Mid- to late-Holocene coastal vegetation patterns in Northern Levant (Tell Sukas,Syria): Olive tree cultivation history and climatic change

Environmental and climatic setting

Tell Sukas (35°18′16″N; 35°55′22″E) is a coastal site in the Syrian lowlands, located between two natural harbours, 7 km southwards from Jebleh and 35 km from Ras Shamra (Northern Levant; Figure 1). The site, excavated by the Carlsberg Expedition to Phoenicia (1958–1963) was inhabited since the middle of the seventh millennium BCE. The artificial hill (24 m a.s.l.) was occupied more or less continuously from the Neolithic up to the end of the Hellenistic period (Lund, 1986, 2004; Oldenburg, 1991; Riis and Thrane, 1974; Riis et al., 1996; Thrane, 1978). Such a location enables a favoured agricultural exploitation with high water availability (Sanlaville, 2000; Weulersse, 1940) and makes this situation particularly convenient for trade and exchange between Inner Syria, Mesopotamia, Anatolia, Central and Eastern Mediterranean. However, the transition between the seashore and mountainous landscapes is rather abrupt along the Levantine coast. The Jabal an Nuşayrīyah (Alawite Mountains) stretches parallel to the coast and is regarded as the northern continuity of the Mount Lebanon. This mountain range reaches 1583 m a.s.l. and is mainly made of Cretaceous–Jurassic limestones (Hardenberg and Robertson, 2007). Along this rugged coastline, only three plains are suitable for a well-developed agriculture in the Northern Levant: the Akkar, the Jebleh and the Latakia plains. Tell Sukas, located in the Jebleh plain, is positioned in prime location for agriculture. Olive trees, vineyards and cereals have been the most important cultivated essences during the Antiquity (e.g. Zohary et al., 2012). Today, olive trees are mainly cultivated on the foothills of the Jabal an Nuşayrīyah. The perennial Nahr as Sukas River, originating from the piedmont 10 km eastwards, outflows in the southern bay of Tell Sukas. Pollen was transported to the coring site mainly by streams from the Jabal an Nuşayrīyah and katabatic winds.

OPEN IN VIEWER

Figure 1.

Location map of Tell Sukas bay in coastal Syria (Northern Levant) and the coring position (red point) in the Southern bay.

A strong precipitation gradient crosses the country from West to East, generating a vegetation gradient through Syria from a thermo-Mediterranean formation on the coast to semi-arid vegetation assemblages to the steppe east of the Jabal an Nuşayrīyah. Average annual precipitation ranges from 700 to 800 mm yr⁻¹ on the coast up to ca. 1400 mm yr⁻¹ in the mountains, to less than 300 mm yr⁻¹ in the arid margins eastwards from the Jabal an Nuşayrīyah and less than 100 mm yr⁻¹ in the extreme East of the country (Traboulsi, 2004). At Tell Sukas, precipitation reaches 800 mm yr⁻¹ with cold and wet winters due to the influence of the mid-latitude westerlies and hot and dry summers (with high atmospheric humidity) stemming from high-pressure systems centred over Iraq and Iran. The contrast between the rainy coast and semi-arid conditions to the East (e.g. in the arid margins) is related to the mountain range. The adiabatic gradient withholds humidity from the west and releases it as rain on the coast, creating a Foehn effect with hotter and dryer conditions to the eastern part of the mountains. Spatial and seasonal precipitation variations on the coast are controlled by temperature and pressure patterns of the Mediterranean sea surface (Sharon and Kutiel, 1986; Ziv et al., 2006). The Westerlies path and the extratropical, mid-latitude, winter cyclonic activity, which formed behind Cyprus (e.g. the Cyprus Low), account to a great deal to the annual rainfall and for major dust transportation to the Levant (Dayan et al., 2008).

