithique ou protohistorique à signaler directement établi sur les rivages de ce marais. Ils ne sont pourtant pas très éloignés et certains sont particulièrement importants. Les coteaux des alentours sont presque tous couronnés de camps du Néolithique récent ou final comme l'important camp fortifié peu-richardien de la Garde à Barzan ou ceux de Jau et Vil-Mortagne à Mortagne (EB8 Figure P16.9), de la Champagne et du Désir à Épargnes, de Bussas et Prézelles à Arces ou encore celui de Chez Reine à Sémussac. Il est difficile d'admettre que les hommes du Néolithique en Saintonge aient obligatoirement évité les zones humides. Les carottes réalisées dans le marais des Monards ont révélé que cette dépression marécageuse avaient été longuement ouverte aux influences de la dernière transgression marine pendant l'âge du Fer, avec une nette accumulation de sédiments. Ce processus peut laisser envisager un possible recouvrement des traces antérieures à l'âge du Fer comme cela a été observé dans le marais de Saint-Ciers plus au sud.

Cette absence de traces est également à étendre aux âges du Bronze et du Fer. Là aussi les établissement humains de ces époques sont nombreux sur les coteaux voisins : habitat du Bronze final et de l'âge du Fer du Fâ à Barzan (EB3 Figure P16.G9) habitat du Bronze moyen des Piloquets également à Barzan, complexe du Bronze final et de l'âge du Fer des Châtelards à Meschers et tous les enclos protohistoriques découverts au cours de prospections aériennes dans ce secteur. A cela il faut ajouter quelques objets isolés épars comme la hache en bronze de Cozon à Épargnes et celles de Meschers.

En fait, il faut attendre la fin des derniers bouleversements liés à la transgression pour qu'un semblant d'organisation apparaisse enfin autour du marais des Monards. Les premières traces concrètes appartiennent donc à l'époque gallo-romaine. Les diverses formes d'occupation du marais et de ses marges sont en grande partie liées à la naissance et au développement de l'agglomération antique du Fâ à Barzan (EB3 Figure P16.9). Il s'agit probablement du port de Saintes et de l'agglomération secondaire antique la plus importante sur les rivages de l'estuaire de la Gironde. Les fouilles et les photographies aériennes y ont révélé une organisation urbaine selon un plan de rue orthonormé, plusieurs temples dont le célèbre fanum du Moulin du Fâ, des thermes, un théâtre monumental bien plus important que celui de Brion en Médoc, de vastes entrepôts (horea), un aqueduc, etc. A cet immense complexe urbain, il faut ajouter un port envisager au premier abord à Chant Dorat, sur la façade maritime de la ville. Mais il est rapidement apparu que le marais des Monards avait aussi pu avoir ce rôle seul ou en parallèle avec le possible port de Chant Dorat. Cette hypothèse repose sur la mise au jour d'importantes structures sur la rive nord du canal de Moque-Souris, dans la partie amont du marais des Monards (EB5 Figure P16.9). Sommairement étudiés, les premiers éléments décrits se résument à des grands appareils interprétés comme de possibles môles d'accostage. Un

19 Palaeo-environmental Study Area P16 East Bank of the Gironde Estuary, west coast, France mobilier gallo-romain abondant accompagnait ces structures, en particulier de nombreux fragments d'amphore d'importation confortant l'éventualité d'une zone de trafic et de transport de marchandises. C'est également sur les marges de ce marais qu'il faut imaginer le passage de l'aqueduc romain de la ville du Fâ. Le point de captage (ou l'un des points de captage) a été localisé au niveau de la source de Chauvignac (EB6 Figure P16.9), à Chénac, sur le coteau dominant le marais sur son flanc oriental. Sur la rive opposée, sur le coteau occidental de Barzan, il est encore possible de suivre un tronçon souterrain de cet aqueduc sur une centaine de mètres, au sud du hameau du pied de l'Oeuf. L'ensemble devait s'étirer sur environ 3 km de long. Entre ces deux structures encore visibles, il reste beaucoup d'incertitudes sur les points de passage des canalisations. Il est cependant évident que l'aqueduc était obligé de franchir le marais et qu'il n'a pu le faire qu'au moyen de canalisations aériennes. Il est probable que les travaux hydrauliques ont dû utiliser au mieux les conditions naturelles du terrain. L'aqueduc devait franchir les zones basses dans leurs parties les plus étroites en amont du marais et peut-être en empruntant le plateau et les coteau de Langlade à Épargnes.

20 Palaeo-environmental Study Area P16 East Bank of the Gironde Estuary, west coast, France

Enfin les coteaux dominant le marais des Monards ont également été appréciés pour le panorama qu'ils offraient sur l'estuaire et pour l'établissement de domaines agricoles du type villa. L'une d'elles paraît avoir été installée à Langlade (EB4 Figure P16.9), sur les premières pentes du coteau dominant les paléorivages. Elles sont encore plus nombreuses dans les campagnes voisines : villas de la Petite Vache à Épargnes, du Taillis à Arces, du Vieux Bourg à Saint-Seurin-d'Uzet, etc., sans compter les habitats plus modestes ou les structures encore non identifiées nombreuses dans ce secteur.

