Mersa/Wadi Gawasis, Mission 2007-2008 - Kathryn A. Bard and Rodolfo Fattovich et alii

Geology, and Archaeozoology - M. Badr, A. Carannante, D. FitzGerald, and C. Hein

Geological investigations were conducted in order to assess the structural stability of the caves and demonstrated that the coral terrace exhibit a complex system of fractures which will have to be carefully monitored for a proper preservation of the site.
Coastal geological investigations were conducted along the wadi bed and the base of the southern and western slopes of the terrace. These investigations are supporting the hypothesis that the mouth of Wadi Gawasis was originally a lagoon.
Geoarchaeological investigations in the harbour area provided evidence of a sequence of sea beaches which were occupied during the Middle Kingdom.

Geology
M. Badr


Introduction
Mersa/Wadi Gawasis lies along the Red Sea coastal plain, a relatively narrow (2-15 km wide) zone lies between the Precambrian basement rocks and the Red Sea shoreline. It is composed of Phanerozoic rocks of Quaternary, Tertiary and Cretaceous clastic and carbonate formations that are unconformably overlained by the Precambrian metavolcanics, metasediments, and intrusive rocks comprising the Red Sea Hills. For approximately 25 million years, this region has been subjected to syn- and post rift sedimentation along the Red Sea rift zone. The substrate encompassing the occupation site developed during the late Quaternary in the proximity to coral reefs and shallow-water carbonate and clastic sedimentation under arid conditions of the Red Sea coastal plain.
Wadi Gawasis is located approximately 500 m from the present shoreline. The site consists of topographically subdued Quaternary deposits. The strata are sub-horizontal to gently eastwardly dipping; Pleistocene reef terraces are overlain by the Holocene fluvial and colluvial deposits, creating a narrow plateau along this stretch of the Red Sea coastline. This plateau is drained by westerly flowing, highly infrequent discharge events from Wadi Gawasis.

Stratigraphy
Late Pleistocene / Early Holocene facies in the study area are represented by a series of carbonate reef strata interbedded with siliciclastic deposits. Three or four distinct siliciclastic / coral terraces, each up to 15 m thick, were identified in the study area. A NNW-SSE trending fault scarp forms a prominent marine cliff, located between 10 and 16 m above mean sea level, where the caves have developed.
The stratigraphy of the Quaternary (Pleistocene and Holocene) facies at the site can be subdivided into the following stratgraphic units:
1) Early conglomerate terrace
2) Calcareous sand
3) Raised coral reef limestone
4) Alluvial conglomerate terrace
1) Early conglomerate terrace: This unit is multicolored, partially compact, with a variable thickness and uneven contacts with adjacent units. The various clasts that compose this unit are derived from the neighboring basement and pre-Quaternary rocks. These clasts are poorly sorted, matrix-supported, composed of rounded to well rounded pebbles, cobbles or boulders, up to 25 cm in diameter. The matrix is calcareous sand, silt and mica.

2) Calcareous sand: The calcareous sand unit is up to 1.5 m thick, yellow-grey in color, partially compact, and coarse- to very coarse–grained with occasional pebbles. It consists of semi-consolidated quartz, pink and white feldspars, dark grey to green mafic ferro-magnesium minerals and mica, cemented by carbonate. This unit is argillaceous at the base becoming calcareous toward the upper part of the sequence.

3) Raised coral reef limestone: The coral reef limestone is yellowish white to yellow in color, hard, porous and caverneous associated with an uneven surface. Carbonate-rich fossils (coral reefs, molluscs, echinoids, and shell fragments) as well as algae and rock fragments are often found within this unit. A 1-m thick conglomerate forms the base of this unit.
Coral reefs are structures produced by the organic secretion of aragonite (a form of calcium carbonate). The ancient raised coral reefs of Wadi Gawasis are located above the present shoreline and were elevated to their present position above mean high water either by local crustal movements (uplifted) or by lowering of the sea level (NOTE: these ideas are developed more completely in the FitzGerald and Hein 2008 Wadi Gawasis report).

4) Alluvial conglomerate terrace: The alluvial conglomerate terrace unevenly overlies the uppermost part of the coral reef limestone. It is composed of subangular to subrounded clasts, derived from local Precambrian basement and late Mesozoic and Cenezoic sedimentary rocks, in a sand with silty clay matrix. Quaternary sediments were deposited in a brackish or highly saline marine environment.

