Using archaeological data for the understanding of Late-Holocene Sea of Galilee’s level fluctuations

In the Jordan Valley, reconstructed changes of the Sea of Galilee level have shown sharp fluctuations of the water elevation during the Holocene. In this paper, we provide new data originating from the excavations of Kursi Beach archaeological site located on the eastern shore of the Sea of Galilee and compare them with other data gathered from the archaeological site of Magdala, located on its western shore. Our data yield to constrain Sea of Galilee level changes between the Iron Age II (10th–9th centuries BCE) and the Crusader period (11th–12th centuries CE), a period of high interest for the archaeological community. We demonstrate that water level was around -212 to -210 m mean sea level (msl) for the Iron Age II period. Lake level rose to -208/-209 m msl during the Late Hellenistic/Early Roman period. Water level remained low (<-213/-214 m msl) from the Byzantine to the Crusader period (from 5th to 12th centuries CE). Our data provide new knowledge for the understanding of variations in the Sea of Galilee level in antiquity. We highlight that water level fluctuations must have been key factors taken into account in the habitation pattern.

Fauna is characteristic of a lake shore. Eucypris virens is, however, reported from temporary water bodies, spring, streams and fishponds (Meish, 2000). This locus is interpreted as a lake shore deposit in the Roman period. The presence of E. virens may reflects the existence of the wadi outlet or a spring nearby. This locus is used as a lake limiting point (Fig. 3).
Locus 28010 is a 75cm thick unit, located between -214.23 and -214.98 m msl. It is mainly composed of silts and clays (83%) with 15% of sands and 2% of gravels. Mollusks are extremely rare and only represented by two individuals of Melanopsis praemorsum and one individual of Theodoxus sp. The ostracod density is high with 466 valves for 10 grams of sediments. Pseudocandona sp. (28%), Ilyocypris sp. (25%), Potamocypris sp. (20%) and Cyprideis torosa (15%) dominate the assemblage. Pseudocandona sp. has a wide distribution and is frequently found in lakes and pounds on fine sediments. As well as Ilyocypris sp. and C. torosa, it is found with relatively high frequencies in both standing and running waters (Mischke et al., 2012). This locus is located at the bottom of the structure into which the foundations were placed. During excavation, this layer has been found to be waterproof. Ostracod assemblage and sediment texture of the unit differs from the species commonly encountered on the shore of the lake. One possibility is that the sediment was intentionally deposited in this area for the construction of the structure. The ostracod assemblage is dominated by C. torosa (77-91%) and Ilyocypris spp. (7-20%) with few Candonopsis (2-3%) and Pseudocanona (1-3%) specimens. Density range from 1160 to 4100 valves for 10g of sediments. Excavation of locus 28205 (-213.97/-214.08 m msl) reveals the presence of a piece of wood in the recess of the western wall ( Fig. 3 and Fig. S3). This wood, likely used as a "cork", has been dated to 1410 -1520 years BP (430 -540 cal. years CE). Locus 28203 is dated to 750 -910 years BP (1040 -1200 cal. years CE) and 28202 to 800 -920 years BP (1030 -1150 cal. years CE). These loci are interpreted as a natural infilling of the structure after lake level rise and indicate higher lake levels at that time. The cork is dated from the same period as the structure. We can consider it as a lake limiting point (Fig. 3). Loci 28107, 28106, 28105, 28104 and 28102 A test pit was excavated on the outskirts of the structure to establish the living level of the period of construction of the structure. A similar picture to that of the inner structure was discerned. Seven loci have been excavated. Locus 28107 is a 10cm thick compact slanting unit, whose bottom is situated between -213.73 and -213.56 cm msl. It is composed of sands (49%) and gravels (30%) with finer sediments (silts and clay=21%). In this locus, we identified Theodoxus jordani and Melanopsis praemorsum representing respectively 42% and 29% of the total assemblage. The third species, counting for 19% of the assemblage (n=19) is Bithynia hawaderiana living under stones and commonly found along with T. jordani and M. praemorsum on the shore of the lake (Pollingher et al., 1978;Dzikowski et al., 2003) Locus 28102 is a pure tight yellow silty sand unit. Silts and clay represents 55% of the total sedimentary texture while the remaining 45% are sands. This sample contains no molluscs. However, the density of ostracods is high (2600 valves for 10 g of sediments). The ostracods assemblage is dominated by C. torosa but it represents less than 50% of the total assemblages. Ilyocypris spp. (27%) and Loxoconcha galilea (26%) are the two other main species of the assemblage. This locus is also interpreted as the deposition of shallow coastal sediment. It may indicate further rise in sea level. The ostracod fauna is typical of the littoral zone of the lake (Serruya, 1978).

Lake level index points
Data used to produce the lake level index points represented in Figure 5 ID in Figure 5 Site

Fig. S1
The breackwater and Tel Kursi (Area A). This image highlight the connection between the breackwater and the tell and reinforce the fact that both of them can be reasonably dated from the Iron Age II period. Photo: Michal Artzy.

Fig. S2
The fishpond/wishing well excavation (Area C). (a) Plan of Area C/Beach area (M. Edelcopp and B. Arubas); (b) Spillage on the N side of Round Structure (view from the north); (c) Inner and outer excavation of round structure (view from the north-west); (d) Two layers of construction and foundation (view from the west). Photos: Michal Artzy.

Fig. S3
Lake-level index points obtained from the study of the fishpond at Kursi (Area C) and details of the loci excavated. Description of the loci is given in Supplementary text.