Present-day vegetation

The distribution of the vegetation and associated phytogeographical belts depends on precipitation gradients and topography (Nahal, 1962; Pabot, 1957; Zohary, 1962):

Coring site and lithology

A 420 cm-long TSII core (35°18′16″N; 35°55′22″E; altitude: 1 m a.s.l.) was collected using a percussion corer in the Southern bay of Tell Sukas (Figure 1). Retrieved directly from the beach, core TSII is mainly composed of sandy clays to clayey/silty sands, with occurrences of pebbles (Figure 2). From 420 to 210 cm, the lithology is homogeneous and dominated by clays. Sediments between 210 and 160 cm are mainly composed of silty clays, with progressive occurrences of fine sandy silts and pebbles between 160 and 105 cm. At 40 cm, coarse and well-sorted sands are indicative of proximal influences and document the slowing down of the late Holocene transgression along with the establishment of a beach facies (Figure 2).

OPEN IN VIEWER

Figure 2.

Lithology and age model for core TSII based on four AMS ¹⁴C dating. The age model was built using Oxcal 4.2 (Bronk Ramsey, 2008; Bronk Ramsey and Lee, 2013). The model parameter k was set to 1 (or 100 m⁻¹). The parameter k describes the relationship between the events and the overall stratigraphical processes – a high value for k would rigidly constrain the data and would be suitable for very simple sedimentary processes with little change in the sedimentation rate, whereas a low k value (such as in this study) would be the opposite (Bronk Ramsey, 2008). The average model agreement index (A model = 103.9%) is clearly above the critical threshold of 60% (Bronk Ramsey, 2008), indicating a robust result for this modelling.

Pollen analysis

Bulk sediment samples were taken about every 10 cm in core TSII, as often as possible in clayey and silty clay layers. A total of 36 samples, each consisting of about 10 g of bulk sediment, were treated for pollen analysis using a standard palynological procedure (Faegri and Iversen, 1989). Pollen grains were counted under 400× and 1000× magnifications. Pollen and spore identification was performed using pollen grain atlases (Reille, 1999) and the palynotheca of the UMR 5554 (University of Montpellier 2). For each sample, one tablet containing a known quantity of Lycopodium spore tablets (18,580 spores ± ca. 1000 per tablet) was added as exotic markers in order to calculate pollen grain concentration (Faegri and Iversen, 1989). Pollen concentration (including unidentified and undeterminable specimens) was calculated by the addition of Lycopodium spores (Bennett and Willis, 2001) according to the following formula:

( Pollen sum of terrestrial plants × Number of L y c o p o d i u m spores in pill ) ( Counted L y c o p o d i u m spores × Volume of sample ) cm 3 ))

All taxa identified are represented in the pollen diagram (Figure 3) except the taxa occurring with less than 10 pollen grains, which have been regrouped in ‘Other Herbs’ or ‘Other Trees’. Pinaceae regrouped all damaged pollen of Abies, Cedrus and Pinus sp.

OPEN IN VIEWER

Figure 3.

Detailed pollen diagram for core TSII from Tell Sukas (Syria, Northern Levant). Pollen zones P1 to P3 were established based on a cluster analysis using the Past software. Ages are given in calibrated ages (cal. BP).

Numerical analysis

Pollen data were analysed using the Neighbour Joining (NJ) method (Figure 4). It is a method initially designed for reconstructing phylogenetic trees; it is based on the same principles as the cluster analysis method (Figure 3). NJ was computed using correlation as similarity measure and final branch as root in absolute data. Pollen types from each cluster were used to create pollen-derived vegetation patterns. ANOSIM analysis was also performed with the aim to define palynological zones (Figure 3). This method allows measuring the dispersion of samples versus the average of the group depending on the dispersion of the average between groups, which provides the significance of the clusters. A SIMPER analysis was further conducted on absolute data in order to extract the main contributor (e.g. taxa) for each group. This method then allows deciphering the most reliable taxa to infer changes in vegetation dynamics and environmental conditions.

OPEN IN VIEWER

Figure 4.

