Roca and its Landscape
The settlement of Roca Vecchia is located on a low promontory protruding in the Adriatic Sea, which is currently roughly 3ha in extension but has been estimated at least 5ha during late prehistory. The promontory area (area Castello Carrare) is closed toward the land by a large fortification wall whose earliest phase is to be dated to the Middle Bronze Age. Similarly to other coastal sites emerging in Apulia during the Bronze Age, Roca Vecchia is surrounded by a large lagoon (Bacino dei Tamari), which is a depression that was at least seasonally filled by water and that must have represented an important attraction area for early occupation (Iacono 2019: 75-76). From a general topographic standpoint, the landscape surrounding the site is completely flat. For this reason and before discussing the fieldwalking survey, the data recovered and their relevance to the study of long-term mobility trends, it is crucial to briefly introduce the area investigated also from a geological and geomorphological perspective as this represents a decisive step in order to understand potential biases in the surface record. The peninsular sector of southern Apulia (Salento) where Roca Vecchia is located, is a low plain whose continuity is interrupted by modest reliefs (locally called Serre) not exceeding 200 m above sea level. Overall, this landscape morphology is a result of the combination of the intense Mesozoic tectonic activity (giving the typical Horst-Graben structure) and the Cenozoic transgressive/regressive cycles (producing the marine infillings of basins, see Cotecchia, 2014 and references therein). The horsts (oriented NW-SE/NNW-SSE, according to the Salento main axis) are well-isolated from the grabens with elevation gradients in the order of tens of meters, and more pronounced on the eastern side than on the western one (Cotecchia, 2014; Largaiolli et al., 1969; Martinis, 1970; Martinis and Robba, 1971; Rossi, 1969a, 1969b; Tozzi, 1993). The geomorphological studies suggest the tectonic stability of this area since the Upper Pleistocene. The uplift rate, indeed, tends to be null or very low (between 0.15-0.3 mm/y), as shown by evidence of the Tyrrhenian Sea (MIS 5e, about 123ky BP, see Cotecchia, 2014; Ferranti et al., 2006). Karst processes are particularly important in this area, with a variety of evidence (e.g., dolines, poljes, other karst depressions, caves). They played an important role in shaping the cultural landscape, in different phases of human history (in particular for prehistoric times). Some of the caves from Salento, indeed, are key-sites for the definition of most of the Italian palaeolithic technocomplexes (Palma Di Cesnola, 1993, 2001). The hydrographic network is almost totally endoreic, with poorly hierarchical incisions of modest depth and occasional torrential nature. Both deep and shallow aquifers are the main freshwater reserves of Salento. Some of these give rise to springs along the coasts or at coastal watersheds. Both along the Ionian and Adriatic sides, there are numerous wetlands or marshes, drained in recent times (Cotecchia, 2014; Largaiolli et al., 1969; Martinis, 1970; Martinis and Robba, 1971; Rossi, 1969a, 1969b). This kind of evidence is well-attested in the territory of Roca Vecchia. Some authors link the name of Grotta Poesia to a local Greek dialect root (“posìa” means “to drink water’’), referring to the possible presence of a freshwater spring (quoted in: Delle Rose and Parise, 2005). As mentioned, an important hydrological basin is present a few hundred meters beyond the wall of the protohistoric settlement (the Tamari Basin, see Auriemma 2004). The stretch of the coastline along the investigated area evolves as a cliff sloping towards the NW. The Pliocene fine calcarenites of the bedrock are particularly friable. This produces rhythmic and massive collapses of the cliff with a consequential erosion of the coastline, including the promontory where the site of Roca Vecchia lies. The magnitude of these collapses can be measured in a historical time scale (as testified by the remains of the ancient/Hellenistic settlement surviving on what are now isolated rocks in the sea). A large strip of plain land extends behind the Tamari Basin, bordered by a morphological terrace parallel to the coastline (distant from it about 15 km). The edge of this terrace is set at an elevation of about 30 m and results in a morphological break. It is more pronounced along the SE edge of the area (with an elevation difference of about 15 m) and more nuanced along the NW edge (with an elevation difference of about 5 m). Further toward the hinterland, a vast plain extends at the height of about 50-60 m above sea level.