Contrairement à de nombreux secteurs où les traces antiques tendent à diminuer ou à disparaître dès la fin du Haut Empire, comme c'est d'ailleurs le cas pour le site du Fâ dont le déclin est prononcé aux IIIe et IVe siècles, la présence de l'homme reste attestée sur les marges du marais des Monards jusqu'aux IVe et Ve siècles. Ce secteur est également le seul à laisser percevoir une continuité dans les formes d'occupation entre la fin de l'Antiquité et la Moyen âge classique. En effet, pendant le haut Moyen âge, un sanctuaire paléochrétien (mérovingien ?) est envisageable à Épargnes ; une nécropole de même époque (Ve-VIIe siècle ?) a été mise au jour dans le bourg de Barzan, sur le flanc occidental du marais et un petit sanctuaire chrétien est encore envisagé à Talmont (EB2 Figure P16.9) à l'époque carolingienne (IX-Xe siècle ?).

Quand les textes médiévaux éclairent enfin cette frange de la Saintonge, apparaissent des communautés humaines anciennes, dynamiques, bien organisées et quelquefois puissantes. De grandes abbayes sont établies dans ce secteur, comme celle de Mortagne plus au sud et presque toutes les paroisses autour du marais des Monards sont le siège de riches prieurés : prieurés augustins de l'abbaye de Mortagne à Épargnes, Saint-Seurin-d'Uzet et Mortagne même, prieurés de la puissante abbaye royale de Saint-Jean-D'Angély à Barzan attestée dès 1098, mais aussi à Talmont. Les terres agricoles autour du marais sont riches et ont centralisées depuis longtemps les établissements humains. Les activités médiévales directement liées au marais sont plus incertaines et l'on ne sait pas trop quand l'éventuel port de Moque Souris a pu glisser vers l'actuel port des Monards. La raison est quant à elle plus évidente et le colmatage de cette zone est sans doute à l'origine du déplacement des ces activités vers l'embouchure du ruisseau. Si l'activité portuaire ne fait pas de doute, les autres activités liées au marais sont rarement décrites dans les textes. De plus le havre des Monards n'a pas été l'un des plus importants de cette côte : ceux de Talmont (EB2 Figure P16.9), Saint- Seurin-d'Uzet (EB7 Figure P16.9) ou de Mortagne (EB8 Figure P16.9) étaient bien mieux placés ou plus accessibles au point de justifier la constructions de puissantes forterresses pour les surveiller et profiter de revenus substantiels.

En conclusion, on constate que du point de vue archéologique le marais des Monards n'a pas vraiment de point commun avec les autres secteurs d'étude de l'estuaire. Il doit son originalité à sa petite taille, à l'histoire de son comblement, à l'absence de traces archéologiques avant l'époque gallo-romaine, puis à son lien avec la grande agglomération du Fâ et la forte occupation de ses marges qui n'a cessé par la suite et qui en fait une unité d'étude spécifique intéressante.

Nb. For discussion, conclusions and references please see sections 7 to 14 of Palaeo- environmental Study Area P15 West Bank of the Gironde Estuary.

5. ACTUAL SEA-LEVEL RISE AT THE MOUTH OF THE GIRONDE ESTUARY

5.1 The impact of storms on the Gironde Estuary water level, lesson from the 27th December 1999 (see also Volume 1, Chapter 6)

Les marais estuariens qui bordent les rives de la Gironde sont situés à une altitude moyenne d'environ +2 m NGF et certaines zones, en particulier dans les marais de la rive droite, ont une élévation inférieure autour de +1 m NGF. Ces zones, proches ou même inférieures, au niveau des plus hautes mers de vives-eaux, sont protégées de l'inondation par des digues. L'altitude

21 Palaeo-environmental Study Area P16 East Bank of the Gironde Estuary, west coast, France maximum de ces défenses est atteinte au niveau de la centrale nucléaire du Blayais à Braud- et-Saint-Louis où les digues frontales culminent à +5,2 m NGF et les digues latérales à +4,75 m seulement (Pialat, 2000). L'altitude du sommet de ces protections est nettement inférieure tout au long de l'estuaire où, de plus, elles sont plus ou moins bien entretenues. Au cours de la tempête du 27 décembre 1999, même les plus hautes digues ont été submergées, provoquant une alerte de niveau 2 à la centrale nucléaire (CNPE du Blayais, 2000). Par ailleurs, elles ont été endommagées ou submergées provoquant l'inondation de plusieurs centaines d'hectares de marais.

Le risque d'inondation des marais au cours d'événements atmosphériques exceptionnels est donc réel. Pour l'évaluer, une simulation du niveau des eaux de l'estuaire dans certaines conditions extrêmes de marée, de pression atmosphérique, de vent, de vagues provoquées par le vent (mer du vent) et de débit des fleuves est indispensable. Cette simulation peut s'appuyer sur des données réelles obtenues au cours de tempêtes. La tempête exceptionnelle qui a traversé le Sud de la France le 27 décembre 1999 a pu servir de base à l'évaluation des surcôtes qui pourraient être atteintes en divers points de l'estuaire sous des conditions extrêmes (Pialat, 2000).

