Models of Neolithisation of Northeastern Iberian Peninsula: New Evidence of Human Occupations During the Sixth Millennium cal BC

: The goal of this article is to discuss the signi ﬁ cance of the archaeological evidence from the sites of La Draga ( Banyoles, Spain ) and Coves del Fem ( Ulldemolins, Spain ) , in the context of the neolithisation of Northeastern Iberia. The 14 C dates have been analysed using Bayesian statistics. The stratigraphy of Coves del Fem covers the transition between the last hunter – gatherers of the region and the ﬁ rst farmers. The chronological sequence covers approximately 1,300 years, from 6065 – 5990 cal BC to 4700 – 4550 cal BC. The site of La Draga was occupied by the ﬁ rst farmers circa 5300 – 5230 cal BC when a wooden platform was constructed and ﬁ rst used. Subsequent repairs of the wooden piles have been dated as well. Another use of the wooden platform is documented around 5200 – 5085 cal BC, although until now new construction evidence has not been documented. La Draga site was reoccupied later, when several travertine structures dated in two moments between the years 5100 – 4900 cal BC and 4950 – 4700 cal BC were constructed and used. The radiocarbon dates of Coves del Fem and La Draga support the existence of two di ﬀ erent models of neolithisation in Northeastern Iberia. In the southern part of the territory, Coves del Fem suggests that the Holocene hunter – gatherer populations remained in the area until the arrival of the ﬁ rst farmers, in a model similar to the one observed at the Ebro basin. On the contrary, the site of La Draga supports the hypothesis of the ﬁ rst farmers colonising a previously unoccupied territory.


Coves del Fem Site
The site of Coves del Fem (Ulldemolins) is located at the Natural park of Serra del Montsant in southern Catalonia (Figure 1). The Montsant is a part of the prelitoral ranges with a higher altitude of 1,115 m. The site is actually a rock shelter 10 m above the current level of the Montsant river ( Figure 2). Currently, the documented cavity surface covers approximately 300 m 2 covered with sediment that potentially preserves archaeological layers. The first fieldwork at the site has revealed a complex, of almost 135 cm deep stratigraphic package, which covers the transitional period from the last hunter-gatherer occupations to the first evidence of early Neolithic occupation in the area (Bogdanovic, Palomo, Piqué, Rosillo, & Terradas, 2017;Palomo et al., 2018Palomo et al., , 2020. The Montsant range is a rich archaeological region and several sites have provided occupation evidence dated at the end of Paleolithic and the Epipaleolithic periods, as it is the case of Abric del Filador, Hort de la Boquera, or Abric del Colls (Garcia-Argüelles, Nadal, & Fullola, 2005, 2014bGarcia-Argüelles i Andreu et al., 2014a). However, the stratigraphical sequences of the region rarely document archaeological layers dated at the sixth millennium cal BC. As a result, the transition to the Early Neolithic period remains unclear.
The excavated surface from the upper layers covers 9 m 2 , whilst from layer 1,008 to the layer 1,014, the excavated surface covers only 5 m 2 due to the presence of a large block of stone. The five upper layers documented at Coves del Fem (UE 1008(UE , 1009(UE , 1011(UE , 1012(UE , and 1013 can be attributed to the end of the early Neolithic period according to the pottery style (Figure 3a), defined as epicardial (Bosch, 1994: 57). The pottery is characterised by globular shapes with plastic decoration, incisions, and channelling. Moreover, several structures have been documented in these epicardial layers: postholes, fireplaces, and pits. Some of the structures were excavated until the underlying soil perturbing the early Neolithic cardial and Mesolithic layers. Carbonised archaeobotanical remains are abundant, including among them are the remains of basketry, charcoal, fruits, and seeds. Wooden remains have also been preserved desiccated, showing woodworking traces. Lithic technology in the epicardial layers is characterised by blade production, always unipolar. The technique of knapping is characterised by indirect percussion and heat treatment on core preforms is often applied. The cores have platforms of percussion with denticulate overhangs. These layers have also provided the evidence of cereal cultivation as well as domesticated caprine.
Layers 1010 and 1014 have provided cardial ware ceramics, while the pots are a combination of impressions and dragged technique (Figure 3b). Combustion features and negative structures have also been identified during the excavation. At the cardial layer, the size of the unipolar blade production remains is shorter than the ones found in the epicardial levels. It is also necessary to emphasise the presence of geometrics of triangular form or segments among the retouched pieces of both Neolithic layers.
Finally, the lower layers UE 106, 107, 108, and 109 of the stratigraphic sequence established in a survey carried out during 2015 (pending unification and renaming with the new sequence established on more recent excavations) can be attributed to Mesolithic occupations. These layers have not been properly excavated and only a test pit was dug in order to obtain samples for radiocarbon dating. Nevertheless, it was possible to document a combustion structure used repeatedly in situ. The archaeological material consists of abundant charcoal, ash and, to a lesser extent, faunal remains and lithic industry. The combustion structure has allowed a very good conservation of the carbonised archaeobotanical remains, among which it is remarkable the presence of an entire cone pine. The lithic remains are scarce and for the moment no geometric elements have been found; however, it has been documented a core for the production of bladelets by hard percussion and other core for the production of flakes (Figure 3c).