Neighbour Joining (or NJ) analysis conducted on pollen data from core TSII and computed using correlation as similarity measure and final branch as root. A1, A2 and A3 are the reconstructed clusters for pollen-derived vegetation patterns.

Age model

For sediment dating, four AMS (Accelerator Mass Spectrometry) ¹⁴C dating were obtained on core TSII (Table 1). The use of radiocarbon dates from bulk sediments should in general be avoided because of reservoir and hard-water effects (Stuiver et al., 1998). Therefore, to avoid such shortcomings, AMS radiocarbon ages were determined using four charcoals, one juvenile marine shell and one palynological residue (Table 1). For each sample, AMS ¹⁴C dating was performed at the Poznań Radiocarbon Laboratory (Poland) and the Laboratoire de Mesure du Carbone 14 (CEA, Saclay, France) using at least 0.8 mg of pure extracted carbon. Using an online calibration program (Oxcal 4.2; Bronk Ramsey and Lee, 2013), the ¹⁴C ages were calibrated to calendar years by referring to the IntCal 13 terrestrial calibration curve of Reimer et al. (2013) for charcoal and palynological residues, whereas the Marine13 calibration curve was used for marine shells with a correction of reservoir effect of 400 years, corrected for Mediterranean samples (Reimer et al., 2013) (Figure 2). Instead of following a regular calibration procedure, a Poisson sequence deposition model (P Sequence; Bronk Ramsey, 2008) was chosen to build the age model (see also the caption of Figure 2 for more details). The advantage of this calibration is that it takes depositional condition changes into consideration. As illustrated in Figure 2, the age of core TSII spans approximatively 6000–2600 cal. BP (4050–650 BCE) and thus covers the late Chalcolithic period, the entire Bronze Age and part of the Iron Age (e.g. Iron Ages I and II). For intervals without radiocarbon dates, ages were assigned by a linear interpolation based on the age depth model. According to our chronology, the average sedimentation rate is ca. 0.065 cm yr⁻¹. The average temporal resolution for the pollen analysis is estimated to 95 years per sample.

OPEN IN VIEWER

Table 1.

Radiocarbon dates for core TSII (Tell Sukas, Syria). Radiocarbon ages were converted to calibrated (cal.) ages using Oxcal 4.2 and CALIB Rev. 6.0.1. They indicate values with 2 standard deviation errors (95.4% of confidence level).

Lab. ID	Composite depth (cm)	Sampling interval	Dated material	AMS 14C age (yr BP)	Calendar year (yr BP)	Using curve
Saclay 2012
*SacA 30840	295	TSII 290–300	Marine bivalve	4230 ± 30	4322 ± 96	Marine13
Poznan 2013
*Poz-58328	185	TSII 180–190	Charcoal	2560 ± 30	2628 ± 126	Intcal13
*Poz-58330	225	TSII 220–230	Charcoal	3095 ± 30	3303 ± 75	Intcal13
Saclay 2013/2014
*SacA 35310	390	TSII 390	Palynological residue	5035 ± 35	5781 ± 117	Intcal13

AMS: Accelerator Mass Spectrometry.

