1. INTRODUCTION
Recent archaeological and anthropological scholarship on colonization has moved the focus of discussions on imperial expansion from the long-term impacts of colonialism for native communities (such as violence, cultural suppression, territorial transformations, settlement destruction and the relocation of peoples) to addressing the various stages of the colonization process. Especially, there is a growing interest in the ‘contact period’, defined as the initial stage of cultural interaction between native and colonial communities.Footnote 1
These studies stress the heterogeneity and complexity of this process, which varies greatly from context to context, while also highlighting a general trend in how colonial newcomers, for their own benefits, may have initially adapted to unknown territories and adopted autochthonous land-use practices to avoid a major restructuring of the landscape. Most conquered territories were not an empty, pristine, untouched wilderness — but were modified landscapes with settlements, infrastructures and services constructed by the native populations for their survival and growth (Ingold, Reference Ingold1993; Kolen, Renes, Bosma, Reference Kolen, Renes, Bosma, Adri van den Brink, Bruns, Tobi and Bell2016). In some portions of the landscape colonists may have thus encountered pre-existing land parcels, drainage systems and modification to the forests, vegetation and soil carried out by other inhabitants (Coughlan and Nelson, Reference Coughlan and Nelson2018). Rather than experiencing immediate destruction, these native landscapes resiliently survived through the inheritance, reuse and/or readaptation of the environment and land-use methods in later colonial periods.
The early conquest of the Americas by Columbus (1492–1502) represents a useful parallel case study that can be examined to understand potential Roman colonization practices. Recent research combining archaeological studies, historical sources and ethnographic surveys (Hofman et al., Reference Hofman, Mol, Hoogland and Valcárcel Rojas2014, Reference Hofman, Hung, Malatesta, Jean, Sonnemann and Hoogland2018; Hofman, Valcárcel Rojas, Ulloa Hung Reference Hofman, Valcárcel Rojas, Ulloa Hung, Beaule and Douglass2020; Herrera Malatesta, Reference Herrera Malatesta2022) illustrated that the Iberian villages established on the island of Hispaniola (modern Dominican Republic and Haiti) after Columbus's first voyage targeted areas that had been previously abandoned or were unused by local communities, but which were easily defended. These areas were not particularly suitable for agriculture but, interestingly, were located at the margins of fertile rural land farmed by indigenous peoples or nearby previous settlements. These Iberian villages were probably established in these locations for the strategic access to indigenous peoples’ resources, building material, infrastructures and land. In contrast to traditional narratives (cf. Kulstad-González, Reference Kulstad González2020), the early stages of the colonization of the Americas by the Portuguese and Spanish Empires probably did not cause an immediate, drastic and costly reorganization of territory. Instead, colonization was underpinned by an opportunistic strategy whereby colonial institutions maintained and exploited the indigenous knowledge of the landscape and the pre-existing structures and infrastructures for their own purposes.
The tendency in human behaviour to reuse what is already available, in terms of (agricultural) know-how, land-use and landscape infrastructures, is in line with the principles of least effort (Zipf, Reference Zipf1949) and ecological inheritance (Foster et al., Reference Foster, Swanson, Aber, Burke, Brokaw, Tilman and Knapp2003; Coughlan and Nelson, Reference Coughlan and Nelson2018) which manifest in our daily lives, history and nature. These concepts may cast a new light on the logic behind the settlement choices made during early colonial periods. Humans generally opt for the optimal strategy, inheriting and taking advantage of readily available resources, building materials and/or structures to avoid the efforts and costs that the destruction and subsequent reconstruction would entail. This may occur, for instance, during the formative phases of ‘start-up’ enterprises — such as in the first colonial explorations and pioneering formations of colonies — when the initial resources available for investments are usually limited or new to the colonizers, and the lack of a reference against which to evaluate decisions hampers the long-term foresight of potentially risky actions. This may apply to situations where the local contexts are unfamiliar and social and environmental factors are unknown or unpredictable, such as during migrations in unexplored territories.
