Introduction

Much recent scholarship has focused on identifying apparent periodic episodes of substantial and longer-lasting drought–so-called megadroughts–which arguably appear to correlate, at least approximately in temporal terms, with episodes of major collapse, change, or reorientation in human civilization in various regions of the world [1–3]. Whether mere approximate contemporaneity is indicative of explanation is another question, and recent discussions of social responses to climatic change have highlighted the unpredictability and variability of possible outcomes, precluding a simplistic association between environmental change and societal collapse [4–11]. Nonetheless, chronological resolution is an essential component of consideration of any possible relationships between environmental change and social response [12], and the absence of an appropriate close temporal association undermines many purported causal connections (e.g. [9, 13–15]). The potential importance of such historically relevant climate forcing episodes, both in explaining history, but also as laboratory of current and future relevance, should not be overlooked–and this general topic has been the subject of much recent research, as in other regions, in the Mediterranean-Near East (e.g. [1–3, 16–23]), but also debate (e.g. [4–11, 24]). In several instances, the exact associations and causal connections between observed proxy evidence for climate change and human history have been demonstrated to be both complex and multi-faceted, and often not as clear in temporal or causal terms as sometimes stated by proponents of one or other specific viewpoint (e.g. [9, 14, 15]).

Two multi-century climate change episodes with special relevance to the Levant/East Mediterranean region are regularly cited and stated as occurring from (i) ca. 4200 BP/2200 BCE and (ii) 3200 BP/1200 BCE (BP here from 2000 CE) (e.g. [16–23, 25, 26]). In each instance, there is purportedly a multi-century episode of collapse, then general hiatus and transformation before a subsequent return to ‘high’ civilization—the Middle Bronze Age (the Middle Kingdom in Egypt, the Old Babylonian period in Mesopotamia) from around 2000/1900 BCE for the first, and the Iron Age ‘renaissance’ from the 9-8^th centuries BCE for the second (the latter perhaps linked to a major solar irradiance minimum and climate shift [3, 27, 28, ref. 29 pp. 112–115]).

The 2200 BCE episode in the later Early Bronze Age (hereafter EB) is well-dated, both in terms of some wider proxies indicating a shift to cooler and more arid conditions, and in terms of the demise and abandonment of the formerly primate site of Tell Leilan in northern Syria just before 2200 BCE [25, 30, 31]. However, the impact of this climate event is more difficult to reconstruct in the Levant and tightly resolved higher frequency records are sparse. Recent syntheses of radiocarbon (¹⁴C) evidence have decoupled the chronology of this event from the long-recognized decline in settlement during the EB IV in the southern Levant, which is now recognized to have begun several centuries earlier, ca. 2500 BCE [32, 33]. It is also increasingly evident that sites in the northern Levantine region, which were in less fragile dry-farming loci with significant access to reliable water resources in the form of karstic aquifers, were variably affected and not all collapsed [25, 34, 35]. Weiss indeed argues that the Euphrates and especially Orontes river valleys, and the area of ‘karstic’ springs, formed refugia during the 2200 BCE episode with growth supplemented by habitat-tracking out of other less resilient regions of northern Mesopotamia/Syria [25, 35]. He points to the take-off of Tell Tayinat as an example of this [35]. While the variable ecological context is undoubtedly relevant, other factors also appear to be in play, and not just survival in the face of climate crisis, including the re-orientation and transformation of interregional trade and local economies, and of the relevant trade routes. In particular, these processes saw Anatolia and the East Mediterranean (and so maritime connectivities, versus, or in addition to, land routes) become more important foci for the Levant in the later/late 3^rd millennium BCE [36, 37]. In fact, contrary to the pattern observed in areas of northern Mesopotamia, the northern Levant witnesses a persistent tradition of urbanism continuing through and following the period at 2200 BCE [34, 38, 39]. As such, this region offers a key venue for observing alternative trajectories and responses to the 4.2k climate episode.

The 1200 BCE episode, around the close of the Late Bronze Age, involves a long-recognized widespread set of site abandonments, destructions and changes across the East Mediterranean and Levantine region, the collapse of the previous ‘palatial’ economies and inter-linked trading systems and an ensuing ‘dark age’ [11, 15, 24, 40–42]. Associated proxy climate evidence indicates another shift around this period to cooler and more arid conditions [20–23, 26, 43]. The widespread collapse of the internationalized political economies of the Late Bronze Age is real, and many sites do experience destruction, decline, or abandonment. But the collapse of the region is not total. It is increasingly clear that this episode is as much a reorientation and shift to differing social and economic models as simple collapse [11, 36], a reorganization that represents a key formative period that shapes the later developments of the Iron Age. The climate proxy data, while relatively consistent in indicating cooler, more arid conditions, are as yet poorly constrained in temporal terms in many cases, and exact associations with cultural and political developments therefore remain problematic [15].

In each of these cases there remains much to be investigated and learnt in order to be able to write a detailed history successfully and convincingly linking archaeology-history and climate. However, even more pressing is the question: what happened around and particularly after these apparent episodes of collapse/change, during the periods of ensuing transformation prior to the generally accepted return of ‘high’ civilization? Despite terms like megadrought and collapse, it is of course widely understood that archaeological and historical evidence continues through these episodes, at least in several areas. We consider here the northern Levantine Early Bronze Age IV as conventionally ca. 2500–2000 BCE (EB IVA ca. 2500–2300; EB IVB ca. 2300–2000 BCE) and the early Iron Age as the period during and after the 12^th century BCE in the northern Levant (ca. 1200–900 BCE, Iron Age I). On the principle that both humans and nature abhor a vacuum, we may also assume a corollary if there is a climate association with the changes occurring ca. 2200 BCE and 1200 BCE (or even if there is simply a key structural change in social and economic systems). Thus, if such climate (or other) changes negatively affect one existing set of sites and their environmental and locational contexts, it is also likely that other loci may find an opportunity to fill this void and to offer ‘alternative’ trajectories and histories. These cases might involve societies centered on a different, more resilient, economic base–like pastoralism in the Levant [4, 44], societies constructed on differing socio-political scales (e.g. smaller units), based in less affected geographic contexts, or with access to alternative trade networks [36, 41]. In other words, we open perspective towards both resilience and alternatives to simple collapse, and more nuanced approaches that incorporate a heterogeneity of response and opportunity at local level in the face of regional-global climate processes.