Pollen grain analysis

Pollen grains are abundant in core TSII, and pollen preservation is fairly reasonable in all the 36 samples analysed. Neither evidence of disturbed sediments nor dramatically corroded pollen grains were reported in core TSII. Pollen grains were generally well preserved in the grey silty clays typical of a shallow marine sedimentation. Conversely, pollen grains were rarely preserved in topmost sediments (e.g. from 161 cm onwards) because of oxidation in sand and beach ridge deposits; only two samples (83 and 79 cm) have delivered pollen grains for identification and counting. Dinoflagellate cysts and foraminifera linings were encountered in nearly every sample analysed, but were not investigated in this study. In total, 81 pollen taxa were identified. A total of 321 pollen grains were counted in average, except for a few samples in which pollen abundances were too low (>200 pollen grains, for 8 samples only). Pollen concentration calculated using Lycopodium spore tablets is high and generally stable: it varies from 8680 to 136,950 grains g⁻¹ (48,811 grains g⁻¹ on average; Figure 3) with noticeable peaks above 100,000 grains g⁻¹ in zone P2 at 263 and 281 cm. Three ecostratigraphic pollen zones (P1 to P3) have been distinguished based on the major changes in pollen assemblages. Results are shown in a detailed pollen diagram (Figure 3), in which pollen taxa are depicted based on their ecological preferences. These ecostratigraphic pollen zones were further validated with a cluster analysis by applying the Bray–Curtis similarity measure (Faith et al., 1987) in paired group algorithm, using the ‘Past’ software. Ecostratigraphical zones P1, P2 and P3 are significant judging from an ANOSIM analysis (R = 0.841 and p < 0.001). A SIMPER analysis revealed that four taxa explain 50% of the total variance between P1, P2 and P3: Olea sp., Poaceae, Chenopodiaceae and Pinus. However, 80% of the total variance is explained when considering the following taxa: Asteraceae Cichorioideae (or Liguliflorae), Pinaceae, Asteraceae Asteroideae (or Tubuliflorae), Apiaceae, cereals, Quercus deciduous, Peganum harmala and Fabaceae.

Numerical results

Three main pollen-derived vegetation patterns were discriminated using numerical analyses: meadow and degraded environment, along with the cereal corpus (A1); olive tree corpus (A2); and Mediterranean woodlands and coastal steppe (A3). Subdivision into groups A1, A2 and A3 is reported with a strong confidence index (ANOSIM, R = 0.624, p < 0.001). In these three significant groups, subgroups can be deciphered using numerical analysis and ecological knowledge (Mouterde, 1983) (Figure 4).

Vegetation dynamics derived from the pollen record

Three ecostratigraphic pollen zones (P1, P2, P3) were depicted in the pollen diagram (Figure 3); palaeoecological association was mainly built based on numerical analyses. The palynological analysis documents vegetation history and dynamics in Northern Levant between ca. 6000 and 2600 cal. BP (ca. 4050–650 BCE), from the Chalcolithic to the Iron Age. Vegetation dynamics are discussed in terms of contrasting environmental states, in which major fluctuations of anthropogenic indicators versus natural Mediterranean vegetation can be detected in the pollen diagram. Periods evidencing a pronounced human impact (e.g. forest clearing, crop cultivation and/or occurrences of ruderal plants such as Plantago lanceolata) are depicted by high abundances of anthropogenic indicators on the natural vegetation, whereas periods of low human activity are characterized by an expansion of a well-developed Mediterranean forest/maquis in the area. It is worth quoting that the influence of local sedimentary processes on the pollinic signal can be neglected since no change of the river trajectory (which would stem from changing coastal morphology during the last 6000 years) was evidenced on the field. Hence, the deposition of fluvial-transported pollen in the bay of Tell Sukas reflects primarily the vegetation assemblages in the drainage basin, that is, from the Western part of the Jabal an Nuşayrīyah.

Between 4600 and 3600 cal. BP (2650–1650 BCE): the most important phase of olive tree cultivation during the EBA III–IV and Middle Bronze Age