Recent reconsiderations of the conventional historical accounts in Roman archaeology have posited new observations that are in line with aspects of these anthropological concepts. Scholars have argued that the traditional image of Latin colonies during the early phase of mid-Republican colonization in central-southern Italy (late 4th–3rd century BC) must be corrected. The idea of agricultural city-states with urban centres and rigidly divided landscapes (typically by centuriation), created through resettlement and land reforms that swept away native landscapes, does not fit with the archaeological record.Footnote 2 The careful reinvestigation of archaeological survey evidence in the early colonial landscapes of Latin colonies founded before the Second Punic War has revealed that clustered settlements (e.g. agricultural villages) targeting specific ecological niches and separated by unoccupied land prevailed, instead of dense and geometrically ordered settlement distributions (e.g. mononuclear farms) placed according to regular land divisions.Footnote 3 Scholars have thus begun to question the logic behind such clustered settlement patterns.
1.1. A NEW INTEGRATIVE SETTLEMENT MODEL FOR EARLY ROMAN COLONIZATION
A new model has been proposed for one colonial territory in Southern Italy to explain the clustered pattern present (Casarotto, Pelgrom, Stek, Reference Casarotto, Pelgrom and Stek2019), which can be more widely tested using other case studies to assess its plausibility on a larger scale. Based on the reassessment of legacy survey data (Casarotto, Pelgrom, Stek, Reference Casarotto, Pelgrom and Stek2016; Casarotto et al., Reference Casarotto, Garcia Sanchez, Stek and Pelgrom2021) and new fieldwork results (Pelgrom et al., Reference Pelgrom, Marchi, Cantoro, Casarotto, Hamel, Lecce, García Sánchez and Stek2014, Reference Pelgrom, Lecce, García Sánchez, Stek, De Siena and Giammatteo2016), it was observed that colonial period settlements in the territory of the colony of Venusia (southern Italy, established in 291 BC) were clustered around the newly founded colonial town and within marginal niches that were not particularly favourable for agriculture. These niches bordered the rural catchment areas of native farming villages, many of which continued to be settled after the conquest during the colonial period. From these places the rural infill possibly infiltrated the native landscape and tapped into existing resources, organically absorbing and reusing previous settlements, their rural networks, roads, infrastructures, agricultural and pastoral practices and land-division systems (Pelgrom, Reference Pelgrom, De Ligt and Northwood2008, Reference Pelgrom2018; Stek, Reference Stek, Düring and Stek2018). Stek has proposed another example of a deviating colonial landscape — the colony of Alba Fucens (303 BC) in the Apennines — where villages around the Fucine Lake and transhumance routes dominated the settlement pattern. These villages were apparently geared towards lacustrine and transhumance economic opportunities rather than the expected agricultural economic framework, with colonization in this area clearly developing ‘beyond the Romanising agro-town’ (Stek Reference Stek2009, Reference Stek, Düring and Stek2018).
Coming back to Venusia, the pre-existing structured landscape created by pre-Roman communities over time was not erased nor immediately reorganized by colonial settlers. This is probably due to the heavy investment and cost its destruction would have required (Pelgrom, Reference Pelgrom, De Ligt and Northwood2008). For agricultural purposes and other necessities, it was presumably easier (and certainly less costly) for the Roman colonial community to exploit the landscape as it was without major changes, at least initially. This can be plausibly argued as digital geospatial analysis has demonstrated that the organization of pre-colonial land-use was the most influential factor in settlement strategies during the Hellenistic period, instead of other physical environmental conditions in the territory of Venusia (Casarotto, Pelgrom, Stek, Reference Casarotto, Pelgrom and Stek2019). In the past, scholars have lent great importance to natural constraints — in particular to the favourable soils for types of agriculture — and have downplayed the influence of the territorial organization of native communities as being irrelevant or having been erased by the Roman conquest.Footnote 4 In contrast, geospatial analysis on data patterning has indicated that settlements in the territory of Venusia during the early Hellenistic and colonial period adapted to a pre-existing situation by preferentially occupying the available (unused or abandoned) space, regardless of its natural properties.