Roca Vecchia: a Mediterranean hub before history
The landscape around the site of Roca Vecchia represents a privileged vantage point for the analysis of widespread phenomena of mobility in the Mediterranean through time, a theme that in recent times has been at the centre of much debate in archaeology (Aldred 2020; Daniels 2022; Heitz and Stapfer 2017; Kristiansen 2014, Kristiansen et al. 2017; Scharl and Gehlen 2017, see also Tab.1). The excavations carried out at Roca in the last thirty years by a research team from the University of Salento have contributed to reconstructing the history of the site (Table 1) starting from the oldest occupation in the area datable to the Middle Bronze Age (Proto-Apennine, around the seventeenth century. B.C, see Iacono 2019; Scarano 2012). Starting from the MBA, the emergence of fortified sites is attested in the whole of southeastern Italy. The monumentality of these fortifications marks the landscape, making these structures not only functional to defensive needs, but also probably deeply imbued with a complex symbolic meaning. Another important characteristic of fortified sites is the resilience of their communities over the long-term, as these have shown the ability to recover even in the aftermath of violent destructions. Roca Vecchia perfectly epitomizes both features. Here fortifications are extremely monumental, reaching a width of some 25 m at the main gate (Scarano 2012). Also, through the second millennium BC it is possible to distinguish at least two massive destruction episodes at the site: the first at the end of the MBA, i.e. 1450 B.C.; the latter, around 1000 B.C (Pagliara 2005; Scarano 2012, 381). From the MBA onward, Roca Vecchia has shown considerable evidence not only for short scale movement but also for long-range connections. The arrival of small groups of individuals coming from the Aegean – Mycenaean world is essentially witnessed by the arrival of imported ceramics, some of which of the Middle Helladic tradition, datable as early as to the LH I period (see Guglielmino 2006; Guglielmino et al. 2010). A moment of intense contact is recorded in the second half of the second millennium BC, in particular in the Subapennine period. Aegean ceramics, both imported and (especially in this phase) of local production (Jung et al., 2021), increase considerably in number and their use seem to have been connected to banquet practices like those recognised at other sites of the Italian Bronze Age, although probably on a larger scale (Iacono 2015; 2019). In the Iron Age, the site shows notable relations with the Corinthian commercial circuit (Corretti et al. 2017; Merico et al. 2010). As it can be seen, much was known of the site through time, relatively little about the landscape that had to be crossed to move from / to it. The Roca Archaeological Survey began in 2019 by a team from the University of Bologna in order to address this issue.
The Roca Archaeological Survey
The fieldwalking survey
1In agreement with best practices (Attema et al. 2020), the general area to be explored (Figure. 1) has been sampled in advance on the basis of three macro topographical recognisable zones: the coastal one inclusive of the former lagoon of the Tamari Basin (see below and Fig. 1.1), the slopes (Fig. 1.2) and the inner plateau (Fig. 1.3). There is a variation in current land use and occupation density among these areas and one of our initial research questions was how the current situation influences the surface record. At the start our goal was to obtain a general view of the characteristics, conservation and surface visibility of the artefacts in the three above-mentioned areas of the landscape. A detailed study of the coastal area, where Roca Vecchia is situated, was initially not our main focus. In general terms, the human impact in this area produced both urbanisation (in particular along the coast and nearby the municipality of Melendugno) and intensive agricultural exploitation. Cultivation includes mainly arable/arboreal crops and olive groves. The latter, in particular, and despite have been decimated by a disease, represents the almost totality of local agricultural production, and seems to be concentrated on landscapes with an outcropping rock substrate and/or very limited soil depth. These areas have been subjected to (recent) heavy agricultural activities, with massive earthworks, rock crushing and ploughing assisted by heavy mechanical equipment. Initially, areas to be explored within the macro areas identified (see above) have been randomly selected (outlined in Figure. 2) except for the most internal area (Fig. 2.1) located in proximity of the two known dolmens (Gurgulante and Placa potentially datable to the BA; Iacono 2019). Although the project has paid attention to all the periods from Prehistory untill today, at the same time it has represented a chance to explore the “hidden landscapes” (Bintliff et al. 1999) of Bronze Age southern Italy. Indeed, based on our knowledge of the geographical area, the remains pertinent to this period are normally characterised by a low concentration of finds. However, as the exploration made progresses, we recognised that the landscape close to the coast and the main settlement showed a clear difference from the hinterland area, as it exhibited a considerable find density for every period. This can be clearly glanced comparing the raw count of artefacts from the 2019 season, which was centred in the interior areas and yielded only some 2500 objects, with those of the year 2020 and 2021. These latter focused on the coastal area and on the surrounding of the former Tamari lagoon and yielded more than 10000 and 60000 finds, respectively. Such a stark difference demonstrates that the landscape around Roca Vecchia shows two different contemporary patterns: a low density and sparsely occupied hinterland along with a coastal stripe that was bulging with evidence of past occupation from late prehistory to post-medieval times.