La Draga Site
The site of La Draga is located in the central part of the eastern shore of Banyoles lake (Girona, NE Iberia) at an altitude of 170 m ( Figure 1). Nowadays, the archaeological site is partially underwater, but the most extensively excavated area is towards the inland. The site has an estimated extension of approximately 15,000 m 2 and has been systematically excavated since 1991. The fieldwork has been focused on three different sectors: sectors A and B-D in the shore of the lake and sector C underwater (Figure 4). One thousand square meters have been excavated so far providing an extraordinary archaeological record.
This open-air site corresponds to a lake dwelling of one of the first farming groups that settled in the northeast of the Iberian Peninsula (Andreaki et al., 2020;Bosch, Chinchilla, & Tarrús, 2000, 2011Palomo et al., 2014). The pottery style documented at the site corresponds to the cardial culture or impressed ware culture. Systematic surveys carried out in the area and archaeological data gathered at regional level until now confirm the scarce evidence of occupations prior to the Neolithic settlement in the region. The closest Mesolithic site, in this case, is the Bauma del Serrat del Pont, previously mentioned, some 25 km away (Alcalde & Saña, 2017).
According to the superposition of the structures built with travertine slabs over a layer of timber logs ( Figure 5), at least two phases of occupation have been proposed for the site, both corresponding to cardial ware culture (Palomo et al., 2014). However, these stratigraphic layers have been clearly documented in sectors B-D and C of the site, where waterlogged conditions have preserved the organic materials, whereas in the area of the site located further inland and at a higher height (Sector A), the stratigraphy shows a more   complex deposition of sediments. Due to the wedged deposits associated with the two occupational phases and the degradation of the timber logs' level, the two phases are not clearly separated in Sector A. In addition, the spatial distribution of the structures shows totally different dynamics.
The distribution of the travertine pavement and structures corresponding to the last moment of occupation are different in this sector. Several negative constructions, of irregular forms and of reduced or extensive dimensions, have been identified. These structures are filled with large quantities of diverse archaeological materials such as charred seeds, animal bones, fragments of ceramics, quartz, flint and bone tools, pieces of ornaments, and grinding instruments. Because of the kind of materials they contain, the pits have been interpreted as landfills for food waste and remains of manufactured objects that were once considered useless.
Combustion features are abundant in sector A where about 40 of them have been identified and they present different morphologies. The combustion features identified are arranged with travertine slabs, sandstones, or burnt pebbles, all in varying quantities, and charcoal, which are the remains of the firewood used.
The particular conditions of conservation at La Draga have allowed the recovery of evidence of some technical productions barely documented on Neolithic sites of the western Mediterranean. Agriculture and livestock, as well as artisanal productions, involved a diversification of tasks and required specific tools and equipment.
Regarding craft productions, the site has provided direct evidence of the use of plant fibres for the production of ropes and basketry and indirect evidence of textile production with vegetal fibres, as well as abundant evidence of tools made of wood ( Figure 6) (López-Bultó & Piqué, 2018;López-Bultó et al., 2020;Palomo et al., 2011Palomo et al., , 2013Piqué et al., 2015Piqué et al., , 2018. The remains of artefacts made with bone materials, shells, and minerals of different nature, provide a complete picture of the diversity of production processes Figure 6: Wooden tools and cordage from La Draga. (Berihuete, Girbal, Piqué, Palomo, & Terradas, 2018;De Diego et al., 2018;Terradas et al., 2017).
The diversity and quality of remains suggest that these early farmers had well-developed agriculture and husbandry (Antolín, 2016;Saña, 2011). They arrived with all their technological knowledge and created a new settlement of big dimensions.