The initial organization of the settlements in Venusia during the colonial period was more sustainable than previously assumed, as it complemented, rather than replaced, the existing territorial organization by adaptively filling the relatively scarcely settled portions and inheriting other portions of the already-structured rural landscape, including some villages. While it is impossible, using survey ceramic evidence, to imply a direct link between the new settlements of the Republican period and an influx of Roman colonists (Dores Cruz, Reference Dores Cruz2011), it is possible to formulate hypothetical scenarios for the observed settlement distribution by assuming that the settlements during the colonial period were inhabited by Roman colonists (these colonial groups were also diverse, having various ethnical and geographical origins; see Bradley, Reference Bradley, Bradley and Wilson2006).
Following this observation, several scenarios are possible. The first is that the genocide or expulsion of pre-existing communities was implemented by the aggressive Roman colonial power. Livy (9.45.17) posits that genocide accompanied the conquest of the wider area surrounding Alba Fucens. Another possible scenario is that indigenous peoples may have been relocated to marginal areas and that the colonial settlers occupied their rural properties, including some villages. Alternatively, local native communities may have been allowed to continue inhabiting their villages, but were forced to share a quota of the food production obtained from farming the surrounding countryside with the new colonial community living nearby in clustered settlements.
A hypothetical scenario explored further in this article is that the initial Roman expansion was an opportunistic more than an oppressive phenomenon of integration, driven by and evolving from the variable factors of local contexts. Roman expansion could have tapped into a wider, pan-Mediterranean phenomenon of rural infill, demographic growth and agricultural intensification also in marginal areas, a phenomenon that had already started in the early Hellenistic period (4th–early 3rd century BC), thus in pre-conquest times, and can be observed archaeologically in both the territories left uncolonized and those that were colonized.Footnote 5 In this scenario, Roman conquest and colonization are seen as an integral part of this large-scale Mediterranean settlement trend that began earlier, and are also conditioned by this trend. According to this integrative scenario, the colonial community, when settling a new territory opportunistically, could benefit from the nearby pre-existing assets by following an adaptive (clustered) rural settlement strategy (in which conflicts, frictions, competitions and negotiation for resources between colonial and non-colonial communities/elites that shared the same contested territory could be expected) that led over time to an organic integration and reuse of the native landscape without major restructuring.
Another scenario is linked to the limits of the survey methodology. The empty zones, and thus the resultant clustering in the survey record, instead of indicating scarcely settled niches might be due to methodological biases (Banning, Reference Banning2002: 39–79; Van Leusen, Reference Van Leusen2002; Attema et al., Reference Attema, Bintliff, van Leusen, Bes, de Haas, Donev, Jongman, Kaptijn, Mayoral, Menchelli, Pasquinucci, Rosen, García Sánchez, Gutierrez Soler, Stone, Tol, Vermeulen and Vionis2020), including the inability of archaeologists to recognize in the field certain site types, such as small, simple farms, through traditional pedestrian survey.Footnote 6 However, recent research has shown that the effects of survey biases, such as surface visibility (Casarotto et al., Reference Casarotto, Garcia Sanchez, Stek and Pelgrom2021) or geomorphological processes (Casarotto et al., Reference Casarotto, Stek, Pelgrom, van Otterloo and Sevink2018), are primarily limited to finer scales of analyses (e.g. intra-site analysis in small-scale off-site surveys) and cannot explain the overall, ubiquitous, large-scale cluster patterns that manifest in a majority of colonial and non-colonial territories.Footnote 7
Independently of whether this clustered settlement pattern can be explained by past oppressive or opportunistic reasons, or by methodological difficulties in modern recording techniques, it is worth assessing whether the new integrative model proposed works as an additional potential scenario in explaining colonial settlement strategies of the western Mediterranean. This is applied through a GIS analysis of the settlement patterns underlying survey datasets collected in early Roman colonial regions of different countries. Previous regional surveys, including the research conducted in Venusia, have indicated clustering based on the survey datasets, which may provide grounding for this new integrative model of Roman expansion, challenging the long-held theory that Roman conquest and colonization entailed the destruction of native landscapes.
2. DATA
The selection of case studies was based on the types of available and accessible survey data. To implement a sound comparison, we attempted to limit the possibility that differences in site patterning between datasets was due to differences in recording techniques. Datasets were selected that had been collected by previous teams using the same or comparable survey methods in the same time-frame of the history of survey methodology (i.e. same level of technological sophistication). Moreover, survey results were accessible (online) or published in a way that allowed for precise GIS digitization and thus reuse. As the aim of this study was to test the new model on a supranational scale using a western Mediterranean perspective, case studies from different countries affected by Roman expansion and colonization were selected.