The large and important site of Tell Tayinat in southeast Turkey offers one such alternative history [45–51]. In this paper, we use radiocarbon dates and analysis of these to investigate and define the chronology of the site of Tell Tayinat, and demonstrate that its substantive history of occupation(s) includes the periods of response to the ‘megadrought’ eras associated with both the 2200 BCE and 1200 BCE episodes. The site is largely not occupied during the ‘high-civilization’ periods of the Middle and Late Bronze Age. Instead, the local regional occupation in these periods is at Tell Atchana (ancient Alalakh), and the switch to Atchana from Tayinat at the end of the Early Bronze Age, followed by a return to Tayinat at the end of the Late Bronze Age, forms a local version of the wider ‘alternative’ paradigm of interest [45–51]. Tell Tayinat thus represents an alternative history and trajectory to the ‘palatial’ mainstream of the Levantine region, and a prime case study illustrating how change—whether forced by economics, climate, geography or other factors—is rarely universal. Instead, negative or positive factors affecting one set of circumstances may well have differing and even opposite effects elsewhere. The context and societal formations at Tell Tayinat—and the contrast, locally, with Tell Atchana, and more widely with the major Middle–Late Bronze Age primate sites across the east Mediterranean and Levant—offer us one rich window into a resilient and successful response, and alternative to collapse, in the face of regional megadrought in the Old World.

Tell Tayinat

Tell Tayinat, located in the Amuq Plain of southeastern Turkey and centrally positioned in the northern Levantine region, comprises a large, low-lying mound within the flood plain of the Orontes River. Tayinat sits on the northern bend of the Orontes, downstream of its entrance into the Amuq Plain, at the point where it turns westward toward Antakya (ancient Antioch) and the Mediterranean Sea (Fig 1). The site has long been a focus of excavation and study, starting with large-scale excavations conducted by the Oriental Institute of the University of Chicago between 1935 and 1938 [52, 53]. More recently, it has been the focus of work by the Tayinat Archaeological Project (TAP) based at the University of Toronto, which began with pedestrian and geomagnetic survey between 1999–2002 [54], further geomagnetic survey in 2003 (http://sites.utoronto.ca/tap/assets/2003geomagneticsurvey_en.pdf), followed by excavations and study from 2004 onwards [45–51, 55, 56]. The earlier excavations demonstrated that the site was occupied during significant portions of the 3^rd millennium BCE, from at least the EB III [52]. To date, the new excavations at the site have produced materials and contexts dating from the later part of the Early Bronze Age (EB IVA-B, late 3^rd millennium BCE) and spanning the Iron Age, with significant attention paid to the monumental structures of the Iron II and Iron III levels [53, 57, 58].

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 1. Map showing the Orontes Valley of Northwest Syria and Southeast Anatolia and all the sites discussed in this study.

This map was produced in ArcGIS v 10.7.1 at the CRANE funded Archaeology Centre Digital Innovation Laboratory of the University of Toronto by compiling GIS Shapefiles and Digital Elevation Data built from publicly available sources including NASA/JPL/NIMA.

https://doi.org/10.1371/journal.pone.0240799.g001

Historically, the site is also well-attested. The Early Bronze Age city is perhaps the location referred to as Alalaḫu in records preserved on clay tablets from the late 3^rd millennium BCE archive found at the site of Ebla in Syria [49, 50, 59]. For the Iron Age, interpretation of inscriptional evidence has led to suggestions that by at least the 11^th century BCE, early Iron Age Tell Tayinat represented the center of a kingdom named Palastin or Walastin, with one king being a certain Taita, “Hero and King of Palastin”, best known from inscriptions in the Aleppo Temple [60–63]. Later historical records, dating to the 9^th-early 8^th centuries BCE, refer to the kingdom as Patina or Unqi [64–66]. The Assyrians took control of the city in 738 BCE and the region became part of their province of Kinalia under an Assyrian governor [65–69].

The end of the first primary period of occupation at Tell Tayinat in the late 3^rd millennium BCE corresponds in general with the period of change and re-orientation across much of the Old World from around 4200 years BP or 2200 BCE [16–19, 25]. In the northern Levant, this episode is in general associated with the EB IVB period. The overall dates for the wider period of societal change, commonly associated with a shift(s) in climate affecting especially the Old World region, are largely uncontested—although the exact timings, scale, extent, coherence and uniformity of the climate episode remains debated even in the Mediterranean-Levant region (e.g. [70–73]). Correlating with the end of the Akkadian Empire and the First Intermediate Period in Egypt, a beginning for this period of change in the later 23^rd century BCE and especially around 2200 BCE is largely agreed. It is generally accepted that there is a period(s) of apparently cooler and drier conditions in this region that lasts until around the 20^th century BCE (ending broadly by around 1900 BCE) [16–19, 25, 30, 31, 70–74, 77, 78]. The area for debate is exactly whether, how, and how comprehensively, the latter causes the former [4, 7–8, 16, 19, 25, 31, 35, 75, 76]. The site of Tell Leilan in northeastern Syria offers a closely resolved date for the beginning of this period [30, 31], as do a variety of proxy climate indicators [25, 70–74, 77, 78]. Its end is marked by the re-emergence of wider trade networks and state-polities (like the Middle Kingdom of Egypt, Old Babylonian Period in Mesopotamia) during the 20^th century BCE.

The issue for Tell Tayinat is to position the site and its EB occupation in terms of this wider general timeframe. In contrast to regions such as northern Mesopotamia, whose chronologies have been more closely tied to the 4.2k event, and its apparent historical effects, northern Levantine chronologies generally have not accounted for this event. The general chronological framework for the northern Levant, largely based on sequences from Hama, Ebla and the early Amuq excavations [79], is generally agreed upon, with the EB IVB period conventionally dated to ca. 2300–2000 BCE. More recently, the creation of an EBA-MBA transitional subdivision, ca. 2100–2000 BCE, has been proposed [34, 80, 81]. This chronological framework, however, has been developed primarily from relative ceramic sequences and has not been anchored to ¹⁴C dates [34, 79, 82]. An independent and direct timeframe is thus lacking. More recently this situation has started to change. ¹⁴C dates pertaining to several late 3^rd millennium BCE sites have been published, including Tell Mardikh/Ebla [83], Rawda [34, 84], Qatna [85] and Umm el-Marra [34]. The suggested date for the destruction of Palace G at Ebla, representing the local transition from the EB IVA to EB IVB, is suggested to date to 2367–2293 BCE, with ca. 54% probability [83], roughly in line with the conventional date for this transition, and placing the 4.2ka (2200 BCE) climate event firmly during the EB IVB period.

The Early Iron Age is more problematic. There has been a long-standing challenge resolving Iron Age chronology in the northern Levant in this period [51, 56, 86, 87]. The close of the Late Bronze Age is characterized by the collapse of the wide-reaching and literate trading systems of previous centuries [15, 40–42, 88–90]. We also lose the well-defined sequence of historical rulers from the unified New Kingdom of Egypt. The result is that archaeologists lack the confident ability to employ replicated material culture associations to tie object types, assemblages, and sites across the eastern Mediterranean to the historical chronology of Egypt. This situation has led to much scholarly uncertainty and debate over dates based on relatively scarce, ambiguous, or contradictory evidence (e.g. [56, 90]). The past couple of decades have, unsurprisingly, seen scholars trying to find alternative means to establish secure chronological timeframes for the early Iron Age in several areas of the East Mediterranean and Aegean, and have–almost inevitably–created controversy as previous hypotheses are challenged, usually by new radiocarbon evidence [86, 87, 91–97].