Core TSII from Tell Sukas reports the most important increase in the Olea pollen curve at ca. 4600 cal. BP (2650 BCE), suggesting a marked rise of olive cultivation coevally to higher regional precipitation based on plant fragments from Tell Breda (Fiorentino et al., 2008; Figure 5c). Olive exploitation flourished in the following centuries, even peaking at very high levels (e.g. 60%) between 3900 and 3600 cal. BP (1950–1650 BCE) indicating that olive oil production attained its apogee during the middle Bronze Age. We propose to link the development of olive cultivation with the ‘second urban revolution’ (EBA III) (Akkermans and Schwartz, 2003), which corresponds to the full-fledged adoption of urban life in Northern Levant (Al-Maqdissi, 2010, 2012). Hence, core TSII from Northern Levant accurately documents the onset of olive exploitation in the middle of the third millennium BC already known in Southern Levant where early olive cultivation began around 6300 BP. In Southern Levant, the development of olive exploitation was associated with early urbanization processes during EBA I (Langgut et al., in press). Therefore, the increase in Olea pollen in core TSII documents the probable development of a specialized economy devoted to olive orchards and secondary products (olive oil). The specificity of Tell Sukas, as a coastal location receiving maritime humidity in Northern Levant, combined with the great agricultural potential of the Jebleh plain, may have stimulated the production of olive oil in coastal Syria. Reports of EBA olive stone remains at Tell Sukas (Helbaek, 1962) and the excavation of an olive press more to the North on the tell of Ras Shamra (Courtois, 1962) further corroborate palynological evidences which argue for olive tree cultivation. However, unlike Inner Syria (texts from Ebla: Archi, 1991), no text documented olive tree cultivation and olive oil production on the Syrian coast during the third millennium BC. It has been suggested that the Levantine ‘Combed Ware’ jars, classical marker of the EBA ceramic, were used for the storage of olive oil, even perhaps for its transportation (Mazzoni, 2002) although this was not confirmed by identification of olive and oil residue in jars. ‘Combed Ware’ was discovered on the North Levantine coast, at Shamiyeh (Al-Maqdissi and Ishaq, in press), Ras Shamra (Courtois, 1962 and unpublished fragments), Tweini (Vansteenhuyse, 2008), Sianu (Bounni and Al-Maqdissi, 1994) and Sukas (Oldenburg, 1991) and well-attested in the northern part of Central Levant, at Arqa (Thalmann, 2006), Fadous-Kfarabida and Byblos (Sowada, 2009; Thalmann and Sowada, 2014). In the Jebleh plain, numerous discoveries of ‘Combed Ware’ jars at Sianu (10 km from Tell Sukas) inferred that this city was probably a trade centre for olive oil production during the EBA III (Bounni and Al-Maqdissi, 1994). The production of olive oil was devoted to local consumption and possibly to trade exchanges. Although this hypothesis has not been confirmed by archaeological or textual data yet, two issues must be considered: (1) during the second part of the third millennium BC, Southern Levant experienced important socio-political changes (e.g. Finkelstein and Langgut, 2014) while (2) at the same time, maritime exchanges developed in Eastern Mediterranean and contacts reinforced between Egypt and Central Levant (Byblos) (Genz, in press; Grimal and Francis-Allouche, 2012). ‘Combed Ware’ jars represented the largest quantity of imported ceramics for the Old Kingdom (Sowada, 2009). Archaeometric studies have shown that some ceramics may have been produced in Byblos, while one vessel may originate from Northern Syria/Amuq (Sowada, 2009; Thalmann and Sowada, 2014: 372). Hence, it is all the more likely that Northern Levant benefited from this economic and cultural impetus and participated in the trade network. But yet, unlike Ebla in Central Syria (Scandone-Matthiae, 1997), no aegyptia dated to the third millennium BC have been discovered in secure EBA III levels on coastal Syrian sites (Matoïan, 2015).

OPEN IN VIEWER

Figure 5.

Palaeoenvironmental records from Northern Levant and the Eastern Mediterranean: (a) pollen-based data from Tell Sukas (coastal Syria) in Northern Levant (core TSII; this study): Olea, Chenopodiaceae, Poaceae pollen data and the arboreal pollen (or AP) abundances; (b) Olea, Chenopodiaceae, Poaceae pollen data at Tell Tweini (coastal Syria; Kaniewski et al., 2010); (c) precipitation reconstruction based on δ¹³C of plant fragments from Tell Breda (Syria; Fiorentino et al., 2008); (d) the δ¹⁸O record from Jeita Cave (Lebanon; Verheyden et al., 2008); (e) Olea pollen content from the Dead Sea (Litt et al., 2012); (f) Olea pollen content from Lake Hula (Israel; Van Zeist et al., 2009). BC (Before Christ), BP (Before Present) and the archaeological period with EBA (Early Bronze Age), MBA (Middle Bronze Age) and LBA (Late Bronze Age). Grey shadings correspond to the three main periods discussed in this study, namely, the onset of olive cultivation at Tell Sukas, the end of the early Bronze Age, and the end of the (late) Bronze Age period.