Field methods and data analysis
In our data-recording workflow, our goal was to document the survey without the aid of paper documentation, using only free (and preferably open-source) software ( Figure. 3b). Two team leaders were responsible for the recording of the limits of the survey units. The points at the corner of each survey unit were recorded using Google’s My Maps application. Influencing factors that affect visibility (weather, vegetation, shadow, ploughing, modern material, rocks) have been recorded on an application purposefully created on five.epicollect.net (a cloud-synced DBMS developed by the Imperial College, see Fig. 3a). The recording was carried out by participants directly through their smartphones but regularly checked by the director and/or team leaders for consistency. The accessible fields within the selected areas have been divided in 20x20m units directly on the field. Such units have been explored following the topographic features of the land (e.g. following the rows of olive trees) recording visibility conditions and collecting archaeological finds separately for each unit. Such division in small units has been facilitated by the general organisation of the property, particularly in the areas explored during the first campaign, normally small properties partially fenced by dry-stone walls, often accompanied by traditional dry-stone constructions (locally known as pagghiare). Our fieldwork has allowed the creation of high-resolution distribution maps of the materials, as well as a considerably detailed evaluation of visibility. At the same time, the intense agricultural use of these plots has affected the level of preservation of the surface record. Usually, the distance between walkers was 5m (a typical distance between rows of olive trees) but a smaller coverage has also been selected (10m of interval/gap) depending on surface conditions and the quantity of finds counted over a strip of 3 units. 2 The interval of 5m was always applied to the areas with presumably good conditions of preservation or where pre-modern material was already evident on the boundaries of the field. Assuming that each walker is able to cover a width of about 2m in front of them (1m to the left and 1m to the right), the coverage then varied between 20 and 40% (being aware of the debate on visual coverage this percentage is thus considered only an approximation, see Banning 2002). The former (20%) sampling is already representative, however given the abundance of the archaeological record of the landscape under study we decided to adopt a 40% sampling strategy in order to have a richer basis for our analysis. In the density maps units with 20% coverage are a minority and the number of potsherds from these has therefore been doubled to make the data comparable with the 40% coverage units. In the maps, interval classes were established separately for each phase using the Jenks natural breaks classification method which seeks to reduce the variance within classes and maximize the variance between classes (Jenks 1967). All the finds have been collected by unit, including post-medieval material. Samples of special materials (special bags) have been collected separately from the materials of the regular coverage, but only in the limited number of cases where these attested the presence of periods otherwise not present in the regular coverage of the unit. As a general rule, it has been decided not to use any discard threshold (based on sherd size) for collecting material. This decision is based on the consideration that in situation of over-abundance of finds, minute prehistoric sherd material tend to be more easily overlooked and left aside. Primary pottery processing has normally been conducted in Bologna also by the means of another five.epicollect.net database purposefully created. Besides the identification of macroscopically recognisable classes of material, the analysis has tried also to assign non-diagnostic undecorated sherd to broad chronological brackets by the means of an assessment of fabric hardness, starting from the basic assumption that older sherds tend to be softer than recent ones. This characteristic is associated with the degree of vitrification of the clays, which is in turn connected to the firing temperature (Whitbread 2017, 204, Lankford 1983). Consequently, it is possible to suggest that older fabrics are normally (although not always) fired at lower temperatures due to the general improvement of the firing process (Ricci 2016, 35-38), and hence, overall, tend to be softer than those of subsequent periods. This consideration is not universally valid, and the attribution of undiagnostic, undecorated sherds is not always easy to implement, particularly for minute material. Despite such limitations, this method represents a useful starting point and can complement traditional criteria based on identification of classes of materials and on shape morphology. Within each category of finds, basic measures (linear length and weight) were collected for the larger artefact in each category, as to allow a quick quantitative comparison.