Methods
The 14 C dates obtained from short live samples from La Draga and Coves del Fem are analysed in the frame of the early Neolithic dates of the northeast of the Iberian Peninsula. The radiocarbon ages provided by laboratory reports have been calibrated using the OxCal program (Bronk Ramsey, 2009) using the IntCal13 (Reimer et al., 2013) curve with the half-life variable of the sample introduced. They were later analysed using Bayesian statistics. Bayesian models were implemented in order to identify the discontinuities in the chronological sequence. In the case of Coves del Fem, a total amount of six radiocarbon dates have been analysed with Chronomodel 1.5.0 software . The data were introduced in stratigraphic order, since the overlapping of strata date is clear and the results of the calibration form a chronological distribution that corresponds to the stratigraphic arrangement.
In the case of La Draga, in addition to the 14 C dates, the dendrochronological analysis of the piles and horizontal woods was integrated, in order to obtain a better picture of the duration and dynamics of the settlement (Andreaki et al., 2020). Fourteen of these wooden piles were selected for the 14 C AMS method in order to date the moment of construction and the duration of the occupation. A total amount of 40 radiocarbon dates have been analysed with Chronomodel 2.0 software according to the stratigraphic order and sedimentary description of the archaeological deposits.

Coves del Fem Chronology
The site of Coves del Fem has provided nine radiocarbon dates on short live samples (seeds, pine cone, animal bone, and charcoal) ( Table 1). The sequence obtained covers approximately 1,300 years, from 6065-5988 to 4699-4545 cal BC. The analysis of the radiocarbon dates shows three sets of consecutive intervals, which in turn can be interpreted as widely contemporary (Figure 7).
The three lower stratigraphic units sampled are dated in 6065-5556 cal BC and correspond to the Mesolithic hunter-gatherer occupations. The radiocarbon dates of the early Neolithic cardial period come from three stratigraphic units and place this occupation between 5623 and 5221 cal BC. The last period of occupation is dated in the interval 4941-4545 cal BC and corresponds to the epicardial layers, the final phase of the early Neolithic period.
The chronological proximity between the Mesolithic stratigraphical layers and the strata attributed to the early Neolithic is noteworthy. On the other hand, there is a discontinuity within the early Neolithic period between the layers characterised as "cardial" and the ones characterised as "epicardial."

La Draga Dendrochronology
The excavations carried out to date at the site of La Draga have made it possible to recover more than a thousand wooden piles from all the excavated sectors. The dendrochronological analysis of the piles is still  not be correlated with any other because there is currently no fixed dendrochronological sequence that continues from the Neolithic period up to the present for the northeastern part of the Iberian Peninsula. The preservation of the last growth ring (cambium) in the 61 piles used to build the wooden platforms during the oldest phase of occupation in La Draga has allowed determining that most of them were cut in the same year and season. The dated piles come from the various sectors of the site, which is why the hypothesis is raised that the wooden platform or platforms on which the dwellings were built were distributed throughout the excavated area and that they were built at the same moment. First of all, the analysis of the global diagram shows that all sectors (A, B-D, and C are occupied at the same cut-off year, the year 237 of the floating chronology (Figure 8). That is, the settlers of La Draga constructed their dwellings at the same time. This evidence is quite unusual at the moment. In Switzerland (and throughout the Alpine Arch) sites dating to the early Neolithic period, settlements are generally smaller and their rate of growth is relatively slow and taken out by neighbourhood. Similar processes are documented for more recent settlements like Cortaillod-Estdating of the Final Bronze (Arnold, 1986;Gassmann, 1984).
The last phase of cut includes 26 samples that correspond to reinforcement and repairs of the structures (platforms and/or dwellings). These samples come from all the excavated sectors. All of them were cut between the years 241 and 265, which suggests a minimum duration of 28 years for the settlement´s life.
The 14 C AMS dates of 14 of these piles allow us to correlate the chronology of the site construction and the phases identified during the excavation according to stratigraphy and the radiocarbon chronology. In all cases, the last ring of the pile has been dated, which has made possible to determine with greater precision the date of construction of the wooden platform. In one of the piles, the 14 C dating does not agree with the dendrochronological results, which is due to a problem of contamination by microorganisms. It, therefore, has been discarded from the Bayesian model.
In the diagram, the dated samples correspond to both the construction of the platform and the subsequent repairs (Figure 8). The six oaks from the initial construction in the year 237 come from four oak stands of different ages. Oak stiffeners made from 241 to 265 come from at least two oak stands.
According to the annual ring growth, between three and five oak forests were exploited. The age of the forest and the anatomical features of the annual rings suggest that this forest was managed before the occupation of La Draga, probably by a population invisible until now in the archaeological record of the region.