Around 4200 cal. BP (2250 BCE), urban disintegration, changes in agricultural and cultural practices became widespread in many places from the Near East (Akkermans and Schwartz, 2003; Weiss et al., 1993; updated in Weiss, 2014), which corresponds to the collapse of the Akkadian empire (e.g. Kaniewski et al., 2008), the end of the Old Kingdom in Egypt (Stanley et al., 2003) and the abandonment of cities in western inland Syria (Schwartz et al., 2000). At the same time, a large-scale climatic event has been observed in the North Atlantic. The 4200 cal. BP (i.e. 4200–3800 BP) event (Bond et al., 1997, 2001) was documented worldwide in climatic archives (Mayewski et al., 2004; Sorrel et al., 2012). At low-latitude settings (such as in the Eastern Mediterranean), the 4200–3800 cal. BP interval corresponds to a notable increase in regional aridity (Bar-Matthews and Ayalon, 2011; Eastwood et al., 2007; Finné et al., 2011; Magny et al., 2013). In Southern Levant, Finkelstein and Langgut (2014) identified an interval with more arid conditions at the end of the early Bronze (~4000–3800 BP) in the Sea of Galilee, which is coeval to lower Dead Sea lake levels (Bookman et al., 2004; Kushnir and Stein, 2010; Migowski et al., 2006). Modelled precipitation estimation from Tell Breda (Syria) also inferred a decrease in regional precipitation (Fiorentino et al., 2008), whereas a prominent dryness beginning at 4200 cal. BP was reported from the Soreq cave record (Bar-Matthews and Ayalon, 2011). However, not far from Tell Sukas, the speleothem record of Jeita cave (Lebanon, Central Levant) does not show compelling evidence for a rapid climate shift around 4200 yr BP (Verheyden et al., 2008). Judging from pollen data in core TSII, the period centred around 4200 cal. BP is tied to a noticeable decrease in the Olea pollen curve, when drier conditions are reported at a large-scale in the arid part of the Levant. However, the effect of enhanced dryness on olive horticulture is not dramatic since Olea abundances remain at fairly high levels (e.g. 26%), therefore suggesting that olive exploitation did not drastically shrink during the 4200 cal. BP climatic event. Other palynological data tend to confirm the importance of olive trees in coastal Levant. Indeed, at Tell Tweini, a pollen grain study documented high Olea pollen content around 4100 cal. BP (Kaniewski et al., 2008; Figure 5). In addition, no dramatic event is reported from archaeological sites. We are therefore inclined to suggest that no ‘collapse event’ happened in the coastal part of Northern Levant around 4200 cal. BP. This is in accordance with the conclusions of Finkelstein and Langgut (2014), who suggested a shift to the North of the boundary between Mediterranean and Irano-Turanian vegetation zones at that time, thereby forcing numbers of people to move to the ‘greener’ parts of the Levant (e.g. the Northern Levant plains, including the Jebleh plain).