Structural Geology
Mersa/Wadi Gawasis is located within a fault/fracture system associated with the Red Sea rift. The faults have two primary orientations: NNW-SSE to NW-SE and NE-SW Caves 2, 3, 4 and 5 were structurally investigated. They are extremely dissected by five major fractures dating to the late Pleistocene/early Holocene, and are associated with about four additional minor fractures (Figure 1). The predominant trend is NW-SE. They are characterized by several bifurcations at different heights with orientations ranging from 10o and 85° for the same fracture. Several masses and blocks have experienced gravity collapse as a result of weaknesses associated with these intersecting fractures. In some locations, these fractures intersect the subhorizontal bedding at the studied outcrop breaking apart units and leaving these sections precariously perched. As an example of the present situation, slight movements were observed forming a narrow space between the fractured blocks along a secondary NNE-SSW fault/fracture plane.


Coastal Geology
D. FitzGerald and C. Hein

Introduction
Coastal geological studies were conducted at the Mersa/Wadi Gawasis archaeological site from 3-10 January, 2008. The primary objective of this work was to determine the geologic setting of the Wadi Gawasis site at the time of occupation, 4000 years before present (BP).
Field studies accomplished during the 2006-07 field season concluded that a bay once existed in the area presently occupied by Wadi Gawasis. The occupation site is located south of the coralline cliffs along the margin of the paleo-bay (FitzGerald and Hein 2007). During this first field season, 28 augers cores were taken along eight transects (3-4 holes per transect, generally spaced ~10 m apart). Eight of these were cored in pre-excavated archaeological pits and the other 20 auger holes were cored in ~1m3 pits pre-dug before augering. From these studies, a coralline/beach rock surface was found and mapped within the wadis sediments along the cliffs. This surface roughly parallels the cliff front and thins in a wadis-ward direction (seaward) such that the seaward most cores into this surface could penetrate the beach rock. At these locations the beach rock was approximately 30 cm thick; here it likely represents a paleo-shoreline.
Efforts during the 2007-08 field season were directed at further defining the bay seaward of the beach rock surface. Deep (> 5 m) cores were taken across the lateral extent of the wadi to better constrain its depth and geometry. Additionally, foraminifera studies and comparisons of the sedimentology and geomorphology of this area to similar sites to the south were used to determine the paleo-environment of the study site at 4 kya (thousand years ago).

Geological Setting (see Fattovich and Bard 2007: 2-6)
Mersa/Wadi Gawasis is located 23 km south of the port of Safaga, at approximately 26o33'05" north, 34o02'08" east. A small embayment exists at the present shoreline; a natural cut in the coral reef within this embayment produces deep water close to shore. Sedimentological and geophysical studies were conducted along the western and southern slopes of a cliff located approximately 450 m from the present shoreline. This formation contains inter-bedded conglomerates, reef rock, and fossil coral and is overlain by fluvial sand and gravel. Reef rock capping the formation and its shallow eastward dip are evidence of a broad uplift, likely related to the cratonic rifting that began in the Eocene Epoch (54.5 ma – 33.7 ma) and accelerated during the Oligocene Epoch (33.7 – 23.8 ma; Sultan et al., 1993). Tectonic rifting split the Arabian–Nubian Shield and created the Red Sea. The Red Sea Basin is considered an active rift where seafloor spreading has occurred for the last 5 million years (Sultan et al. 1993).
Occasionally during the Oligocene, the Bab el Mandeb, which is a narrow strait between the Arabian Peninsula and African continent connecting the Red Sea to the Indian Ocean, closed. During closure, the Red Sea produced evaporite sequences, which are found throughout Wadi Gawasis and the surrounding region (Sultan et al. 1993).