La tempête du 27 Décembre 1999, paramètres et impact sur le niveau de l'eau

Conditions atmosphériques le 27 Décembre 1999

Une violente tempête a traversé le Sud-Ouest de la France dans la soirée du lundi 27 décembre 1999. Elle a été provoquée par une dépression atmosphérique très importante (965 hPa) qui s'est brutalement creusée sur le proche Atlantique (Figure P16.1) et a traversé la France d'Ouest en Est en moins de 10 heures (entre 19 h et 4 h du matin). La rapidité de la chute de pression (Figure P16.2), comparable à celle des régions soumises aux cyclones, a provoqué des vents d'Ouest exceptionnels qui ont balayé la côte atlantique, l'Aquitaine et le Poitou-Charente en rafales atteignant partout les 100 km/h et dépassant parfois 190 km/h (Météo-France, 2000).

Le 27 décembre, le courant jet, qui souffle en haute altitude en principe à la latitude de l'Islande et de la Scandinavie, était anormalement décalé vers le Sud au niveau de la France. De plus il atteignait une vitesse exceptionnelle de 512 km/h au-dessus de Brest où il souffle en moyenne à 250 km/h. C'est cette configuration atmosphérique qui a occasionné le creusement de la dépression et son déplacement très rapide, à près de 200 km/h au travers de la France (Météo-France, 2000).

Les vents générés sur l'océan ont déferlé sur les côtes à des vitesses records : 198 km/h sur l'île d'Oléron, 173 km/h à l'embouchure de l'estuaire, 158 km/h au Cap-Ferret et 144 km/h à l'aéroport de Bordeaux-Mérignac, plus à l'intérieur et où le vent avait déjà été freiné par son passage au-dessus des terres. Outre ces vitesses de rafale, les vents moyens à 10 m au- dessus du sol ont atteint 120 km/h au Cap-Ferret (Figure P16.27).

Conditions hydrologiques le 27 Décembre 1999

Parallèlement à ces conditions atmosphériques, le débit de l'estuaire est durant cette période en accroissement rapide par suite de fortes pluies tombées sur les bassins versants des fleuves passant de 1000 m3/s environ le 25 Décembre à 4100 m3/s le 27 Décembre.

Enfin, le coefficient de marée pour la marée haute correspondant à la période de la tempête est de 77. Ceci se traduit normalement par un niveau des hautes mers variant de 1,90 m NGF à l'île de Cordouan à l'extérieur de l'estuaire, à 2,99 m NGF à Bordeaux (Tableau 5).

22 Palaeo-environmental Study Area P16 East Bank of the Gironde Estuary, west coast, France

Surcôtes de pleine mer observées le 27 Décembre 1999

L'estuaire de la Gironde est équipé d'un réseau de 10 marégraphes entre l'île de Cordouan et Bordeaux (Figure P16.28). Deux d'entre eux ont été détruits ou ont cessé de fonctionner en cours de tempête, les autres ont enregistré la marée réelle qui a pu être ainsi comparée à la marée prévue pour des conditions atmosphériques normales (Anonyme, 2000). Le Tableau 5 résume ces comparaisons et montre que la surcôte a été dans l'estuaire de plus de 2 m avec un maximum de 2,32 m à l'entrée du fleuve Garonne et "seulement" 2,01 m à , c'est-à- dire au niveau de la centrale nucléaire mais sur la rive opposée.

Cette surcôte est due d'une part, à la faible pression atmosphérique (987 hPa à Bordeaux au lieu de 1013 hPa -pression de référence utilisée pour les prévisions-) et surtout au vent. La surcôte liée à la pression atmosphérique locale devait être de 29 cm (Pialat, 2000). Par analogie, il peut être établi un graphique des surcôtes ou sous côtes en fonction de la pression atmosphérique (Figure P16.29).

La part du vent dans la surcôte observée est d'environ 85 %. Elle est liée au rafales de vent du secteur Ouest-Nord-Ouest qui ont poussé les eaux océaniques dans l'estuaire provoquant une onde gravitaire qui s'est propagée vers l'amont. Il n'est pas possible pour l'instant d'établir une abaque des surcôtes en fonction du vent ; cependant les données enregistrées au cours de la tempête du 27 décembre peuvent être prises comme maximum de référence.

La submersion des digues situées à +5,20 m NGF, alors que le niveau des eaux enregistré à leur droit n'atteignait que 4,52 (Tableau 5 Pauillac) prouve l'effet des vagues liées au vent (mer du vent). L'amplitude des vagues durant la tempête n'a bien sûr pas été mesurée, elle a été cependant évaluée d'une part en fonction de l'heure de submersion des digues à 5,20 m NGF et d'autre part par des abaques (Feuillet et al.), ces estimations pouvant être comparées aux observations de certains témoins (Pialat, 2000).

D'après les rapports officiels (CNPE du Blayais, 2000), l'inondation de la centrale nucléaire a débuté à 19 h 30 pour s'achever à 23 h. A ces heures là, les niveaux marégraphiques enregistrés étaient respectivement de 2,90 et 3,70 m NGF ce qui implique donc une amplitude des vagues de la mer du vent de 2,30 et 1,50 m ou encore des hauteurs de vague de 4,6 et 3 m pour que le niveau de l'eau dépasse celui des digues.