The late Bronze (ca. 3600–3200 cal. BP; 1650–1250 BCE) and the Iron Age (ca. 3100 to ca. 2600 cal. BP; 1150–650 BCE)

The LBA (from ca. 3600 to ca. 3200 cal. BP; 1650–1250 BCE) is marked by a spectacular decrease (from 60% to 5%, albeit in two steps: 3600–3500 cal. BP, then 3450–3200 cal. BP) in olive tree cultivation in Northern Syria around coastal Tell Sukas (Figures 3 and 6). Sukas (Suksi) is now mentioned in the texts of Ugarit as a town located near the southern border of the Kingdom of Ugarit (Lund, 2004; Van Soldt, 2005). Ugarit provides the main archaeological and textual documentation concerning olive horticulture in LBA Northern Levant. Olive press installations were discovered in different areas on the tell of Ras Shamra (Callot, 1994) and texts (13th and beginning of the 12th) documented olive orchards, production of olive oil in large quantities, as well as distribution and trade of olive oil (Heltzer, 1999: 446, 450). Olive horticulture thus represented the most important tree crop besides vineyards in the Kingdom of Ugarit. Hence, the end of the LBA and the transition with the Iron Age likely correspond to the most severe, abrupt decline of Olea pollen cultivation (along with a strong decrease in Mediterranean trees and an almost total absence of pollen grains in sediments) between 3250 and 3100 cal. BP (ca. 1300–1150 BCE), although the decline of olive exploitation already started at ca. 3450 cal. BP (1500 BCE). This relatively prolonged interval is, therefore, the most pronounced dry episode during the Bronze and Iron Ages in Northern Levant. A similar decrease – or persistently low values of Olea pollen (e.g. Langgut et al., 2013; Van Zeist et al., 2009) – in olive horticulture was reported in Southern Levant (Langgut et al., 2014; Litt et al., 2012; Neumann et al., 2007a) and in the Golan Heights (Neumann et al., 2007b). A dry episode during the LBA–Iron Age transition was well documented elsewhere in the Middle East (e.g. Carpenter, 1966; Ward and Joukowsky, 1992; Weiss, 1982), especially in palynological records from the Nile Delta (Bernhardt et al., 2012), the Dead Sea (Kushnir and Stein, 2010; Langgut et al., 2014; Litt et al., 2012; Migowski et al., 2006; Neumann et al., 2007a), the Sea of Galilee (Langgut et al., 2013) and Lake Hula (Van Zeist et al., 2009) in Southern Levant and from coastal Syria (Kaniewski et al., 2008, 2010) and Cyprus (Kaniewski et al., 2013), while harsher conditions in Turkey (Bottema et al., 1994; Woldring and Bottema, 2003) further suggested a dry spell spreading over a large geographical area. A former study that focused on the region of Tell Tweini, near Sukas (Kaniewski et al., 2008), concluded that a complex shift occurred from an open deciduous forest towards an arid/saline-tolerant vegetation at ca. 3175 cal. BP (1225 BCE). In western steppic Syria, an adverse drastic reduction in rainfall at ca. 3200 cal. BP (1250 BCE) was inferred at Tell Breda, concurrently to the precipitation trend proposed by Bryson and Bryson (1997) near Tell Leilan (eastern Syria). Other studies from the Eastern Mediterranean have documented a drier trend during the LBA–Iron Age transition (e.g. Drake, 2012; Roberts et al., 2011; Schilman et al., 2002). Modelled precipitation based on plant δ¹³C from Tell Breda (Fiorentino et al., 2008) and the δ¹⁸O speleothem record from Jeita cave (Verheyden et al., 2008; Figure 5d) has proposed that a dry trend started at ca. 3000 cal. BP. Coevally, lower rainfall is also inferred in carbonate deposits from Soreq cave (Bar-Matthews et al., 2003).

OPEN IN VIEWER

Figure 6.

Map showing the archaeological sites mentioned in the text for Northern Levant and the palaeoenvironmental archives used in Figure 5 (e: Dead Sea and f: Lake Hula are outside of the map).