In the northern Levant, especially, there has been lack of both data and scholarly focus. After the collapse and transformation of the well-known Late Bronze Age civilizations and their major, often ‘palatial’, centers around the close of the 13^th century BCE and into the early 12^th century BCE [11, 15, 42, 88, 89, 90], the social, political and economic processes of the subsequent earlier Iron Age are much less well understood [51]. This is particularly true in the period stretching from the late 12^th through 10^th centuries BCE, where historical documentation has been scarce. Although new inscriptional evidence is beginning to provide a historical framework for this period, the precise chronology of these historical developments remains fluid, uncertain, and largely based on paleographical grounds as newly emerging finds regularly require the revision of historical chronologies and king lists [60–62, 98, 99]. In the 9^th–8^th centuries BCE, historical documentation becomes more frequent from Neo-Assyrian records, as a result of their increasing contacts with this region, and then their takeover and administrative control from the late 8^th century BCE onward [65–68, 100]. Following the conquest of Kunulua–the ancient name in this period for Tell Tayinat–by the Assyrian ruler Tiglath-pileser III in 738 BCE, Tayinat became the capital of the Assyrian province of Kinalia [58, 64, 65, 101]. The available textual evidence suggests that in the intervening period, the larger political structures of the Late Bronze Age had vanished and in their place was a collection of relatively small territorial states [45, 46, 51, 55, 100–103]. The trajectories and chronologies of these profound changes and of the emergence of these new socio-political formations remains largely unclear. Hence establishing a defined and robust chronology for a key site like Tell Tayinat represents a critical step towards a more nuanced understanding of this time period.

In contrast to the southern Levant and a few other areas, where the widespread use of radiocarbon evidence to address issues of Iron Age chronology has led to the development of a large dataset of chronometric evidence (e.g. [87, 91–97]), available radiocarbon data in the northern Levant is extremely scarce [15, 23, 51, 104]. This has led to the development of a series of relative regional chronologies, often employing widely varying chronological terminologies based on locally-defined or somewhat arbitrary subdivisions [86, 105–108]. These are often based on linkages in material culture and styles within the Levant and with the Aegean and Cyprus [56, 109]. As a result, the issue of chronology has been problematic and largely based on circular reasoning and imprecise criteria.

One of the key discoveries during recent excavations at Tell Tayinat is the articulation of a sequence of early Iron Age remains, after a settlement hiatus during the Middle and Late Bronze Ages, that were unattested in the earlier excavations at the site [45–47, 51, 55, 110]. With an area from ca. 12–20 hectares in extent, Tell Tayinat is one of the larger early Iron Age sites known in the eastern Mediterranean, and represents an important regional sequence for the early Iron Age in the northern Levant. As a result, we report work aimed at remedying the lack of chronological data for this region through establishing a high-resolution calendar timescale for early Iron Age Tell Tayinat. The use of data from Tell Tayinat for the development of a regional sequence is particularly appropriate, given the important role the site played in the early development of radiocarbon dating. A sample of charcoal from Tell Tayinat in fact appeared on Libby’s famous ‘curve of knowns’ as a measurement by the University of Chicago. Libby wrote:

“The next sample, which is marked “Tayinat,” is from a house in Asia Minor which was burned in 675 B.C. It was wood from the floor of a central room in a large hilani (“palace”) of the “Syro-Hittite” period in the city of Tayinat in northwest Persia. Its known age is 2625 ± 50 years” [111].

This statement is historic, but also both remarkable and (retrospectively) distressing. A charcoal sample of the scale required at this time for radiocarbon dating likely had a considerable number of tree-rings. Thus it could quite well today have offered a potentially important dendrochronological sample, or least a highly resolved dendro-¹⁴C-wiggle-match sample. Such a sample could have enabled precise dating (as undertaken at some other Anatolian Bronze and Iron Age cases [78, 112–116]) of some considerable value, perhaps tied to a specific construction episode at the site (its context from the floor in a central room suggests a fallen roof beam). Sadly, no charcoal recovered so far in the recent Tayinat excavations has been of anywhere near such a scale for dendrochronological analysis, and the whereabouts of Libby’s original sample or any others that may have been collected in the 1930s remains unknown. The confidence expressed in the dating of the context and sample from the site is also striking, and almost certainly misplaced, given the complex history of the specific bīt-hilāni palace from which the sample was taken (Rooms I-J, 1^st floor in Building I) (e.g., [53, 57, 117, 118]).

In view of these research opportunities and the current limitations in terms of dating, our project therefore seeks to integrate the archaeology with the radiocarbon evidence to achieve a secure timeframe for Tell Tayinat in both the late Early Bronze Age and the Early Iron Age. We use organic samples (identified wood charcoal and especially short-lived seed material) from well-defined archaeological contexts at Tell Tayinat for a program of radiocarbon dating. In particular, we have carefully selected sets of short-lived sample material where possible–offering ages contemporary with use–from stratigraphic sequences. These circumstances, where we have an archaeologically-ordered sequence of contexts and samples, allows in addition the application of Bayesian chronological modeling approaches [119–123], where prior archaeological or historical knowledge can be integrated with the radiocarbon probabilities, in order to achieve more precise and more robust chronologies.

Materials and methods

Samples and ¹⁴C dates

We reviewed and identified the available organic sample materials from the Tell Tayinat Archaeological Project (https://tayinat.artsci.utoronto.ca/) and its excavation program. The Tayinat landowners, in particular the Kuseyri family, permitted work on their land. All necessary permits were obtained for the described study, which complied with all relevant regulations. The Directorate of Cultural Heritage and Museums of Turkey granted the research permits necessary to conduct each of the Tell Tayinat excavation seasons. All fragments of wood charcoal chosen for examination that were larger than 2 mm were fractured by hand or with a steel razor blade to create fresh transverse, radial, and tangential planes, in order to examine the wood anatomical structure. After fracturing, samples were supported in a sand bath and examined under a Motic K-400P stereo microscope at x6 to x50 magnification and an Olympus BX51 polarizing microscope at x50 to x500 magnification. The micro-anatomical features of each section were documented, photographed, and compared with those from modern reference collection materials, standard reference texts [124, 125], and the InsideWood online database (https://insidewood.lib.ncsu.edu/). Seed and non-wood botanical remains were identified by comparing sample morphological characteristics with modern reference materials and reference seed atlases (e.g., [126, 127]). A LEO 1550 field emission scanning electron microscope (FESEM) was used for high magnification observation of plant micro-features and high-quality image capture.