Ces hauteurs de vague sont nettement supérieures aux valeurs h1/3 théoriques calculées par les abaques de prévision en milieu peu profond en fonction du fetch, de la vitesse du vent, de sa durée et de la profondeur (Feuillet et al.). Ces valeurs théoriques sont de l'ordre de 2,20 m, c'est-à-dire de 1,4 à 2 fois moins hautes que les estimations précédentes. Cependant on pourrait admettre que la morphologie complexe du plan d'eau puisse induire un phénomène "d'interférences constructives" au cours desquelles l'eau apportée par une vague n'aurait pas le temps de se disperser avant qu'une autre vague n'arrive faisant ainsi monter le plan d'eau (Pialat, 2000).

Par ailleurs, de nombreuses personnes ont observé les vagues de l'estuaire pendant la tempêtes. Pialat (2000) rapporte que des habitants de Pauillac auraient estimé la hauteur des vagues à 7 m, valeur qui semble confirmer par le témoignage, plus crédible, du Capitaine du ferry qui traverse la Gironde entre Blaye et Lamarque et dit avoir vu "s'abattre sur son bateau des vagues dépassant 7 m". Ces témoignages semblent donc valider le calcul précédent fait à partir de l'heure de submersion des digues d'altitude connue.

En conclusion, la surcôte marégraphique de 2,01 m observée à Pauillac peut se décomposer en 0,29 m par effet de la pression atmosphérique (987 hPa) et 1,72 m par effet du vent.

23 Palaeo-environmental Study Area P16 East Bank of the Gironde Estuary, west coast, France

Le niveau, par rapport au zéro NGF, ayant été réellement atteint par les eaux sur les rives de l'estuaire au cours de la tempête en face de Pauillac, a donc été de : 2,51 m (le niveau prévu) + 2,01 m (la surcôte mesurée ou observée) + 2,30 m (l'amplitude des vagues) = 6,82 m NGF.

Cette valeur est à considerer comme un maximum : elle correspond au niveau de la crête des vagues à la côte.

6.2 Simulation des niveaux des eaux de l'estuaire dans des conditions météorologiques et marégraphiques extrêmes

La tempête du 27 décembre 1999 a eu lieu sous un coefficient de marée de 77 ce qui est une marée moyenne, les coefficients variant de 20 à 120 selon une probabilité d'occurence donnée par le tableau 2. Le niveau de la haute mer à Pauillac (ou Blaye) pour une marée de 77 est de 2,51 m NGF, il est de 3,66 pour une marée théorique de 120. Sans tenir compte du débit des fleuves, si tous les paramètres étaient à leur maximum, le niveau statique des eaux de l'estuaire pourrait atteindre 6,25 m NGF, se décomposant comme suit:

• coefficient de marée de 120 niveau 3,66 m • pression atmosphérique de 955 hPa surcôte 0,58 m (situation d'Octobre 1987 -Hontarrede et Duvere, 1988-) • vent 130 km/h surcôte 2,01 m (situation Décembre 1999) TOTAL 6,25 m

Cette valeur ne prend pas en compte les vagues. Dans le cas de vagues comparables à celles estimées et observées lors de la tempête de décembre (amplitude égale, voire supérieure à 2,3 m), on peut donc estimer la hauteur maximum pouvant être atteinte par l'eau sur les rives de l'estuaire médian (au niveau de Blaye ou Pauillac) à:

6,25 m + 2,3 m = 8,55 m NGF

6.3 Conclusions

Il apparaît donc que la conjonction de conditions atmosphériques exceptionnelles avec les plus forts coefficients de marée peut entrainer une élévation des eaux de l'estuaire (estuaire médian entre Pauillac et Blaye) jusqu'à l'altitude de 6,25 m NGF. Cette valeur est nettement supérieure à celle des estimations passées ainsi qu'à l'altitude des plus hautes digues actuelles (5,20 m NGF). Les marais estuariens qui bordent la Gironde sont donc particulièrement exposés au risque d'innondation et le "périmètre à risque" devrait être limité par la courbe de niveau de 6,25 mètres, et non par celle des 5 mètres comme c'est en général le cas.(Figure 16.30).

Par ailleurs la possibilité d'occurrence de vagues d'amplitude supérieure à 2 mètres, liées à des vents exceptionnels (mer du vent), fait que les eaux estuariennes peuvent atteindre, sur les rives, une altitude de 8,55 m NGF. Cette estimation devrait faire réfléchir les aménageurs, et les conduire à revoir le dimensionnement des digues de protection des marais estuariens.

La tempête du 27 Décembre 1999 a été particulièrement défastatrice et néfaste pour le Sud- Ouest de la France et il semble qu'aucun événement atmosphérique de cette puissance ne se soit produit au cours des temps historiques. Elle aura cependant servi d'expérience en vrais grandeur dans l'évaluation des risques côtiers d'innondation sur les rives de l'estuaire de la Gironde.