Coastal Setting
Mersa/Wadi Gawasis is located in the northwest corner of the Red Sea, approximately 140 km from the southern end of the Gulf of Suez. This microtidal coast has a mean range of 0.5 m. Changes in wind regime produce a lower summer mean sea level (0.5 m) than in winter. The northern part of the Red Sea is dominated by north-west winds, with velocities ranging between 7 and 12 km/hr. During summer months, persistent NW winds lower mean sea level and drive surface waters southward at velocities of 15-20 cm/sec. This current coupled with wave action is responsible for transporting sediment southward (Lindquist 1998).
The present mean wadi surface is situated approximately between 0.5 and 1.5 m above mean sea level (sea level data from Tilia, personal communication 2008). A poorly constrained sea level curve for the northern Red Sea is based on δ18O foraminifera records from a sediment core taken south of the Sinai Peninsula (Figure 2; Siddall et al. 2003). The curve suggests several relative sea level excursions on the order of 10 m during the past 10 kya. However, little confidence is placed on the details of this curve due to the large error bars. The curve indicates that sea level has been rising slowly during the past 4 ka.
Though a dearth of well-constrained sea level data exist for the Wadi Gawasis region of the Red Sea, it is possible that a higher than present stand of sea level (up to +4 m) may have existed at the study site 4 kya. Siddall (personal comunication 2008) has suggested that such a sea level highstand could have existed in the Red Sea and fit within the uncertainties of his sea level curve. A mid-Holocene highstand along equatorial and south hemisphere coastlines is well documented in the literature. It is suggested that this was caused by isostatically-induced equatorial siphoning of waters from far field oceans to coastal regions that underwent glaciation during the last ice age (i.e., Mitrovica and Milne 2002; Peltier 2002; Milne et al. 2005; Lambeck et al. 2003). Peltier (2002) cites a number of middle latitude northern hemisphere Pacific Ocean islands that also show evidence of this highstand. Lambeck et al. (2003) provide evidence for a similar highstand in the nearby Persian Gulf at the time of occupation of Wadi Gawasis (4 kya); however, the Persian Gulf represents a very different tectonic regime. Despite its proximity to the Red Sea, sea level in the Persian Gulf cannot be used as a direct proxy for sea level in the Red Sea. There is some local evidence of a higher than present sea level exists along the shoreline (Figure 3); however, the mode of formation of these features is unknown. They could also be explained by storm processes.
Therefore, no firm data presently exist for a higher than present sea level at Wadi Gawasis and additional analyses are needed to determine the position of relative sea level 4 kya.

Data Collection
During the 2007-08 field season, a pulse auger system was used to collect 11 deep (3-6 m) auger cores to supplement the 28 short (1 – 3 m) auger cores taken during the previous field season. Several auger holes from the previous field season were re-entered and three of the previous transects (T3, T5, and T6) were extended across the Wadi (Figure 4). Core sites were separated by several hundred meters such that southerly-trending transects ended half way across the wadi.
Due to the inability of the pulse auger system to penetrate hard strata (e.g. salt and pebble layers), pits were excavated to a depth of between 1.5 and 2 m by workers. This allowed auger cores to start below the wadi surface and below any hard surfaces. A hand auger was used to dig to the water table. The pulse auger system was then used to retrieve sediment from the core holes. This method recovers sediments at known depths but does not allow for the observation of sedimentary structures. The detailed stratigraphy of the sides of the pits was recorded; coarser stratigraphic units from the auger and pulse auger holes were also recorded. The cores generally terminated between 4-6 m below the wadi surface. Sediment samples were taken from several pits for foraminiferal analyses.

Results
A total of eleven cores was taken, reaching depths between 3 and 6 m below the present wadi surface. Cores WG-T3A4 and WG-T5A5 were taken in pits by the same name from last year; this method served to extend previous cores. Stratigraphic data from the cores were logged by transect and data converted to graphic core logs (Figure 5). Cores were subdivided by the sediments found at the bottom of each hole and these data were added to the database from the 2006-07 field season (Figure 4; Table 1). Twenty of the cores terminated at an impermeable layer found several meters below the ground level. This layer consisted predominantly of carbonate cemented coral and beach rock (Figure 6). The remaining eighteen auger cores ended in a grey, medium-fine sand; the depth of these holes was constrained only by the ability of the coring equipment to penetrate deeper below the water table.
Shells were collected from all auger holes during both years, and they generally increased in abundance, especially gastropods, from the sheltered to the more open water setting (T1 to T8). Shells found in the protected areas (T1-T4) as well as those found in the lagoonal sediments in the main wadi were generally more delicate, consisting of thin bivalves compared to the thicker shells of the higher energy environments (T5-T8). However, one relatively fragile bivalve occurs throughout the study area. Results of malacological investigations by Alfredo Carannante of shell samples collected are presented in Table 2. Nearly all shells (bivalves and gastropods) collected typically live in intertidal or subtidal muddy or sandy bottoms in the very nearshore or lagoons.
Sediment samples were retrieved from several cores in the wadi. These samples were studied for foraminiferal abundances (Tables 2-3). All results of these foram analyses suggest that the sediments were deposited in a lagoon with infrequent freshwater inputs. These species live in a salinity of 35-53 parts per thousand in depths of less than 50 m in warm, lagoonal environments.