The Landscape of Roca Vecchia through time
A first report has been published in 2020 but it was uniquely based on preliminary pottery processing and not on specialised analysis of materials (Iacono et al. 2020). The following discussion will present only data related to the periods from early and later prehistory.
Early Prehistory
The earliest phase documented in the territory around Roca Vecchia is represented by evidence of lithic industry, dated to a final moment of Upper Palaeolithic or to the Mesolithic, mainly clustered on the boundary of the Tamari basin ( Figure. 4). The absolute paucity of findings (mostly non-diagnostic) might be due to their average small size which hinders accurate chrono-cultural framing. Limited to the 2019 field season, the sample comprehends only 35 pieces. The fine chert is the most represented raw material (with very rare siliceous limestone, radiolarite and quartzarenite). A large part of the lithic implements (91%) is patinated, as result of a long exposure of these materials to weathering. Evidence of dark coating on one third of findings is also relatable to post-depositional alteration, nevertheless, this kind of alteration is usually due to Fe/Mn precipitation, with presence of water, but in anoxic environment (e.g., Goossens et al., 2015). No other chemical alteration is detectable. The physical alterations comprehend mainly post-depositional fracturing (15 pieces), fire alteration (1 piece), and smoothing of edges (1 piece). This suggests that the modern mechanical stress (e.g., agricultural activities) constitutes the main destructive process affecting the original site, while water-flow processes (with long-distance transport of material from primary position) should be negligible. The clustering of main lithic finds around a relatively small area, near the Tamari basin, appears also consistent with this reading. From a technological point of view, most of the pieces are represented by management flakes and indeterminate fragments. Possible target flakes, retouched tools (8 pieces) and cores (2 pieces) are less represented. In particular, the presence of “short endscrapers” (2 elements), blades (2 elements), and bladelet cores (3 elements, Fig. 4) point to a final Upper Palaeolithic or Mesolithic period. The location of these lithic findings in the nearby Tamari basin seems to reflect a precise settling choice, selecting peri-coastal marsh environments. Other similar Upper Paleolithic and Romanellian/Epiromanellian contexts, indeed, are recognizable along the Eastern side of the Apulian peninsula, such as the sites near Brindisi of Torre Bianca, Torre Canne, Porto Fetente, Gallico I-II and Torre Testa (Punzi, 1968), as well as in a closer range San Foca (just north of Roca, see Ingravallo, 1980), Malapezza (Dell’Anna, 2010), and some others around the Alimini lakes (near Otranto some 10 Km to the south, see Milliken and Skeates, 1989; Piccinno and Piccinno, 1978). This settlement pattern could be the result of economic and behavioral practices shared by a common cultural background. The marsh environment may have acted as an attractor for late-Paleolithic/Mesolithic human groups, due to the rich resources of this kind of biome, during a sharp climate changing phase. This is consistent with the idea of a wide-range economy, during the final Paleolithic and Mesolithic periods, aimed at the exploitation of diversified basins and resources (Tagliacozzo, 2003).