24 Palaeo-environmental Study Area P16 East Bank of the Gironde Estuary, west coast, France

6.4 References

Anonyme (2000) Tempête du 27 Décembre 1999 : cahier d'observation de la marée aux différents point de maesure de l'estuaire de la Gironde. Unpublished report, Port de Bordeaux, Département Hydrographique, 12p. Cnpe du blaysis (2000) Dossiers d'informations suite aux intempéries du 27 Décembre. Unpublished report, Rapport EDF, 16p. Feuillet J. Coeffe Y., Bernier J., Chaloin B. Dimension des digues à talus. Unpublished report, Rapport EDF n° 64, 169p. Hontarrede M. et Duvere J-P.Ouragan sur la France, 15-16 Octobre 1987. Météo Marine, n° ?, 11-19. Météo-France (2000) Tempêt sur le Sud-Ouest de la France. Lundi 27 - Mardi 28 Décembre 1999. Météo France, suppl. Dec.99, 16p. Pailette. Pialat S. (2000) Centrale du Blayayais : comment l'inondation du 27 Décembre 1999 a-t-elle pu se produire? Unpublished report, Rapport de Maîtrise Sciences de l'Environnement, Université Bordeaux 1, 42p.

25

Borehole LIII X LIII Y Z m NGF Latitude Longitude HOREST 9519 362.19 330.77 2.08 45° 14’ 18.50” 0° 41’ 28.61” HOREST 9520 362.28 330.73 1.56 45° 14’ 17.32” 0° 41’ 24.42” HOREST 9607 363.40 338.09 1.60 45° 18’ 16.86” 0° 40’ 45.47” HOREST 9608 365.28 338.73 1.70 45° 18’ 39.80” 0° 39’ 20.31” HOREST 9609 366.33 339.29 1.30 45° 18’ 59.16” 0° 38’ 33.08” HOREST 9611/12 367.38 330.77 2.00 45° 13’ 02.72” 0° 37’ 26.67” HOREST 9613 367.22 328.30 2.00 45° 13’ 04.48” 0° 37’ 34.10” HOREST 9702 366.05 338.76 1.10 45° 18’ 41.68” 0° 38’ 45.04” HOREST 9703 363.00 338.20 2.50 45° 18’ 19.95” 0° 41’ 04.00” HOREST 9704 362.20 338.05 2.30 45° 18’ 14.14” 0° 41’ 40.44” HOREST 9710 363.62 330.12 2.50 45° 13’ 59.16” 0° 40’ 22.01” HOREST 9711 364.50 329.42 2.50 45° 13’ 37.55” 0° 40’ 21.13” BRGM 755-1-01 365.90 338.81 2.00 45° 18’ 43.12” 0° 38’ 52.01” BRGM 755-1-05 365.36 338.51 1.00 45° 18’ 32.78” 0° 39’ 16.28” BRGM 755-1-14 365.30 335.88 1.10 45° 17’ 07.58” 0° 39’ 14.65” BRGM 755-1-20 363.72 334.86 1.00 45° 16’ 32.70” 0° 40’ 25.38” BRGM 755-5-24 363.78 333.62 0.75 45° 15’ 52.64” 0° 40’ 20.55” BRGM 755-5-26 363.58 328.50 2.52 45° 13’ 06.68” 0° 40’ 21.13” BRGM 755-5-71 362.69 331.52 2.00 45° 14’ 43.37” 0° 41’ 06.97” BRGM 755-5-63 365.80 328.66 0.89 45° 13’ 14.48” 0° 38’ 39.73”

Table 1 Locations of HOREST cores and BRGM boreholes used in this study. Italics indicate auger drills.

Borehole Upper Clay-Silt Peat Lower Clay-Silt Sand HOREST 9519 10.3 8.6 absent absent HOREST 9520 11.5 no data 11.2 absent HOREST 9607 14.0 23.5 8.6 absent HOREST 9608 12.4 no data 11.0 284.3 HOREST 9609 10.2 13.0 absent no data HOREST 9611/12 absent no data absent 204.4

Table 2 Mean value of grain-size in the lithological units (in µm).

Borehole Depth Age Cal. B.C. Material Sample ref. Meas. type (cm) (B.P.) HO 9520 500 - 503 5540 ± 70 4505 - 4250 Peat Beta 95392 Trad. HO 9607 350 - 355 5030 ± 70 3975 - 3665 Peat Beta 99067 Trad. HO 9608 311 - 316 4120 ± 40 2665 - 2475 Peat Beta 99068 AMS HO 9611/12 210 - 215 5630 ± 70 4605 - 4340 Peat Beta 104801 Trad. HO 9611/12 160 - 165 5160 ± 70 4090 - 3790 Peat Beta 104802 Trad. HO 9613 50 - 55 2670 ± 70 930 - 775 Peat Beta 104803 Trad. HO 9613 180 - 184 5630 ± 70 4605 - 4340 Peat Beta 104804 Trad.

Table 3 14C age estimates obtained from the upper peat unit.