See Table 1, Table 2 and Table 3

Discussion
Preliminary analyses suggest that during the time of occupation (radiocarbon dating of shells, currently being analyzed, will produce exact date precise dates) the wadi was a shallow, semi-enclosed bay (~1.5 km2) with an open connection to the Red Sea.
The base of the present cliffs consists of a narrow coral-beach rock platform. Archeological excavations during the 2006-07 and 2007-08 field seasons found a number of occupation sites within one meter above this beach rock surface. From the cliffs toward the wadi of the coral-beach platform a medium-fine grey sand is found at depth in all auger cores. The grey color likely results from anoxic conditions of previous water table levels resulting in iron oxide coatings. The occurrence of this sediment below the thin beach rock layer found in WG-T7A3 indicates a ubiquitous distribution of this sediment at depth throughout the study area. Based on the sedimentology and relative abundances of mollusk and foram species, this sediment is interpreted to have been deposited in a tidal lagoon. Shell and foraminifera distributions indicate that much of paleo-bay was protected and a low energy environment. Wadi processes have since infilled the bay.




Depth of the Bay
The depth of the bay is unknown and will be determined through dating methods and infilling rates of the bay during the past 4,000 years. Dating of shell samples will provide minimum dates for which the embayment existed. Additionally, deposition rates can be determined from a series of dates taken from shells vertically stacked within a single core. From the radiocarbon dates infilling rates and depth of the bay can be roughly determined. Shell samples collected over the two years of field studies at Wadi Gawasis have been sent to the French Institute for Oriental Archaeology (IFAO) in Cairo for radiocarbon dating.

Quseir al-Qadim: A Geological Equivalent of Wadi Gawasis?
During the 2007-08 field season Myos Hormos / Quseir al-Qadim was visited. Quseir is located approximately 50 km south of Wadi Gawasis and contains archeological evidence for the existence of both Roman and Islamic ports (Blue 2007; Figure 7). This site shows a number of similarities with Wadi Gawasis. Stratigraphic sections created from cores taken at this site are nearly identical to those at Wadi Gawasis (Figure 8). The paleo-environment of the Quseir site has also been interpreted as an open lagoon. Over several thousand years, the lagoon was filled by wadi sedimentation processes and by 900 years BP had become a small embayment (Figure 9). This bay was subsequently filled, shifting the shoreline several hundred meters seaward of the paleo-ports. Though wadi infilling processes alone can explain the evolution of this bay, a change in relative sea level since the time of occupation could complicate this scenario. Evidence at Wadi Gawasis suggests that this site followed a similar evolutionary history from a Middle Kingdom port to a filled, dry wadi.

Future Research Directions
Future research will follow two lines:
1) Deep drill cores taken during a future field season can provide valuable information in sea level and paleoenvironmental reconstructions at Wadi Gawasis. Additional coring transects (auger cores and drill cores) across the eastern side of the Wadi will be useful in determinations of the size of the opening of the bay to the Red Sea and the geometry of the lagoon.
2) An ongoing effort continues to determine the paleo-fluvial hydraulics of the wadi. The climate of the region during the past 4,000 years is currently being investigated to determine periodicity of the wet-dry cycles. A preliminary investigation of the up-valley wadi terraces, indicate evidence of major floods that topped the terraces. These floods would have transported huge quantities of sediment to the coast and perhaps led to the filling of the former port.
Determinations of wadi infilling rates from samples collected during the 2007-08 field season will seek to determine potential discharges and sediment transport rates during large magnitude flood events. Determination of the periodicity of flood events will help to provide a sediment budget for the region during the past 4,000 years.


Archaeozoology
A. Carannante

Data collection
Archaeozoological remains from the excavated areas were taxonomically identified and catalogued in a data-base. Data were recorded with information about the archaeological context (excavation unit, square, stratigraphic unit), anatomical identification, presence of marine erosion or bioerosion marks, presence of natural or anthropogenic holes, and information about the sediments context.
The database at present includes information about 3,226 archaeozoological remains: molluscs, crustaceans, and barnacle shells, and also fish and marine reptile bones.
Concentrations of large shell remains (mainly Lambis) found on the coast of Mersa Gawasis have been studied in addition to studies of the archaeozoological remains from the excavated units during this field season. Large shell remains were recounted, their state of preservation was studied together with bioerosion marks, and their taphonomy was recorded.
Ethnographic information was also collected in interviews with Red Sea fishermen and local people in order to verify the present use of Lambis sp. molluscs and shells, and the recent distribution of this species.