The territories of the towns of Cosa (modern Ansedonia) in central Italy and Tarraco (modern Tarragona) in northeastern Spain (corresponding to survey projects A and B in Fig. 1) were selected for analysis. Cosa (273 BC) and Tarraco (218 BC) were among the earliest territories to be conquered by Rome in the Italian and Iberian peninsulas (from the 3rd and 2nd centuries BC respectively) and are characterized by similar settings given their locations in key positions on the Mediterranean coast (Brown, Reference Brown1980: 58–9; Celuzza and Regoli, Reference Celuzza and Regoli1982: 44; Moreno Escobar, Reference Moreno Escobar, Cristilli, Gonfloni and Stok2020). The survey project results were published in Carandini et al., Reference Carandini, Cambi, Celuzza and Fentress2002, and Carreté, Keay, and Millett, Reference Carreté, Keay and Millett1995, as well as online in the Fasti Online Survey portal (Carandini et al., Reference Carandini, Cambi, Celuzza and Fentress2012) and the Archaeological Data Service (Keay and Millett, Reference Keay and Millett2003). Survey research undertaken later in Tarraco (Prevosti and Guitart, Reference Prevosti and Guitart2011) was not considered in this paper as it is only partly accessible (the volume with the full dataset is forthcoming, Schneider, Reference Schneider2017: 15). Moreover, since the dating of land-division systems is notoriously difficult to establish (Celuzza and Regoli, Reference Celuzza, Regoli and Carandini1985: 49–53; Pelgrom, Reference Pelgrom, De Ligt and Northwood2008, Reference Pelgrom2018; Rodríguez-Antón, Magli, González-García, Reference Rodríguez-Antón, Magli and González-García2023), the previous hypothetical reconstructions of centuriation systems were not considered in settlement pattern analysis.Footnote 8
Location of the survey projects in the region of Cosa (A, based on Carandini et al., Reference Carandini, Cambi, Celuzza and Fentress2002) and Tarraco (B, based on Carreté, Keay, Millett, Reference Carreté, Keay and Millett1995) in relation to modern major cities in Italy and Spain. Basemap: ESRI World hillshade.
Systematic field-walking survey projects were conducted in both regions during the late ’70s, ’80s and early ’90s by teams that belonged to the same emerging Anglo-American survey school. These surveys used comparable recording techniques and levels of methodological sophistication.Footnote 9 The meticulous publication of the results of these two projects is outstanding, allowing for the reuse of their datasets for further analyses, such as the analysis presented in this study (Witcher, Reference Witcher2008; Casarotto, Reference Casarotto2022). Both projects applied an intensive survey strategy that consisted of systematically walking within all accessible and visible field units with three–five people spaced 5–20 metres apart, who recorded the locations of notable concentrations of archaeological materials (e.g. pottery sherds), structures and off-site materials onto topographic maps. The projects also implemented a sampling strategy based on 1 km-wide transects located every 5 km in the territory, with larger portions being investigated around the main centres in the ager Cosanus (e.g. the Valle d'Oro facing the colonial town). These transects covered 135.3 and 53.7 sq. km of the hinterlands of Cosa and Tarraco. In the Tarraco survey project the precise extent of the area walked within the transects was provided and corresponded to 11.3 sq. km.
In both projects the term ‘site’ was used loosely when structural remains and scatters with particularly high material densities were found on the surface of the terrain.Footnote 10 These material concentrations may have indicated that archaeological sites, such as ancient settlements (e.g. farms, villages and villas), are located underneath or in the surroundings of their discovery. The dating of sites was based on chronological classifications of the ceramic that was observed in the field and studied in the laboratory. In both territories, the clustering of sites during the pre-Roman (4th–3rd century BC) and conquest/colonial periods (3rd–1st century BC) was observed by previous studies.Footnote 11 In this paper we assessed whether this clustering during the Hellenistic period appeared randomly across the landscape or preferentially targeted areas that would ease the access to or use of legacy resources, settlements and infrastructures.