CORE NUMBER LAMBERT III GRID REFERENCE ALTITUDE X Y m N.G.F.* HOREST 9417 362.50 351.50 3 HOREST 9418 362.45 351.10 3 HOREST 9502 363.06 350.95 3 HOREST 9504 362.40 351.70 3

* Nivellement Général de la France (a)

Core number Depth (cm) Age (years B.P.) Sample type Ref. labo Meas. type HOREST 317 4640 ± 100 Plant debris BA95083 AMS HOREST 860 5480 ± 100 Plant debris BA95084 AMS HOREST 255 2840 ± 70 Plant debris BA95085 AMS HOREST 515 6130 ± 130 Plant debris BA95086 AMS HOREST 890 5560 ± 190 Plant debris BA95087 AMS HOREST 506 9180 ± 260 Scrobicularia shells BA95088 AMS

(b)

Table 4 (a) Core locations and (b) 14C dates obtained on samples from the Monards marsh Figure P16.1 Location of the Gironde Estuary and of the Holocene marshes studied along the Estuary. Figure P16.2 Geological map of the Médoc peninsula and schematic West-East cross section through the Médoc peninsula. Figure P16.3 Comparaison between observed (A) and 2DH computed (B) turbidity maximum at low tide under mean river flow conditions (1000m3/s). Figure P16.4 Suspended sediment concentration in the navigation channel simulated with the 3D model. (a) with salt effect, (b) without salt effect. Figure P16.5 Horizontal distribution of simulated maximum bottom shear stress. Figure P16.6 Geomorphological map of the Monards marsh and core location. Figure P16.7 Morphology and topography of the Monards marsh and adjacent tidal flats. Figure P16.8 & Figure P16.9 Maps of the archaeological sites on the East Bank of the Gironde Estuary. Figure P16.10 Location of cores, auger drillings and boreholes used in this study. Figure P16.11 Altitudinal distribution of archaeological sites across the Saint Ciers-sur-Gironde marsh and its hinterland. Sites are presented by chronological period. Minimum occupiable datum (MOD) is calculated from each distribution. Figure P16.12a Fence diagram of the northernmost transect. Figure P16.12b Fence diagram of the median transect. Figure P16.12c Fence diagram of the southernmost transect. Figure P16.13 Lithology, median grain and diatom distribution in core HOREST 9608. MP marine plankton; MT: maine tychoplankton; MEB: marine/brackish epipelon; BP: brackish plankton; O: others. Figure P16.14 Location and nature of archaeological sites in the Saint Ciers marsh area. Figure P16.15 Palaeogeography of the Saint Ciers-sur-Gironde marsh pre-6000 B.P. Figure P16.16 Palaeogeography of the Saint Ciers-sur-Gironde marsh 6000-2200 B.P. Figure P16.17 Palaeogeography of the Saint Ciers-sur-Gironde marsh 2200-1800 B.P. Figure P16.18 Palaeogeography of the Saint Ciers-sur-Gironde marsh 1800-300 B.P. Figure P16.19 Lithology and grain-size distribution of cores HOREST 9407 and 9418. Figure P16.20 Lithology and grain-size distribution of cores HOREST 9502 and 9504. Figure P16.21 Distribution of foraminifera species and groups in core HOREST 9417. Figure P16.22 Distribution of foraminifera species and groups in core HOREST 9418. Figure P16.23 Distribution of foraminifera species in auger T3. Figure P16.24 Interpretative cross-section across the Monards marsh showing the internal distribution of holocene facies. See Table P16.5 for age range of 14C dates. Figure P16.25 Situation météorologique le 27 Décembre 1999 à 12h30. Figure P16.26 Evolution horaire de la pression atmosphérique (en hPa) et de la vitesse des rafales de vent (en km/h) à l’aéroport de Bordeaux-Mérignac du 27 Décembre 1999 à 6 h au 28 Décembre 1999 à 6 h. COGNAC - Charente ROYAN - Charente-Maritime 110 110 100 100 90 90 80 80 70 70 60 60 50 50 40 40 30 30 20 20 10 10 0 0 12 13 14 15 16 17 18 19 20 21 22 23 00 01 02 03 04 05 12 13 14 15 16 17 18 19 20 21 22 23 00 01 02 03 04 05

MERIGNAC - Gironde LEGE-CAP-FERRET - Gironde 90 120 80 100 70 60 80 50 60 40 30 40 20 20 10 0 0 12 13 14 15 16 17 18 19 20 21 22 23 00 01 02 03 04 05 12 13 14 15 16 17 18 19 20 21 22 23 00 01 02 03 04 05

Figure P16.27 Evolution horaire de la vitesse en km/h des vents moyens du 27 Décembre 1999 à 12 h au 28 Décembre 1999 à 5 h. PLAN DE SITUATION DES STATIONS MAREGRAPHIQUES

Figure P16.28 Localisation des stations marégraphiques de l’estuaire de la Gironde. Figure P16.29 Relation entre la pression atmosphérique (en hPa) et la surcôte marégraphique (en cm). Figure P16.30 Altimetric zonation of the littoral of the Gironde Estuary which put in evidence “submerged” regions for a level of the esturine waters at 5 and at 6.25m N.G.F. Palaeo-environmental Study Area P16 East Bank of the Gironde Estuary, west coast, France

Plate P16A Marshes of the East bank of the Gironde Estuary

(a) Coring in the Roman causeway at the La Melonne site

(b) Coring in a salt working in Le Cahour site

27 Palaeo-environmental Study Area P16 East Bank of the Gironde Estuary, west coast, France

( c ) The Fréneau site: a salt working site in the border of the estuary.