La Draga Radiocarbon Dating
The 40 14 C AMS dates (Table 1 and Figure 9) were documented throughout the excavated sectors A, B, D and C. The dates come from faunal remains, seeds, and 14 of them from horizontal piles and timber logs (Andreaki et al., 2020). Except for an aberrant date around 3700 cal BC, the other dates correspond to the early Neolithic period. This way, a first Bayesian model was built in order to estimate the probability of the different documented stratigraphic events, which are summarised in Figure 10.
The best estimation suggests a 68% confidence interval between 5301 and 5229 cal BC for the moment of construction of the platform and for the first use of the wooden pile dwellings. The temporal distance between the cutting moment of the logs and successive moments of repair of the original platform can be determined, dendrochronologically, in around 29-30 years. It is less than the calibrated error estimated using radiocarbon dates. At least one date on a cereal seed is located in this interval and can be related directly to the repair of the wooden piles. According to the model, the moment of construction associated with wooden pile dwellings, their repair, and use (Platform) could be placed in the confidence interval at 68% interval of 5292-5216 cal BC.
The subsequent use of the platforms (Post-platform) corresponds to at least four stratigraphically contemporary events documented in sectors A, B-D, and C and dated at 68% interval of 5213-5085 cal BC.
After the construction and use of the wooden platforms during the first occupations, a reoccupation of the site has been documented, characterised by its stratigraphic association with anthropic accumulations of building blocks of limestone tuff -travertineand not with wooden structures. The new occupation, therefore, would not be related to the first. In sector A, eight structures correspond to occupations related to the use of travertine slabs. In sector B and in stratigraphic association with the travertine slabs that here adopt a configuration similar to that of a constructed pavement, three dates were sampled from three different stratigraphic units (III, IV, and IIa), with stratigraphic overlap between them.
The model suggests two moments of occupation related to the use of travertine slabs, the oldest moment (Travertine 1) ranging in the interval of 5093-4907 cal BC (68% confidence interval), while the most recent one (Travertine 2) in the interval of 4943-4713 cal BC (68% confidence interval).