3. TESTING THE HINTERLANDS OF COSA AND TARRACO
Previous settlements and agro-pastoral investments (e.g. drainage, terracing, road and land-division systems; vegetation and forest clearance; soil improvements) may have been seen as an opportunity for new settlers, thus influencing subsequent rural patterns (Verhagen et al., Reference Verhagen, Nuninger, Bertoncello, Castrorao Barba, Campana, Scopigno, Carpentiero and Cirillo2016). Operating under this assumption, we described critical zones within the landscapes around Cosa and Tarraco whose clustered settlement distributions in the pre-Roman and conquest/colonial periods could fit the new integrative settlement model. To identify these zones, we conducted a point-density analysis in ArcGIS and focused on the areas that displayed high and localized site densities (i.e. point clusters) or nucleated settlements interpreted by the survey teams or other scholars as villages (technical details are explained in Casarotto, Pelgrom and Stek, Reference Casarotto, Pelgrom and Stek2016).
For the density analysis in the region of Cosa we considered the Hellenistic settlement sites (200 sites in total, dating c. 350–50 BC) located in the survey transects that covered the reconstructed hypothetical territory of the colony (Cardarelli, Reference Cardarelli1924–5). The same methods and criteria were applied to Tarraco, with a total of 44 Hellenistic sites identified (see Prevosti and Guitart, Reference Prevosti, Guitart, Prevosti and Guitart2010; Houten, Reference Houten2018: 248–50 for the hypothetical colonial territory) (Table 1). In general, site density was found to be slightly higher in the ager Cosanus both within the survey area and the site agglomerations. When we considered only the areas that had been pedestrian surveyed (for Tarraco this spatial information was provided in the survey publication and online: Carreté, Keay, Millett, Reference Carreté, Keay and Millett1995), or were most likely to have been pedestrian surveyed (for Cosa this was determined by excluding the forested, steep and urban zones and considering only the farmed land through a GIS simulation), we observed similar density patterns and comparable inter-distances between villages and localized clusters in the two survey areas.
The percentages of rural territories as defined by different settlement densities (see also Fig. 2 and Fig. 4). In total 200 (for the region of Cosa) and 44 (for the region of Tarraco) Hellenistic settlement sites were studied. The sites were identified by the survey teams in the survey sample area (i.e. within the survey transects).
What was different, however, was the size of the settlement sites within the clusters and the sizes of the possible villages. In general, larger sites were identified in the landscape around Tarraco than Cosa. For example, material scatters classified as villages in Tarraco were identified as 6–10 ha in size, and in Cosa of 1–3 ha. Small settlements in Tarraco were measured as c. 0.3–3 ha in size, whereas in Cosa a settlement site was considered small if the material scatter size measured less than 0.2 ha. This remarkable difference in surface scatter sizes may reflect the differences in rural settlement types, the variable impacts of modern mechanized agriculture (and differences in surface material preservation) or different density thresholds adopted by the two survey projects in the field for site identification and off-site and background materials (see above) (Attema and Schörner, Reference Attema and Schörner2012). Based on the available legacy data, it was not possible to estimate reliably the effects of these factors on site classification. If resources become available in the future, targeted field visits could examine this specific aspect (e.g. Seubers and Tol, Reference Seubers and Tol2016; Casarotto et al., Reference Casarotto, Garcia Sanchez, Stek and Pelgrom2021). To navigate this obstacle, we focused on spatial settlement trends as displayed by point-density distributions rather than by size (for the same approach, see Casarotto, Pelgrom, Stek, Reference Casarotto, Pelgrom and Stek2016).
3.1 THE REGION OF COSA
In the survey transects more distant from the city, ten site agglomerations were identified through point-pattern analysis (Fig. 2). Inside or near these habitation clusters, six villages were recognized by the survey team: two villages were founded during the early colonial period (3rd century BC) and four villages had already existed during the previous Etruscan period in the 4th century BC and were continuously occupied after the conquest in the 3rd century BC. Surrounding these villages, a few isolated small settlements (potentially farmsteads) were located. These villages had a minimum inter-distance of 2 km and a maximum inter-distance of 5 km from their nearest neighbours (Casarotto, Pelgrom, Stek, Reference Casarotto, Pelgrom and Stek2016). The zone around the town of Cosa (c. 2 km radius) was largely devoid of early colonial sites and may have corresponded to land that was directly exploited by the city.