Following identification, we selected a number of samples (both wood charcoal and short-lived seeds) for which reasonably secure archaeological associations are available (as is inevitable at a complicated multi-period tell site with such a multi-phase stratigraphic sequence, the work of this project has in fact led to the reassessment of a few contexts–see below). The general Tell Tayinat site stratigraphic sequence is set out in Table 1; further stratigraphic description and the discussion of associated material culture can be found in [48, 50, 51, 53, 110, 118]. The samples ultimately selected for ¹⁴C dating and any comments related to their archaeological contexts are listed in Table 2. These samples were then radiocarbon dated at the Oxford Radiocarbon Accelerator Unit. Acid-Base-Acid (ABA) sample pretreatment, target preparation, and Accelerator Mass Spectrometry (AMS) ¹⁴C dating were performed following methods described previously [128–130]. Isotopic fractionation has been corrected for employing the δ¹³C values measured on the AMS–the quoted δ¹³C values were measured independently on a stable isotope mass spectrometer (±0.3‰ relative to VPDB). The 49 new ¹⁴C dates acquired are listed in Table 3.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 1. The Tell Tayinat general stratigraphic sequence indicating contexts of radiocarbon dated samples.

Descriptions of associated stratigraphy and material culture can be found in [48, 50, 51, 53, 110, 118]. The conventional dates for the associated periods are indicated following [105–106] for the Iron Age and [34] for the Early Bronze Age, while previously published dates for the main Tayinat phases are indicated in parentheses and italics following [53].

https://doi.org/10.1371/journal.pone.0240799.t001

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 2. Samples for ¹⁴C dating from Tell Tayinat obtained by this project.

For comments on some of the sample contexts, see the notes below the table. Tayinat General Period Scheme refers to the General Site Phases as identified in Table 1. Local Field Phase refers to the individual phasing schemes devised independently for each excavation area, outlined in Table 1.

https://doi.org/10.1371/journal.pone.0240799.t002

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 3. ¹⁴C dates on the samples in Table 2.

https://doi.org/10.1371/journal.pone.0240799.t003

Results

We first considered Model 1, which employs all the data, and then the very slightly revised Model 2 removing the outlier, OxA-32139, as noted above. Model 1 does not quite achieve satisfactory OxCal A_model and A_overall values given the outlier just noted (typically ~59–60), with one very low individual OxCal Agreement value (<6) for OxA-32139. The Model 1 run values quoted in Table 4 and the model run shown in S1 and S2 Figs achieved satisfactory Convergence (C) values ≥95 for all elements. However, it should be noted that, especially without use of a high initial kIterations value, runs of Model 1 in a number of cases fail to achieve ≥95 Convergence values for some of the late elements in the model (Phase 2 Late 1 onwards). This problem usually does not occur once OxA-32139 is excluded in Model 2. Fig 2 shows the Phase 6 part of Model 1 and indicates the extreme outlier date of OxA-32139 on sample SA5113. For the full Model 1, see S1 File, S1 and S2 Figs. The placement of the Model 1 dated elements versus IntCal20 (and with the previous IntCal13 shown for comparison) is shown in S3 Fig.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 2. The portion of Model 1 showing the period 6 data to illustrate the very large outlier OxA-32139 on sample SA5113 (see text).

This sample is excluded from the remainder of the modeling in our study. Data from OxCal 4.3.2 and IntCal20 with calibration curve resolution set at 1 year. The OxCal Agreement (A) values, the Posterior v. Prior values from the OxCal General Outlier model (O) for the short-lived samples, and the Convergence (C) values are all shown. The dates on wood charcoal samples with the Charcoal Outlier Plus model applied always have an outlier value of 100/100. The light-shaded red probability distributions for each dated sample are the non-modeled calibrated age probability distributions for each sample in isolation. The dark red probability distributions are the modeled (posterior density) calendar age probability distributions. The lines under each probability distribution indicate the modeled 68.2% and 95.4% highest posterior density (hpd) ranges. (Note: OxCal from version 4.4.1 uses 68.3% hpd ranges, however, since we employed OxCal 4.3.2 in this paper, we list 68.2% ranges following OxCal version 4.3.2.).

https://doi.org/10.1371/journal.pone.0240799.g002

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 4. Selected modeled calendar age ranges from the models and outputs shown in Figs 3–5.

TPQ refers to a date solely from a long-lived sample(s). Typical results shown; very small variations (often of around 1 year) occur between different model runs (we illustrate by giving results from a very similar but different model run in S3 File, where a number of start/end dates for some of the ranges are 1 year different). For comparison of the results for Model 2 with the previous IntCal13 [134], see below in Table 6. Whole ranges only are listed (compare with S3 File where sub-ranges are listed). The Time Span Phases 4&3 Date estimate combines the separate start (Phase 4) and end (Phase 3) Date estimates in the model.

https://doi.org/10.1371/journal.pone.0240799.t004

Figs 3 and 4 show Model 2. This is a revision of Model 1 removing OxA-32139. Model 2 achieves acceptable OxCal agreement index values overall, with A_model around 81 and A_overall around 81, above the accepted threshold value of 60 (typical values based on several runs). Usually all elements now achieve satisfactory Convergence values ≥95. There are no outliers among the dates on short-lived samples above the 6% level. These are very minor discrepancies, and the outlier modeling slightly down-weights their influence. We thus employ this model as our best estimate for the Tell Tayinat chronology. The calendar age ranges for various of the elements from both Model 1 and Model 2 are detailed in Table 4. They are very similar.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 3. Model 2 (excluding OxA-32139): Bayesian chronological model for Tell Tayinat Iron Age sequence, part 1.

The A_model and A_overall values are satisfactory versus the threshold value of 60 (Table 4). Data from OxCal 4.3.2 and IntCal20 with calibration curve resolution set at 1 year. The Individual OxCal Agreement values (A), the Posterior v. Prior values from the OxCal General Outlier model for the short-lived samples (O), and the Convergence (C) values are all shown. The wood charcoal samples with the Charcoal Plus Outlier model applied all have a Posterior/Prior value of 100/100. The light-shaded red probability distributions for each dated sample are the non-modeled calibrated age probability distributions for each sample in isolation. The dark red probability distributions are the modeled calendar age probability distributions. The line under each probability distribution indicates the modeled 95.4% highest posterior density (hpd) range. Cyan color indicates the start and end Boundaries of the model. Green color indicates the Boundaries calculated within the Tell Tayinat Sequence. Blue color indicates an OxCal Date estimate query for a Phase (this quantifies the time period within the start and end Boundary for the relevant Phase).

https://doi.org/10.1371/journal.pone.0240799.g003

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 4. Model 2 (excluding OxA-32139): Bayesian chronological model for Tell Tayinat Iron Age sequence, part 2.