Preliminary results
WG 32
This excavation unit is in front of Caves 2, 5, and 6, on the top of the sandy terrace slope. A few shells, mainly Anomiidae (25 MNI), were found, most of which had wood impressions on the bottom side. These shells had been scraped away from well planed wood surfaces, probably ship timbers. A few Nerita shells (with natural holes) were also found.

WG 55
This area is contiguous to WG 32 and in front of Cave 7. Several gastropod shells with holes were found there. Most of them had been pierced by predators, but a Conus shell with a drilled hole and a cowrie shell with a percussion hole were also found.
A few Anomiidae and barnacle shells with wood impressions were excavated. Also found were several valves of different Veneridae species, some gastropods, and two Sparidae fish vertebra without erosion marks, which may be the remains of food.
Twenty large fragments of the flat bones of a sea turtle were found at the entrance of the Cave 7. The fragments were restored by Pasquale Musella into four parts of a turtle shell from an Eretmochelys imbricata specimen: two shields of the lower plastron, a marginal plate, and a costal shield with two ribs from the carapace. The xiphiplastron and hypoplastron show corneous tortoise remains on the external side, while the external side of the dorsal costal shield shows a reddish coloration due to fire exposure. No other turtle bones or tortoiseshell fragments have been found in the excavation. Such evidence points to exploitation of the corneous tortoiseshell, which was generally obtained by exposing the carapace to heat.
Five mother-of-pearl spoons made from the nacreous valves of large specimens of Unioniacea freshwater species have also been excavated. The spoons, of different sizes, were shaped using a saw and then were ground down to obtain smooth edges. Other fragments of mother-of-pearl suggest that more nacreous spoons had also been made.
Another broken shell spoon, made from the last whorl of a large Charonia tritonis shell whose columellar callus is still visible, was also excavated.

WG 45, WG 46, WG 47, WG 49, WG 50, WG 54, WG 57
These excavation units lie in what is believed to be the beach of the "harbor area." Several hearths and many potsherds from at least two occupation phases were excavated here. Many specimens of edible gastropods, several Circe crocea valves (MNI 25), many fish bones (mainly parrotfish and sea-bream), and several stingray remains were found in these strata, usually associated with layers of charcoal and ash. They probably represent food remains.
Percussion marks were found on a Lambis shell fragment associated with an ash layer, evidence of alimentary use of this species. A few shells have anthropogenic holes (two Conus and a Polinices).
The most peculiar find was a large valve of a Tridacna shell with clear traces of fire exposure. Only the internal side of the valve is reddened at the bottom and blackened on the edge. This evidence suggests that the large shell was used as a lamp.

Lambis shell middens on the coast:
Study of the large shell concentrations on top of the fossil coral terrace which borders the coastline was conducted to explain their use. More than 1,050 Lambis shell fragments were found, with a minimum number of individuals (MNI) of at least 703 specimens, which were associated with several Tridacna shells. Most of the Lambis shells show consistent patterns of fragmentation and only a few are complete. A large part of the assemblage was amassed in a structure made of large blocks of fossil coral. Lambis shells were also concentrated in other structures built with large slabs of conglomerate stone covered by coral blocks.
Different uses have been suggested in order to explain the large shell concentrations:
1) Alimentary use: study of the bioerosion marks revealed that more than the 72% of the Lambis shells were collected after the death of the mollusc. Only 17 MNI appear not to be eroded and with a different pattern of fragmentation; they are the only Lambis shells that might have been collected alive for food.
2) Use as a raw material: no working marks are present on the Mersa Gawasis remains. Lambis shell bangles obtained from the spire or from the last whorl were used in the ancient Egypt and the Near East, but none of these parts of the Mersa Gawasis Lambis shells appears to have been worked.
Seventy cap portions of the last whorl have been found in the Lambis shell concentrations. This evidence excludes the use of this part of the shell to produce spoons similar to those found in the site.
A survey along the coast and interviews with local people revealed that the constant pattern of fragmentation that characterizes the Lambis shell remains from the archaeological site is different from that produced by percussion. The pattern of fragmentation of the archaeological Lambis shell remains is typical of Lambis shells eroded by waves. Dozens of similar Lambis shell fragments can now be collected on the beaches a few kilometers from Mersa Gawasis.
3) Use as a building material: along the Red Sea coast fishermen's huts are found which utilize Lambis shells as a building material instead of mud-bricks. The Mersa Gawasis Lambis shells, however, were found in different contexts: below the coral blocks of the shrine structures, on the top of the structures, and around them. Many of the shells were also found between the coral blocks. Thus, the shells were deposited during the construction of these structures.

Pleases find attached .pdf files with the full report and the related figures