Point-density analysis of the Hellenistic settlements in the survey sample area of the Ager Cosanus (marked by the white line). Each cell (measuring 20 x 20 metres) of the resulting raster density surface received a value (d) indicating the number of sites located in a circle of 1 sq. km. The labels correspond to the site codes (UT codes) reported in the survey publication (Carandini et al., Reference Carandini, Cambi, Celuzza and Fentress2002). Basemap: shaded relief (DEM 10 m TINITALY 1.1, Tarquini et al., Reference Tarquini, Isola, Favalli, Battistini and Dotta2023).
A wide site agglomeration dated to the Republican period was observed in the southwestern portion of the Valle d'Oro, about 3 km from the city of Cosa (A in Fig. 2). Some scholars have interpreted this concentration as farms regularly distributed according to land-division systems,Footnote 12 while other scholars have posited this as a possible nucleated settlement or village (Pelgrom, Reference Pelgrom, De Ligt and Northwood2008; Casarotto, Pelgrom, Stek, Reference Casarotto, Pelgrom and Stek2016). Whether this agglomeration represents an agricultural village, a loose group of dispersed farms or a combination of various settlement types (settled contemporaneously or at different intervals during the Republican period) (Chisholm, Reference Chisholm1968; Roberts, Reference Roberts1996), it is interesting that it occurred in the area within the ager that shows a remarkable use in the Archaic period (6th–5th century BC) and an abandonment or scarce use in the 4th century BC before the arrival of the colonists (Carandini et al., Reference Carandini, Cambi, Celuzza and Fentress2002: 291–3; Attolini et al., Reference Attolini, Cambi, Celuzza, Fentress, Pasquinucci and Regoli1982; Celuzza and Regoli, Reference Celuzza and Regoli1982) (Fig. 3).
Republican period settlement agglomeration corresponding to cluster A in Fig. 2. Red dots are settlement sites: full dots probably date from the 3rd century BC (Carandini et al., Reference Carandini, Cambi, Celuzza and Fentress2002: 117), while empty dots probably date from the 2nd century BC (Carandini et al., Reference Carandini, Cambi, Celuzza and Fentress2002: 162). The buffer sizes have been calculated based on the dimensions of the site types reported in the survey publication (Carandini et al., Reference Carandini, Cambi, Celuzza and Fentress2002: 59). The yellow polygons and dots indicate Archaic-Etruscan period settlements (Carandini et al., Reference Carandini, Cambi, Celuzza and Fentress2002: 84–5). The village codes correspond to those reported in the survey publication (Carandini et al., Reference Carandini, Cambi, Celuzza and Fentress2002: 379–409; Carandini et al., Reference Carandini, Cambi, Celuzza and Fentress2012). The location and size of Village 3 (located on the top of the Settefinestre hill, with eroded material having been washed downhill, explaining the elongated shape in the figure) is described in Celuzza and Regoli, Reference Celuzza, Regoli and Carandini1985: 57 and Celuzza, Reference Celuzza and Ricci1985. Villages 1 and 2 are described in Celuzza and Regoli, Reference Celuzza and Regoli1982: 36; Attolini et al., Reference Attolini, Cambi, Celuzza, Fentress, Pasquinucci and Regoli1982: 369; Carandini et al., Reference Carandini, Cambi, Celuzza and Fentress2002: 291–2. Basemap: shaded relief (DEM 10 m TINITALY 1.1, Tarquini et al., Reference Tarquini, Isola, Favalli, Battistini and Dotta2023) and 20 m contour lines.