Otherwise, see caption to Fig 3.

https://doi.org/10.1371/journal.pone.0240799.g004

Discussion

The EB IV dates indicate that the terminal EB occupation at Tell Tayinat lies in the 23^rd to 22^nd centuries BCE. This suggests that the site was active as the 4.2 ka climate event began (23^rd century BCE) and likely continued at least into the early part of the following century. We have relatively few data as yet from the more substantial earlier phases of occupation, including the construction of a major structure in FP8b [48, 50], for which only a terminus post quem can be provided. The destruction of this structure, however, appears to be dated somewhere between 2300–2200 BCE, relatively early in the EB IVB period. Furthermore, no dates have yet been obtained from the recently excavated preceding levels, most notably from another substantial construction that appears to have been destroyed by fire and has been tentatively dated to the EB IVA period (FP10). More data, however, pertains to the terminal phases of EB occupation, after the destruction of the FP8 structure. The final EB IV Phase 7, in particular, exhibits some spread in ¹⁴C ages (Figs 3 and 5) among the olive pits represented, with two (OxA-32138, OxA-32134) perhaps indicating a date range around/after 2100 BCE. Phase 7 appears to represent a relatively drawn out period of reducing circumstances at the site following the destruction of the more substantial architecture of Phase 8. We lack any constraint on the end of Phase 7, since there is then a gap in site occupation. The end Boundary could reach, in round terms, to ~2100 BCE at 68.2% hpd and ~2000 BCE at 95.4% hpd. The data to hand suggest that Tell Tayinat Phase 7 occupation likely ran at least well into the 22^nd century BCE, and perhaps further.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 5. The Tell Tayinat Phase 7 data from Model 2 in Fig 3 shown in more detail.

The lines under each probability distribution indicate the modeled 68.2% and 95.4% highest posterior density (hpd) ranges.

https://doi.org/10.1371/journal.pone.0240799.g005

This phase is associated with the decline and eventual abandonment of the site at the end of the phase, representing the terminal EB occupation at the site, after which occupation shifts to the neighboring site of Tell Atchana (ancient Alalakh) for the Middle and Late Bronze Ages, beginning in the terminal phases of the EB. This abandonment thus appears to occupy a period around and shortly after the onset of the 4.2 ka cultural-climate episode. It has been increasingly recognized that this climate episode did not significantly impact sites in the northern Levant, and specifically sites in the Euphrates and Orontes river valleys, in the same manner as observed in northern Mesopotamia [34, 38]. This does not contradict the general climate shift to more arid conditions observed in a number of records, including directly from barley finds from archaeological sites in the region [144]. But it does highlight that such a general regional impact has very different effects locally, depending on a range of human, geographic/environmental and technological factors [144]. Even during the climate episodes around 4200 BP (~2200 BCE) and 3200 BP (~1200 BCE), barley grains from coastal sites, for example, show little substantive drought stress [144]. In the northern Levant, in particular, it has been proposed that the karstic geology of these regions and their resulting access to underground aquifers as more stable water sources in times of drought, positioned them as “refugia” for “habitat-tracking populations” in times of climate crisis [25, 145, see also ref. 35]. Favorable physiographic and edaphic features, or irrigation technology and low-risk, sustainable agricultural practices may also mitigate local impacts.

Such comparatively favorable conditions, as well as possible mitigation strategies, may be reflected in the archaeobotanical evidence from Tayinat. In both FP8 and FP7, barley and free-threshing wheat are well-represented, and emmer wheat is also present in significant quantities [48, 146, 147]. Although barley is the dominant cereal crop, the frequencies of (more water-demanding) wheat are notably higher compared to sites further east, which typically relied much more heavily on barley [148]. This suggests better water availability in the Amuq compared to other inland areas in the late 3^rd millennium, but the frequency of emmer may simultaneously represent a strategy aimed at minimizing yield variation. Similarly, Tayinat displays very high ubiquities in FPs 8–7 for a range of water-demanding species such as olive and grape, particularly when compared to other inland sites [146, 147]. Notably, no significant shift in species representation appears between FP8 and FP7 [147]. Carbon isotope analysis of cereal grains in addition indicates no evidence of drought stress in either barley or wheat during either phase, and crops may in fact be slightly better watered in FP7 compared to FP8 [147].

On the other hand, the zooarchaeological assemblages associated with FP8 and FP7 show notable differences [48]. Sheep and goat have a greater focus in absolute numbers compared to pig and cattle in FP7 in comparison to FP8, although cattle appear to have contributed the most significant amount of meat to the diet in both phases. Kill-off patterns for sheep and goat suggest a mixed animal management strategy was employed in both phases, with meat, milk and wool all likely playing a significant role. Notably, hunting is frequent in FP8, but becomes much less common in FP7. Hunting in FP8 focused on red deer (Cervus elaphus), although roe deer (Capreolus capreolus), gazelle and hare/rabbit are also present. Similarly, fish is found in very high proportions in FP8, and although it remains high in FP7 it declines significantly. Similar patterns are seen in bird, turtle and amphibian remains, although these are all found in much lower numbers than fish.

The results suggest a much more varied subsistence strategy in FP8, relying on a combination of domestic livestock and significant supplementation from hunting, fishing and fowling. In FP7, in contrast, hunting and fishing decrease notably and reliance on livestock increases. This is particularly true of sheep, goat and cattle, although pig declines slightly in FP7. Despite the decline visible architecturally after the destruction of the major building at the end of FP8, these observations suggest that FP7’s subsistence economy became more intensively focused on domestic livestock than in the preceding period.

Overall, despite their approximate temporal coincidence, a direct causal relationship between Tayinat’s late EB decline and this climate episode is difficult to support, as it seems much more likely that the site failed to recover from its destruction during the preceding century (or more–since the length of Phase 7 is unclear). This destruction is positioned alongside evidence for similar destruction levels in the northern Levant, most notably at the major regional center of Tell Mardikh-Ebla, where the Palace G complex was destroyed ~2350–2300 BCE [83]. The agents responsible for such destructions remain debated [149–151], but in the aftermath, Ebla was rapidly reconstructed and regained its status as a primary regional center, although perhaps on a more modest scale than observed in the preceding period [34, 39, 152]. Tayinat, in contrast, although re-occupied, does not seem to have recovered its primate position in the region, with the principal settlement shifting to nearby Tell Atchana, which then remained the central settlement throughout the Middle and Late Bronze Ages. This shift, in contrast to wider regional reconstructions of major climatic crisis, has often been postulated as the result of local factors, such as a decisive shift in the course of the Orontes River [153, 154]. Indeed, in light of the close proximity of Tell Tayinat and Tell Atchana, this change may be more reflective of a local political and spatial reorganization than a major break in settlement.