Later Prehistory
After the late Palaeolithic / Mesolithic horizon it seems to be possible to notice a considerable gap in anthropic activities attested. Neolithic traces are ephemeral in the record. Considering both lithics and ceramic artefacts, no material of the sample analysed so far seem to belong to the Copper Age-EBA. This is not entirely surprising as within the area this horizon is often rather ephemeral in the surface record and is usually recognised only in a limited number of contexts (see Cazzella and Recchia 2021). Also, in terms of general distribution of the material ( Figure. 5 Figure. 6), the hinterland seems to be remarkably poor in terms of representation of the various prehistoric periods. A “void” seems to be recognisable, particularly during prehistoric times, while later periods are somewhat better attested. Such a skewed distribution of the evidence demands an explanation. One of the potential factors affecting this trend could be (at least partially) the intense agricultural activity conducted in this zone, particularly around the internal plateau (Fig. 2.1, 2.2) during subsequent periods. Also, erosion might have played an important role in erasing traces of earlier frequentation. Still, as mentioned, the relative sparseness of the surface record from the hinterland is paired with the richness of the material coming from the coastline and the margin of the Tamari basin. This trend is so apparent that has led us to reformulate some of the main assumptions on which the survey was based. The model that represented the benchmark for the interpretation of occupation within the region for late prehistory was based on the results by Recchia and Ruggini (2009, see also Ruggini 2009) on the hinterland of the Brindisi area (immediately to the north). This assumed a two-tier landscape occupation, with hamlets dispersed in the interior accompanied by the attestation of long-lived specialised, fortified settlements (Cazzella 2009) ( Figure. 7). Yet, according to our current knowledge, the landscape around Roca Vecchia seems to preserve very faint traces of such small settlements. Besides unobtrusive material scatters or isolated sherds, the only evidence belonging to this period is the looted chamber tomb recovered not far from one of the known dolmens around the site (see Iacono et al. 2020). The structure bears some resemblances with LBA structures from western Greece as well as to another from Roca Vecchia itself and other sites in the Salento region (Iacono and Guglielmino 2021, Scarano 2021). Beside this structure, a seamless distribution of prehistoric ceramic fragments is evident in a radial pattern surrounding Roca Vecchia. The majority of these fragments are dated to the Bronze Age, with their density gradually diminishing as the distance from the site increases. (Fig. 6). Rather than with the presence of multiple occupation cores, a similar situation seems to be more in line with a unique settlement which has its main nucleus within the peninsula area, but that extends considerably outside the fortifications. Obviously, a similar reconstruction has some limitations, not the least because the material retrieved often cannot be precisely dated and seem to be part of a flat horizon with little possibility to operate more precise distinctions. The spatial distribution of diagnostic finds can be of some help in reconstructing the diachronic dimension of this activity. While well-dated MBA material such as the decorated Apennine fragment (Fig. 6.1, 7.1 with a decoration comparable to Macchiarola 1987, Mot. No. 177a) seem to be spatially located in an area relatively close to the main settlement (but still outside the main walls), other later materials have been found further from the site, within and beyond the area enclosed by later Hellenistic fortifications. This can be evinced, for instance, by the horizontal roll handle (Fig. 6.2, 7.2) a feature that is more frequently attested from the RBA onward (Cocchi Genick 2004, 78, also note the lack of such feature in the MBA levels and its attestation in those dating to the RBA: see Pagliara et al. 2008, no.5, 27; Scarano 2012). Likewise, the only Aegean type vessel fragment identified so far in the landscape (Fig. 6.3, 7.3) belongs to a vessel dated in all likelihood to LH III B-C early/middle. The date is based on the decoration that includes a panel motif (FM 753) which is likely dated to this period, particularly in the light of what seems to be the ending part of a chain of quirks in central position within the motif. This feature that is attested in Aegean type productions at the main site (e.g. Coluccia 2010). Incidentally, this period corresponds to the highest attestation for Aegean type pottery at Roca Vecchia (Guglielmino et al. 2017; Iacono 2019), which makes the dating to such a horizon more probable. Such situations might tentatively be considered indicative of a gradual expansion over a long horizon that goes from at least the 15th cent. BC (according to the radiocarbon dates obtained from layers with Appennine material at Roca Vecchia, see Scarano 2012), toward at least the Recent Bronze Age (up to the 12th cent. BC). Naturally enough, we are well aware that the surface record does not mean habitation, and therefore we should pay attention not to equate the presence of anthropogenic traces with an expansion of the dwelled area tout court. However, it is possible to safely assert that the range of human activities centred around the main settlement of Roca considerably extended beyond the customary 3-5 ha. traditionally assigned to it.