(d) The Fréneau site: a detailed view

28 Palaeo-environmental Study Area P16 East Bank of the Gironde Estuary, west coast, France

The Monards marsh:

(e) General view of the marsh

(f) General view (some cores were realised in the field in the foregound)

29 Plate P16b Foraminifera of the Monards marsh (scale: 100 µm) 1-Trochamina inflata; 2-Haynesina germanica; 3-Ammonio tepida; 4-Elphidium germanica; 5-Elphidium williamsoni; 6-Elphidium gerthi; 7-Elphidium incertum; 8-Elphidium magellanicum. Plate P16c Foraminifera of the Monards marsh (scale: 100 µm) 9-Asterigerinata mamilla; 10-Planorbulina mediterranensis; 11-Discorbis nitida; 12-Quinqueloculina sp.; 13-Fussurina levigata; 14-Fussurina orbignyana; 15-Cibiides lobatulus; 16-Miliolinella subrotunda; 17-Gavelinopsis praegeri. Plate P16d Foraminifera of the Monards marsh (scale: 100 µm) 18-Bulimina marginata; 19-Cassidulina crassa; 20-Cassidulina laevigata; 21-Lagena apiopleura; 22-Lagena striata; 23-Lagena sp.; 24-Bolivina sp.; 25-Trifarina angulosa; 26-Harynaosigma lactea. Archaeological Inventory: East Bank of the Gironde Estuary

Coastal Sea-level Environment Climatic Running Cultural Visible Unvalidated Coastal Unit Coastal Coastline Altitude Name Classification1 Classification 2 Source 1 Source 2 Description X Lamb2ext Y Lamb2ext Fragility Managment Change al Change Change Chronology Amenity Amenity Code Risk Context Type

Gallo-Roman villa, fortification from the XVII and XIX centuries, german La Pointe de 340.191 to 2069.832 to 0 to 30 m Villa, settlement settlement, building EB1 / bunckers from 1940-1945 destroyed in 1945. First anthic 3 1 3 1 3 3 2 3 I-D-M E 1 and 2 hard coast Suzac 340.497 2069.558 discoveries in the XVIII century, excavations in 1891 and 1936.

Anthique settlement indeterminated, medieval town with fortifications Le Bourg de 0 to 5 m Settlement building EB2 / (XIII century) with roman church (XII century), fortificated again 346.579 2064.627 3 1 3 2 3 2 2 3 I-D-M E 1 and 2 hard coast Talmont during the XVII century then abandonned

Anthique town (harbour of Saintes) occupied from the end of the soft 348.822 to 2064.714 to 5 to 30 m Le Fâ Town building, roads,… EB3 / Protohistory to the Bas Empire (III , IV centuries A.C.), theater, therms, 4 2 3 1 3 3 2 2 I-M F 2 and 3 estuarine 349.783 2063.888 streets, aqueduct, harbour, temple,… coast

Gallo-Roman settlement, possible villa with occupation from the IV soft building, ceramic 351.269 to 2063.604 to 5 to 10 m Langlade Settlement, villa EB4 / and the V centuries discovered during agricultural works around 4 1 1 0 1 2 0 3 I F 3 estuarine sherds, coins 351.240 2063.477 1995. coast

building, material Harbour, possible location of the harbour of Le Fâ (jetty?), abundant soft Le Canal de 3 m Harbour (amphorae, EB5 / materail with amphorae from the Haut Empire (I-II centuries A.C.) 351.255 2063.432 4 2 3 1 1 2 0 2 I F 4 estuarine Moquesouris ceramic sherds) discovered during agricultural works around 1995. coast

soft La Source de building, ceramic Anthique aqueduct of Le Fâ with ceramic sherds and coins (I-II 4 m Aqueduct EB6 / 351.875 2063.149 4 0 1 0 1 2 1 0 I-D-M F 3 estuarine Chauvignac sherds, coins centuries A.C.), discovered in 1955. coast

Château de St Castle (fortress) from the Medieval period reoccupied during the 2 to 10 m Castle building EB7 / 352.205 2060.451 2 1 1 1 1 2 2 0 M E 1 and 2 hard coast Seurin Modern period

Settlement, settlement, tools, Settlement occupied from the Neolithic to the Gallo-Roman period, 354.491 to 2058.740 to 4 to 40 m Vil Mortagne EB8 / 4 2 2 1 3 2 2 3 D-I E 1, 2 and 3 hard coast harbour material buildings and anthic harbour, abundant and rich material 355.115 2058.388

soft Salt working site from the end of the end of the first Iron Age 360.947 to 2049.276 to 3 m Les Granges Salt working site briquetage EB9 / 4 3 3 1 1 2 0 2 I E 6 estuarine discovered during agricultural works in 1970. 361.121 2049.031 coast