Our Iron Age data and modeling provide a refined and robust absolute timeframe for the early Iron Age at Tell Tayinat, independent of cultural and historical assumptions, running from around the 12^th through mid-8^th centuries BCE (Table 4, Figs 3 and 4). The site’s earliest Iron Age occupation thus lies squarely in the period following the collapse of the Late Bronze Age ca. 1200 BCE, and represents an alternative model and context for this period of transformation and climate challenge in the Old World [51]. Discussion here will focus primarily on the dates from the early Iron Age (i.e. Iron I-early Iron II, ca. 12^th-9^th centuries BCE, Table 4); further discussion of the late Iron II dates here will await a future publication about the Iron II-III transition at the site (i.e. 8^th-7^th centuries BCE) [155]. The dates calculated are consistent with the broad chronological sequence constructed based on linkages to Aegeanizing ceramics of the Late Helladic (LH) IIIC tradition [45, 46, 51, 55, 56]. The earliest levels (Field Phase [FP] 6c), which largely precede the widespread use of Aegeanizing LH IIIC-style ceramics at the site [51, 110], begin in the 12^th century BCE. Aegeanizing influences begin to appear in the late 12^th century (FP6b) and proliferate in the 11^th century (FP6a-5), continuing in declining frequencies into the mid-10^th century (FP4-3, 2 Early) [51, 56, 110]. These dates are consistent with a more or less conventional chronology and do not support recent suggestions for much earlier (or ‘higher’) dates for the end of the Late Bronze Age (and the LH IIIB to LH IIIC transition) [97]. We place Tell Tayinat Phase 6b-a and its assemblage and associations with the Late Helladic IIIC tradition [51, 56, 110] from the late 12^th century BCE onwards, in line with other recent ¹⁴C based work in the Aegean and East Mediterranean [94, 96, 156–158].

The original excavations at the site produced a monumental structure known as Building XIV that was assigned to Building Period 1 [53]. The current excavations have identified the foundations of this structure [45–47, 118], but have not yet produced any datable contexts for radiocarbon analysis. The stratigraphic position of the foundational remains of Building XIV, however, allow the proposition of a narrow date range, based on the model presented here, in the mid-late 10^th century BCE, contemporary with Phase 2 Middle A1. The latter phase is associated in Field 1 with the construction of a major ditch and sherd-paved street, located immediately to the south of Building XIV [118]. The material assemblage associated with this building and with the street contains Red-Slipped Burnished Ware, whose fluorescence is conventionally dated to the Iron Age II, although only small quantities were found [118]. Instead, the ceramic assemblage for phases 2 Middle A(1) and 2 Middle A(2) is dominated by plain wares, suggesting a date between Phases 2 Middle A and 2 Middle B (ca. 900 BCE) for the Iron I-II transition. Samples analyzed that post-date the Iron I-II transition at the site all originate from the Iron II occupation at Tell Tayinat, ending in Phase 2 Late 2. This appears likely to precede closely the 738 BCE date when the Assyrians conquered the site and assumed its control [65, 66, 118]. The modelled dates for Phase 2 Late (Table 4) are consistent with this assessment. The Date estimate is 772–753 BCE (68.2% hpd) and 791–735 BCE (95.4% hpd), indicating the period lies shortly before the Assyrian conquest. The Boundary estimate for the Phase 2 to 1 transition, 764–743 BCE (68.2% hpd), 771–721 BCE (95.4% hpd) is either just before, or at about the same time as, the Assyrian conquest. We currently lack any ¹⁴C data from Phase 1 to clarify the dating after Phase 2. It seems inherently likely that the Phase 2 to 1 transition probably is associated with the Assyrian conquest and associated changes following 738 BCE and this assumption appears compatible with the available ¹⁴C-based timeframe.

No Iron III context has yet been the subject of ¹⁴C dating, but the dates might thus be anticipated to fall in the period of Assyrian control following 738 BCE and must continue until after at least 672 BCE [66, 118, 143].

Our chronology allows comparison of some persons known from the epigraphic record and likely linked with names thought to refer to Tell Tayinat or its territory [47, 60–62] (see Table 5). The Aleppo Citadel inscription of king Taita, hero and ruler of Palistin, is dated to the 11^th century BCE on the basis of paleography and iconography [60–62]. The currently proposed chronological scheme would link this ruler to Tayinat Phases 6a to 5. A putative second Taita, reconstructed from inscriptional evidence from Meharde and Sheizar, has been attributed on similar grounds to the (early?) 10^th century BCE [62]. This ruler would thus most likely be associated with the later Iron I materials at the site (FPs 4–3). As such, contrary to earlier assumptions, the reigns of these rulers likely preceded the monumental constructions of Building Period 1 (Buildings XIII and XIV). This time period is designated Phase 2 Middle A(1). Dates are estimated within the model from the surrounding data and constraints, since there are no ¹⁴C data for this period, at about 955–911 BCE, 68.2% hpd, and 961–866 BCE, 95.4% hpd (Table 4). Any potential monumental constructions associated with the reigns of the earlier 10^th century rulers remain to be discovered. Other possible 10^th century rulers attested on the Arsuz stelae [62, 98], notably Suppiluliuma (I), who claims to have conducted a war in Cilicia, may be more plausibly related to the construction of the monumental structures associated with Building Period 1 in the mid-late 10^th century BCE. These prominent figures also appear to be associated with a significant reconfiguration of urban space at the site, and an expansion of their kingdom’s territorial extent. By the 9^th century BCE, Tell Tayinat had become the city of Kunulua, the apparent royal city of the Neo-Hittite kingdom of Patina (or Unqi) [47, 61, 159]. The ruler Suppiluliuma (II), attested in a monumental inscription recently discovered at Tell Tayinat, likely corresponds to the Sapalulme mentioned in the campaign records of Shalmaneser III [62, 63, 159, 160]. Qalparunda, a ruler of Patina, paid tribute to Shalmaneser III in both 857 and 853 BCE, and may correspond to the Halparuntiyas mentioned in Tayinat Inscription 1 [47, 159–161]. These two rulers would likely correspond with the time of Phase 2 Middle B.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 5. Suggested correlations between Tayinat Phases, absolute dates as reconstructed here by Bayesian modelling of ¹⁴C dates, rulers attested in historical sources, and conventional northern Levantine Iron Age periodization.