Mobility at Roca Vecchia in Early to late Prehistory
In this section we will sew together the many threads presented so far, showing how the data of the Roca Archaeological Survey can help us to understand patterns of mobility and their effect in the landscape. The earliest phase recorded by archaeological evidence is framed between a final moment of the Palaeolithic and the Mesolithic, testifying the presence of people with a highly mobile life-style. In this chronological horizon, a wide-range subsistence economy, exploiting not only the large game, but also small preys and marine resources, is attested in the region (Tagliacozzo 2003). This strategy could be related to a contraction in the extension of the play-radius, with a consequent extensive exploitation of the local resources and a possible characterization of the foraging mobility (characterized by frequent moving of the camp for relatively short range). As collateral consequence, areas like the marshy environment could have acted as main attractors for the camps, given the richness of exploitable resources therein. Secondly, the intrinsic presence of different siliceous raw materials is robust evidence of long-range mobility. The raw materials identified in the lithic assemblage, indeed, are completely absent in Salento (e.g., Spinapolice 2012 and reference therein). At this stage of the research, two possible hypotheses can explain the presence of these raw materials: 1) a direct trip to sources far more than 100Km from the communities living around the Tamari basin aimed to recover these raw materials; or 2) a collecting mediated by a number of shorter down-the-line exchanges reaching the closest sources. As we move to later periods, the contrast between a void hinterland and a populated stripe around the coast and the Tamari basin emerges quite clearly. Both patterns are likely far from being unrelated, as it is possible to suggest that, beyond distortions in the representation of the surface record, there was a dependency relation between these two distinguished landscapes: the interior is depleted because consistently, through time, the coast is attracting the majority of human activity. No doubt this “attraction” entailed also various forms of movement, primarily in the short range. The picture can be reliably reconstructed only starting around the mid second millennium BC. Such a configuration of spatial engagements with the landscape could have been potentially the outcome of the relocation of people and groups close to the peninsula of the main site, a kind of short-range migration from other areas (arguably outside of the surveyed area), potentially due to the increasing nodal role of the Roca in local networks of interaction and its increasing ability to accumulate surplus (Iacono 2019, 156-160). The presence of material showing clear “foreign” features such as Aegean type pottery and Apennine pottery (within Salento these two classes are almost equally rare during the late MBA, see Iacono 2019, Scarano 2006), give us an idea of connections further afield, either direct or (more likely) mediated by the main site.
Conclusions
The variable nature of mobility as attested in the landscape around Roca Vecchia through time allows us to explore broader issues related to the relationship between the human landscapes and mobility on a much broader canvas than what is normally accomplished through studies with a much narrower chronological focus. The analysis of the surface record has allowed to highlight how the scale of mobility has considerably changed through time, from the immediate remits of different environmental niches exploited by local prehistoric communities, to the building up of larger and larger communities that already toward the end of the Bronze Age seem to have been more akin to the “urban” status than what was previously thought. Such communities were, in turn, also the product of much larger networks of mobility and interaction that toward the second half of the second millennium BC were transforming this portion of the Central Mediterranean involving areas as distant as the state societies of the Aegean and the Balkans as well as northern Italy (Iacono 2019). Of all these long-range movements of individuals very little is directly recognisable in the material culture collected in the landscape with only one sherd of Bronze Age Aegean type pottery recovered from the surroundings of the site, a figure that marks a sharp contrast with the abundance of such material from the main site. This can potentially suggest that the circulation of this material was somewhat limited to the area of the main site although further investigations are needed in order to prove this.