Here, we question to which extent the drier period widely recognized during the late Bronze–Iron Age transition could be related to the Near Eastern textual documentation dated to the 13th and 12th century BCE, which provides a vivid record of the desperate pleas for food in the Hittite empire (Bryce, 2005; Singer, 1999) and the Levant (Singer, 1999). The archives discovered in the so-called houses of Rapanu and Urtenu (Ras Shamra – Ugarit) are of special interest in this regard. Letters from the Hittite court, Pharao or the king of Qadesh documented demands for great quantities of grain and even famine (see RS 20.212, RS 34.152, RS 94.2002+, RS 94.2287; Lackenbacher, 2002; Lackenbacher and Malbran-Labat, in press; Singer, 1999). Moreover, as previous studies underlined, the question arises regarding a link between this dramatic climatic event and the period of turmoil and crisis of the end of LBA in the Eastern Mediterranean. Significant Aegean, Cypriot, Anatolian and Levantine centres (Figure 6), among which Ras Shamra (Yon, 1992, 2006), Ras el Bassit (Courbin, 1986), Ras Ibn Hani (Bounni et al., 1998), Tell Tweini (Bretschneider and Van Lerberghe, 2008), Tell Sukas (Lund, 2004; Riis et al., 1996), Tell Sianu (Al-Maqdissi, 2006) or Tell Kazel (Badre, 2006) suffered major destructions and the Hittite empire collapsed. However, the causes are still largely debated. Among other mechanisms, migrations, invasions of the Sea People, economic and social disintegrations or even environmental disasters, such as a dry event or earthquakes (Kagan et al., 2011; Nur and Cline, 2000; Oren, 2000; Ward and Joukowsky, 1992; Yasur-Landau, 2010), have been considered as proximate causes for the LBA collapse. On one hand, palynological data from core TSII corroborate the dry event occurring at the transition between late Bronze and Iron Age, with a dramatic shift in vegetation assemblages in coastal Northern Syria. At Tell Sukas, minimum values of Olea pollen were recorded around 3100 cal. BP (1150 BCE), thus inferring a weakening of olive oil production. On the other hand, the decline of the Olea pollen curve at Tell Sukas occurred around 3450 cal. BP (even at ca. 3600 cal. BP) at the time resolution used here, that is, long before the LBA collapse. Therefore, other factors (not documented in texts) must be considered for the decline of olive oil production as soon as about 3450 cal. BP (1500 BCE) in coastal Levant. Note also the increase in Tubuliflorae pollen abundances and low occurrences of Cyperaceae between ca. 3400 and 3150 cal. BP (1450–1200 BCE), all of which being indicative of a prolonged dry interval.

Tell Sukas was reoccupied at the beginning of the Iron Age (Riis et al., 1996). From ca. 3100 to ca. 2600 cal. BP (1150–650 BCE), pollen data from Tell Sukas suggest a climatic resumption with higher available moisture judging from increased values of oaks and cultivated olives and relative high abundances of Poaceae and Cyperaceae (Figure 3). Greater available moisture during the Iron Age was also reported from the Dead Sea pollen record in Southern Levant (Langgut et al., 2014; Litt et al., 2012; Neumann et al., 2007a) and from the Soreq cave (Bar-Matthews and Ayalon, 2011), probably because of increased precipitation in the entire Levant. Additionally, in core TSII, maximum values were recorded for cereals along with the spread of non-palatable herbs (e.g. the increase in Liguliflorae and Tubuliflorae), suggesting the onset of cereal agriculture from ca. 3100 cal. BP (1150 BCE) onwards, as it was also recognized in Southern Levant (Cordova, 2010; Litt et al., 2012). A slight increase in Olea pollen at 2600 cal. BP thus suggests a recovery of olive tree cultivation (e.g. Figure 3), even if our sampling resolution is too fragmentary in this interval. Nevertheless, olive orchards continued to be exploited after the long-term drought of the LBA–Iron Age transition, although the relatively low olive values are likely indicative that horticulture activity was primarily devoted to personal consumption, rather than for commercial purposes. Forthcoming analyses would be needed to help refine the history of olive cultivation on the Syrian coast from the Iron Age onwards (i.e. the time interval covering the Hellenistic, Roman, Byzantine and early Arab periods).