For Iron I, alternating pale orange and white coloration denotes the four-period division as outlined in [51]; for Iron II-III, gray and white coloration denotes the separation between Tayinat Phases.

https://doi.org/10.1371/journal.pone.0240799.t005

Here too, a direct association between various reconstructions of 13^th-10^th centuries BCE climatic crisis in the eastern Mediterranean and Near East (e.g. [3, 20–23, 26, 43, 162–164]) and the trajectory of development at the site of Tell Tayinat is difficult to discern. The decline of the Late Bronze Age settlement of Tell Atchana begins at least a century earlier than the proposed major crisis that occurs ca. 1200 BCE and afterwards, and the site’s occupation shrinks significantly during the 13^th century BCE [165, 166]. In contrast, during the proposed Early Iron Age shift to more arid conditions, the site of Tell Tayinat appears to flourish [51, 110]. Historical records suggest the formation and expansion of a major political entity centered around the Amuq Plain and Tell Tayinat during the 11^th-10^th centuries BCE [60–62], the height of the reconstructed climate crisis. Although monumental architecture contemporary to the earliest historically attested rulers has not yet been uncovered by the current excavations, the archaeological evidence suggests a prosperous settlement that remained tied into long-distance trade networks. Major monumental constructions and a significant reorganization of urban space are attested in contexts now dated to the late 10^th century (Building Period 1, Phase 2 Middle A(1)), culminating in the formation of the major Neo-Hittite royal city of Kunulua in the 9^th century BCE. In combination with the historical evidence, this suggests that the rise of Tayinat, and the foundations of the later Iron Age social and political institutions at the site, had their origins during a period frequently defined (elsewhere in the wider region) as one of crisis and collapse.

The archaeological evidence from Tayinat points toward a variety of mechanisms that may have contributed to this Iron Age fluorescence. The stability of the location of the central settlement in the Amuq in the south-central part of the plain throughout the Bronze and Iron Ages suggests that the short-distance shift in site location from Tell Atchana to Tayinat during the 12^th century BCE could not have been a response to large-scale climatic degradation or to major shifts in economic networks. Rather, it seems likely to represent an adaptive response to locally changing conditions, which may have included a shift in the course of the Orontes River [153–154]. An Orontes paleo-channel identified between the sites of Tayinat and Atchana has been suggested to date to the Iron Age, and the shift of the main settlement to the north bank of the river would have maximized the site’s access to agricultural land in the plain north of the river [154]. Furthermore, the site of Tayinat al-Saghir, a small mound artificially constructed between Tayinat and Atchana at some point during the Iron Age, may represent a quay that would have allowed the main settlement to control riverine traffic, cementing its economically strategic position [118].

Palaeobotanical evidence likewise provides additional detail to this complex picture. Crop isotope data from Tayinat suggest minimal evidence for drought stress during the early Iron Age, particularly for barley, and generally suggest improvement in crop water status over the course of the Iron I [147, 167]. Furthermore, continued cultivation of water-demanding species such as grape, olive and fig at Tayinat during the Iron I does not suggest that water availability was a significant issue [167]. These arboricultural taxa reach their highest ubiquities and frequencies in FP6 and generally decline somewhat thereafter [51]. Free-threshing wheat also remains the most frequent crop plant throughout the Iron I, although barley increases in ubiquity during FPs 6–5 [51]. This increased focus on free-threshing wheat has been suggested to represent a labor optimization strategy [167]. Emmer is notably more ubiquitous during the Iron I than in the preceding LBA or during the later parts of the Iron Age, which may represent a strategy employed to minimize the risk associated with fluctuations in yield [51]. The frequency of large-seeded vetches such as bitter vetch (Vicia) and grass pea (Lathyrus) are similarly interpreted as indicative of a risk-spreading strategy that prevents crop failures, stabilizes crop yields and maintains soil fertility [167]. In the Iron Age, the frequent co-occurrence in northern Levantine sites of a focus on free-threshing wheat and large vetches has been interpreted as an attempt to achieve balance between the water input and labor requirements of agricultural production [167].

The zooarchaeological evidence points to a reasonably consistent strategy of animal raising through the Iron I, although differences exist between FP6b-a and the later phases of the Iron I. Pig consumption is at its most frequent during FP6b-a, although cattle consistently represents the most significant contributor of meat to the diet [51]. Mortality curves for sheep and goat in the Iron I suggest strategies aimed at exploitation of secondary products (dairy and particularly wool/hair), and there is evidence for significant textile production beginning already in the late 12^th century [51, 110]. Hunting evidence is found in very low frequencies in all phases, while fishing is noticeably more frequent in FP6b-a compared to later Iron I phases [51].

In addition to the political decline of Alalakh during the 13^th century BCE, textual sources suggest that problems with grain production and/or distribution may have resulted in the diversion of local agricultural products to elsewhere within the Hittite Empire [15, 42, 168], as well as a concurrent decline in regional population within the Amuq as a result of Hittite deportations that preceded the climate crisis [168]. The frequencies of wild taxa associated with moist environments suggest a wetter environment at Tayinat during FP6 [51, 147]. The Amuq has historically experienced a fluctuating degree of marshiness [118, 153], and this may explain the increased appearance of fish in the zooarchaeological assemblage during the same period [51]. In contrast, FPs 5–3 are associated with wild plant taxa that may indicate an expansion in agricultural area during the late 11^th-10^th centuries BCE through the use of new arable fields that were likely left uncultivated during the 12^th-early 11^th centuries BCE [51]. This suggests an agricultural extensification strategy during this time that may be related to increasing population in the Amuq Plain (and stable political-economic circumstances), as identified from settlement data [51, 56, 169], and which may reflect increasing urbanization connected to the formation of the kingdom of Palistin. Indeed, comparison of the apparent lower town plans extant and accessible via geophysics at Tell Tayinat would suggest that Tayinat records long-term organic town growth through the course of the Iron Age, whereas the nearby site of Zincirli, in contrast, perhaps largely records a new post-Assyrian conquest layout [169]. This may indicate only limited post-Assyrian-conquest changes beyond the monumental constructions of the elite zone in the upper city at Tayinat [118].

As suggested for the late EB period (see above), the local environment (through a combination of the Orontes River and the karstic geology) may have provided a suitable context in this region for greater resilience despite the arguments and evidence for climate change (more arid conditions) following ~3200 BP (~1200 BCE). Crop isotope data from Tayinat mentioned above are consistent with broader regional patterns that suggest the Orontes and coastal regions were somewhat less affected by drought stress [144, 167]. Comparative studies would also indicate that the nature of the association between climate change and social change is, predictably, complex and contingent on the nature and history of the relevant societies and their specific vulnerabilities (e.g. evolved socio-environmental mis-matches [170]). In particular, climate change alone is often not the key element that overcomes societal resilience; rather, it is sharp, volatile, fluctuations with a duration of several years that seem the greater threat to established complex agrarian societies in the preindustrial period (e.g. [29, 171–176]).

A final issue is the robustness of the ¹⁴C calibration record with respect to the Tell Tayinat samples. The new IntCal20 [133] ¹⁴C dataset is greatly enhanced (in terms of data for several periods, and quality) compared with IntCal13 [134]. However, for the periods of time relevant to Tell Tayinat and its dated elements, there are in fact only relatively small changes (see S3 Fig) and no major new underlying data contributions. Table 6 compares the results for Model 2 (in Table 4) with the same model run with IntCal13 (an example is cited with good OxCal A_model and A_overall values and all Convergence, C, values ≥95 –note that many runs do not achieve good Convergence in the final part of the model unless the kIterations value is increased, from about or after Phase 2 Late 1). There are few substantial differences; date ranges are largely similar.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 6. Comparison of the modeled calendar age ranges from Model 2 with IntCal20 [133] (Table 4) versus runs of Model 2 using the previous IntCal13 calibration curve [134] and the Hd GOR Mediterranean dataset [180].