Fems and La Draga in the Frame of the Early Neolithic of Northeast of Iberia
La Draga and Coves del Fem, the two presented sites in this article, offer some interesting data for the understanding of the models of the Neolithisation process in the region. They present clear differences with respect to this process. In the case of Coves del Fem, the Neolithic occupation reaches chronologically very close to the Mesolithic occupation as it happens in other sites located at the basin of the Ebro river. The site presents a complete chronostratigraphic sequence covering this transitional period, including some occupations attributable to the 6000-5500 cal BC period. In this case, however, the little excavated area neither allows the technological and cultural affiliation of the Mesolithic layers to be absolutely precise, nor helps to determine a possible relationship with the geometric settlements of the Ebro valley, dating close to the Neolithic period. In Northeastern Iberia, the late Mesolithic evidence documented so far (Can Sadurní, Bauma del Serrat del Pont, and Font del Ros) was characterised by notches and denticulate lithic assemblages; however, all these sites are located at the northern part of the territory, in the prelitoral ranges and the Pre-Pyrenees (Figure 1). The only exception is Cova del Vidre, the only site with geometrics that is placed at the low Ebro Basin. It is noteworthy that Coves del Fem is also placed in the Ebro Basin.
The same rock shelter was occupied later by a population with a very distinct technology. The cardial pottery is well represented, which suggests the arrival of the farmers at the area very early around 5623-5221 cal BC. However, despite the introduction of Neolithic elements, there is a continuity in the exploitation of faunal resources, as the same wild taxa occur throughout the sequence: Cervus elaphus, Oryctolagus cuniculus, and Capra pyrenaica (pers. communication M. Saña). The place where the site is located is not very appropriate for agriculture, as the rock shelter is located in a narrow canyon of the Montsant river. So far, scarce evidence of cultivation has been documented at the oldest Neolithic layers, although the archaeobotanical remains are still under study. The state of the current research does not allow yet determine the relationship between these early farmers and the last hunter-gatherer population. Later, Coves del Fem was recurrently occupied by farmers who dug silos and pits and built combustion structures probably related to the heat treatment of siliceous rocks, in order to improve their properties for knapping. The abundance of siliceous rocks in the area probably was one of the main attraction for the reoccupation of the shelter.
In the case of La Draga, the construction of a settlement of considerable dimensions by a Neolithic population occurred in a place without previous Mesolithic occupation. Evidence of the Mesolithic population is also absent in the closest area. Not far from La Draga is placed the site of La Bauma del Serrat del Pont (Alcalde & Saña, 2017) where the cardial layers, dated 5480-5380 cal BC, lay directly over the Mesolithic layers, dated 6300-6060 cal BC, which confirms the chronological distance of more than 500 hundred years between both occupations.
The lack of previous archaeological evidence suggests, in this case, the colonisation of a previously unoccupied territory. This colonisation implied such a massive felling of trees, both to construct dwellings and to enable cultivation fields, which had a significant impact on the environment (López-Bultó, 2019; Revelles et al., 2014Revelles et al., , 2015. Although the dendrochronological analysis has provided evidence of another interesting fact: the felled trees correspond to different forests, one of which was already cut 200 years before and the other one about 40 years before the construction of the platform. These data are very interesting because they show that a significant amount of human activity took place in the surroundings of La Draga before the first documented occupation at the site. It is very possible that this mass felling of the forest corresponds to the activity of Neolithic populations of which we have no other data. That is, the first Neolithic population settled at the lake surroundings before the dates that the archaeological evidence suggests. The evidence of forest management prior to the first occupation of La Draga supports the hypothesis of an early neolithisation in the region, also suggested by the chronology of the early Neolithic layer of the abovementioned site of La Bauma del Serrat del Pont (5480-5380 cal BC).
The earliest occupation of La Draga was performed by people carrying all the Neolithic kit. They completely depended on the cultivation of cereals and livestock breeding, whose remains are abundant in the site (Antolín, 2016;Saña, 2011). Moreover, they had a deep knowledge of local resources, both biotic and abiotic, that allowed them to remain in the same place for such a long period of time as well as to reoccupy the same place several times after a short interval of time.
To sum up, La Draga and Coves del Fem are examples of clearly different processes of Neolithisation between the northern and the southern parts of the Northeastern Iberia, on one side, regarding the relationship between the last hunter-gatherers and the arrival of the first farmers and on the other, the pattern of settlement or the economic strategies. Coves del Fem can be related to the sites of the Ebro Basin with respect to the close chronological distance between the last hunter-gatherers and the first farmers, and the pattern of settlement was characterised by the reoccupation of rock shelters. Unfortunately, due to the current state of the research, we are not able to characterise with more precision the technological tradition of these hunter-gatherer groups. The case of La Draga, on the other hand, confirms that the Neolithisation in the northern part was characterised by the arrival of people who colonised a previously unoccupied or scarcely occupied region. The first occupants, carrying cardial pottery, settled shortly after their first arrival at the Mediterranean litoral region and as a result, introduced the Neolithic way of life and its material culture to all the Pyrenean and pre-Pyrenean lands.
Nevertheless, the data allow us to reflect on the limitations of interpreting complex processes such as those of a regional occupation, at the time. We only have indirect data of the presence of a Neolithic population, more numerous and with a greater capacity to modify the environment than hunter-gatherer societies, surely much less numerous and also less intrusive. Of course, we cannot conclude that the absence of data from a Mesolithic population means an absolute population gap.

Conclusion
The radiocarbon dates of Coves del Fem and La Draga support the existence of two different models of neolithisation in Northeastern Iberia. In the southern part of the territory, the site of Coves del Fem suggests that the Holocene hunter-gatherers remained in the area until the arrival of the first farmers, in a model similar to that observed in the Ebro Basin. The early farmer occupation lays over the Mesolithic layers and the radiocarbon dates suggest that both occupations are close in time.
On the contrary, the site of La Draga supports the hypothesis that the first farmers colonised a previously unoccupied territory carrying already with them all the Neolithic kit. There is a gap of more than 500 hundred years between the late Mesolithic layers dated in the area and the first farmer evidence. However, according to the dendrochronological results, the first farmers probably arrived at the lake surroundings before the date suggested by the current radiocarbon dates, and shortly after their first arrival to the Mediterranean litoral.
Funding information: This work was supported by Ministerio de Ciencia, Innovación y Universidades, Spain (grant numbers PID2019-109254GB-C21, PID2019-109254GB-C22, and PCI2020-111992). The archaeological excavation at La Draga and Coves del Fem was funded by the Departament de Cultura (Generalitat de Catalunya) CLT009/18/00026 and CLT009/18/. The authors are members of the research group supported by AGAUR-Generalitat de Catalunya: TEDAS 2017 SGR 243 and ASD 2017 SGR 995. Raquel Piqué also acknowledges the support of Catalan Institution for Research and Advanced Studies (ICREA, Generalitat de Catalunya, Spain), through the ICREA Academia Program.
Conflict of interest: Authors state no conflict of interest.