Important fortress dated from the IX century, partially destroyed 361.539 to 2048.277 to 5 to 35 m Conac Castle building EB10 / 4 0 1 0 2 2 2 1 I-M F 1 and 2 hard cost during the XIV / XV and the XVI centuries and rebuilt 361.702 2048.119

soft Les Salt working site from the 2nd Iron Age, discovered during 363.019 to 2045.658 to 2.5 m Salt working site briquetage EB11 / 4 2 3 1 1 2 0 2 D-I F 3 estuarine Cheminées agricultural works around 1972. 363.163 2045.520 coast

soft Two salt working sites from La Tène III discovered during 365.291 to 2043.232 to 2 to 3 m Le Petit Marais Salt working site briquetage EB12 / 4 3 3 1 2 2 0 2 I F 3 and 6 estuarine prospections in 1996 365.487 2043.009 coast

soft Salt working site from La Tène III discovered during agricultural 2 m L'Ardente Salt working site briquetage EB13 / 361.485 2042.065 4 3 3 1 1 2 0 2 I F 6 estuarine works around 1980 coast

soft Le Bois de 366.345 to 2041.308 to 2.5 m Salt working site briquetage EB14 / Salt working site from the 2nd Iron Age, discovered in 1996. 4 2 3 1 1 2 0 2 D-I F 3 estuarine Bouteille 366.334 2041.076 coast

soft Harbour and anthic depots (I, II centuries A.C.) excavated in 1984, 365.970 to 2039.398 to 2 to 3 m L'Ile Sèche Harbour building, ceramics EB15 1084 4 3 3 1 2 2 0 2 I F 3 and 6 estuarine first discovered around 1856. 366.323 2039.502 coast

soft Salt working site from the end of the of the Iron Age and the Haut 363.260 to 2039.007 to 2 m Le Cahour Salt working site briquetage EB16 / 4 3 2 1 2 2 0 3 I F 6 estuarine Empire discovered during agricultural works in 1991 and 1996. 363.711 2038.975 coast Archaeological Inventory: East Bank of the Gironde Estuary

soft Neolithic settlement with lithic material, discovered during agricultural 367.336 to 2034.607 to 2 to 3 m Les Ferrés Settlement flints, tools, material EB17 8615 4 1 1 0 1 2 0 1 D-I F 3 estuarine works before 1974. 367.717 2034.480 coast

Settlement, salt Neolithic settlement, two salt exploitation from La Tène III, harbour flints, briquetage, soft working site, from the Haut Empire (I, II centuries A.C), reoccupation with drainage 362.109 to 2030.951 to 1.5 to 3 m Fréneau ceramic sherds, EB18 4466 3 3 3 1 3 3 0 4 D-I F 6 estuarine harbour, finds, works during the XVII century discovered around 1960, excavated 362.248 2030.868 building, wreck coast drainage in 1991

soft 2.5 to 5 m Le Moulin Neuf Depository pot, material EB19 8601 Depository of the Late Bronze Age, discovered in 1907. 368.829 2030.159 4 1 1 0 1 2 0 0 D F 3 estuarine coast

Setlement, salt Settlement from the Bronze Age, salt exploitation from the 2nd Iron soft buiding,ceramics, 2 m La Patte d'Oie working site, antic EB20 8544 Age, anthic harbour reoccupated during the Medieval and/or the 365.747 2029.096 4 3 2 1 3 2 0 3 D-I F 6 estuarine briquetage, flints harbour Modern periods discovered around 1970, 1992, and 1993. coast

Small island in the middle of marshes occupied during the end of the soft Le Terrier settlement, flints, 3504 and Neolithic, the end of the 1st Iron Age or at the beginning of the 2nd Iron 367.349 to 2028.543 to 2 to 4 m Settlement, castle EB21 4 3 3 1 3 2 0 2 I F 3 estuarine Ricard metals 794 Age; reoccupation and fortification during the XVI century, 367.433 2028.404 coast discovered in 1975, excavated in 1878. Settlement, salt Lithic material from the Neolithic, salt exploitation from the 2nd Iron soft building, 367.242 to 2028.206 to 2.5 to 5 m La Barrière working site, EB22 8551.8552 Age, anthic settlement reoccupated by a castle (fortress) during the 4 1 2 0 3 2 1 3 D-I-M F 3 estuarine briquetage, flints 367.374 2028.095 castle Medieval period discovered in 1991 and 1992. coast

soft Two salt working sites from the 2nd Iron Age, discovered in 1992 365.407 to 2027.490 to 2.5 m Berdot Salt working site briquetage EB23 8588 4 3 2 1 1 2 0 2 I F 3 estuarine and 1996. 365.435 2027.302 coast

soft 3 m Vrillant Salt working site briquetage EB24 8592 ,8591 Two salt working sites from the 2nd Iron Age, discovered in 1992 365.41 2026.142 4 3 2 1 1 2 0 2 I F 3 estuarine coast

soft Two salt working sites from the 2nd Iron Age, discovered 364.879 to 2024.637 to 3 m La Sègue Salt working site briquetage EB25 / 4 3 2 1 1 2 0 2 I F 3 estuarine discovered in 1991 and 1992. 364.886 2024.498 coast