Data from example runs with satisfactory A_model and A_overall values (>60) and with all dated elements with satisfactory Convergence values (≥95). Whole ranges listed. Phase 4 and 3 Date estimates combined as start Phase 4 to end Phase 3.

https://doi.org/10.1371/journal.pone.0240799.t006

A potentially greater concern is recent work indicating the possible relevance at times of a modest/small Mediterranean growing season-related ¹⁴C offset for high-resolution calendar age determinations from the lower elevation Mediterranean region [177–180]. There is an intra-annual (i.e. seasonal) atmospheric ¹⁴C cycle, with a winter low and a summer high. Thus plants growing (and photosynthesizing) in the Mediterranean region at lower elevations in the winter through spring period (and stopping growth by the summer) may yield a recognizably slightly different ¹⁴C history versus the IntCal record derived from trees from central and northern Europe and North America that grow primarily from later spring and right through the summer [179, 180]. However, the fact that the overwhelming majority of the short-lived samples in this Tayinat study are olive pits (~86% of such samples) likely partly mitigates this issue. Unlike many trees or cereals and other field crops in lower elevation eastern Mediterranean environments, olive fruit grow from later spring through the autumn and hence comprise a ¹⁴C record that is only partly (not largely) out of kilter with the IntCal record. Grapes also grow through, and are harvested late in the summer to start of autumn, and again, in contrast with field crops like cereals, thus minimize any likely growing season offset [179–181] (in all, olive pits and a grape seed include 89% of the short-lived samples dated at Tayinat). Nonetheless, for interest, we compare the Model 2 results run against the Mediterranean-Anatolian Hd Gordion (GOR) dataset [180] in Table 6. This Hd GOR record is only a sketch for the Mediterranean—much more work is needed—and lacks data for the EB part of the Tayinat record and ends during the period of the late Tayinat samples. Thus it offers only a partial indication of possible differences. The Hd GOR record is largely similar to IntCal20 for most periods, and some of the changes in IntCal20, versus IntCal13, reduce what were previously further instances of differences when comparing the Hd GOR record versus IntCal13 [180], for example especially in the earlier 16^th century BCE (see the region labelled 1 in S4 Fig). Nonetheless, within general similarity, there are some periods, notably at the times of reversals and plateaus in the radiocarbon record, where the Hd GOR record exhibits some offset. Two examples are indicated (labelled as 2, 3) in S4 Fig. A possible additional area of minor offset might also exist at the reversal/plateau covering the earlier to mid-9^th century BCE labelled with the? in S4 Fig. The last two of these offsets could have minor relevance and effect on the Tayinat dates—but, noting that the types of short-lived samples dated likely minimize any effect (see above), this is likely insignificant.

As with the IntCal13 model, a number of model runs fail to achieve satisfactory Convergence values especially for the last part of the model from the Phase 2 Late 1 Date Estimate onwards unless the kIterations value is increased. Over multiple runs, there is also more noise in the late part of the model. The results in Table 6 are for a typical successful run with good Convergence. The results are generally similar to those from IntCal20 and IntCal13. However, in line with the observations of a small growing season offset issue and its possible consequences [177–180], we notice some modest effects, and especially during periods of reversals and plateaus in the radiocarbon calibration curve [179, 180]. For the periods where the Hd GOR [180] dataset applies (thus only for Tayinat Phases 6 onwards), the date ranges for some of the Tayinat phases (from Phase 6c to Phase 2 Middle B) are a little later, variously by around a decade to several decades considering the 68.2% hpd ranges. Consistent with previous observations [179, 180], the largest shift indications (of ~12–47 years in the 68.2% hpd ranges) occur in the 10^th and early to mid-9^th centuries BCE when there are reversals in the ¹⁴C calibration curve [133, 134] (S4 Fig). In view of the comments above about olive fruit and grapes, the effective (i.e. real) offset for Tell Tayinat is likely a little smaller. Nevertheless, this exercise highlights an area of possible minor chronological variation. Where this offset does apply, the effect is to achieve slightly later (more recent) calendar age estimates (something of potential relevance to debates over Iron Age chronology in the southern Levant, for example [179]). The very substantial change in atmospheric radiocarbon levels (the steep slope in the calibration curve) from the late 9^th through mid-8^th centuries BCE (linked with a major change in radiocarbon production and thence changes in solar activity processes [182]) (S4 Fig) clarifies that, regardless of any minor variations, the Tayinat Phase 2 Late 2 data are mid-8^th century BCE and thus likely represent the last pre-Assyrian conquest (738 BCE) phase at the site.

Conclusions

The absolute dating of the later Early Bronze Age and earlier Iron Age occupation periods at Tell Tayinat and associated northern Levantine sites has been the subject of debate and ambiguity for many years. The regional chronological frameworks for the northern Levant during both periods have never been adequately addressed in absolute chronological terms, but rather have been largely based on relative chronologies derived from regional ceramic sequences. The chronology of the early Iron Age, in particular, has been linked only approximately to material and stylistic associations and thence to debates in other areas, in both the Aegean and the wider East Mediterranean, concerning the centuries following the collapse of the 13^th century BCE palace-era Late Bronze Age civilizations. Our chronology, based on the integration of the archaeological sequence at Tell Tayinat with radiocarbon dates, provides for the first time a directly relevant, refined, and robust timeframe for this important site and its region. The chronological framework thus developed places the two major occupation phases at Tell Tayinat firmly within temporal contexts relevant to ongoing debates about two periods of supposed climate crisis (around and following ca. 4200 BP/2200 BCE and around and following ca. 3200 BP/ 1200 BCE). The complex and contrasting responses observed at Tell Tayinat during these two transformative periods positions the site as a locus strategic to understanding the diverse ‘alternative’ developmental trajectories observed during these two intermediate eras.

Supporting information

Acknowledgments

This work is a contribution from the CRANE project (https://www.crane.utoronto.ca/). We thank all our excavation collaborators and the team at the Oxford Radiocarbon Accelerator Unit. We thank the Directorate of Cultural Heritage and Museums of Turkey, which has graciously awarded the research permits necessary to conduct each of the Tell Tayinat excavation seasons. We thank the ministry representatives who have supported the campaigns each season. We thank the directors and staff of the Antakya Museum, and the Tayinat landowners, in particular the Kuseyri family, who have generously permitted work on their land. All necessary permits were obtained for the described study, which complied with all relevant regulations. BL thanks the Cornell Center for Materials Research Shared Facilities for use of facilities for organic materials identification work.