This preference in the Republican period to target marginal, abandoned or scarcely settled zones within the previous settlement system has also been observed in Venusia in southern Italy (Casarotto, Pelgrom, Stek, Reference Casarotto, Pelgrom and Stek2019) and in other zones of the western Mediterranean affected by Roman conquest and colonization (Nuninger et al., Reference Nuninger, Verhagen, Bertoncello and Castrorao Barba2016; Verhagen et al., Reference Verhagen, Nuninger, Bertoncello, Castrorao Barba, Campana, Scopigno, Carpentiero and Cirillo2016). The Monte Alzato and Poggio Settefinestre area in the Valle d'Oro (Fig. 3) may comply with this type of adaptive and organic rural settlement patterning. Abandoned, unoccupied settlements and agricultural land may have provided building materials and infrastructures for reuse to subsequent communities (e.g. roads and drainage systems). Previous investments in land and soil improvements by Iron-Age/Etruscan communities in the territory surrounding their villages (probably located in the Monte Alzato and the Settefinestre hills)Footnote 13 may have been an opportunity for successive settlers to cluster strategically in this area and utilize, integrate and develop the legacy land-use heritage (see discussion in Pelgrom, Reference Pelgrom, De Ligt and Northwood2008, Reference Pelgrom2018). While the Archaic-Etruscan villages may not have been occupied, new Republican settlements may have been created in their surroundings within previously exploited areas. This is the same area where major Roman villas were constructed from the 2nd century BC, thus attesting to the possibility of a longer continuity in land use.
3.2 THE REGION OF TARRACO
Five localized, high site densities were identified by the survey team, one of which was classified as a probable village (Fig. 4). Near to these densities a couple of isolated small settlements (potentially farmsteads) were also located. These nucleated settlements had a continuous occupation from the Iberian period to the late Republican period (5th–1st century BC). In general, the entire landscape displayed remarkable settlement continuation (Carreté, Keay, Millett, Reference Carreté, Keay and Millett1995: 241–52, 272–82). These habitation clusters had an inter-distance of 6–8 km (i.e. 3–4 km catchment radii), which is a typical value for a village-based socio-ecological settlement model (Bintliff, Reference Bintliff and Barker1999, Reference Bintliff, Bailey, Charles and Winder2000, Reference Bintliff, Favory and Nuninger2009), and exhibited a clear preference for dominant positions in the landscape (Carreté, Keay, Millett, Reference Carreté, Keay and Millett1995: 245). The clusters were located on less fertile soil, whereas the surrounding landscapes had more fertile soil for agriculture (Carreté, Keay and Millett, Reference Carreté, Keay and Millett1995: 244–6). This preference to cluster in less productive portions of the landscape and to travel to the agricultural fields may indicate a land-use pattern that is typical of a multiple-core nucleated settlement system (Pelgrom, Reference Pelgrom, De Ligt and Northwood2008), or a tendency to settle scarcely used portions of the landscape to ensure control and increase productivity during episodes of increased population growth (Terrenato, Reference Terrenato, Alcock and Osborne2007; Stek, Reference Stek and DeRose Evans2013). It was also noted that the zone around the town of Tarraco (c. 4 km radius) was largely devoid of sites and may have corresponded to land that was exploited directly from the city (Carreté, Keay, Millett, Reference Carreté, Keay and Millett1995: 251).
Point-density analysis of the Hellenistic settlements in the survey sample area of the Ager Tarraconensis (all survey transects are within the supposed ager, the limits of which are not shown in the figure as they extend beyond the investigated area). Each cell (20 x 20 metres) of the resulting raster density surface received a value (d) indicating the number of sites located within a circle with an area of 1 sq. km. The perimeter of the sites as reported in the survey publication (Carreté, Keay, Millett, Reference Carreté, Keay and Millett1995) is displayed by the red lines. Basemap: shaded relief (DTM 25 m of Centro Nacional de Información Geográfica – CNIG).
The lack of observed marked differences in site distribution from the survey results may indicate an ecological inheritance of land-use practices from the pre-Roman to the Roman period, such as in land-settlement strategies, agricultural methods and general territory use around these clustered settlements. It is true that the broad chronological resolution adopted by the survey project for dating these sites did not allow for the identification of the possible intervals of site abandonment (see the discussion in Carreté, Keay, Millett, Reference Carreté, Keay and Millett1995: 57–61). Regardless of who lived within these large settlements and during which moments of this long period, the available evidence for site distribution does not fit comfortably a scenario of drastic landscape restructuring or destruction of the landscape. Rather, the available evidence points to a long-term continuation (or re-occupation) of settlement types (and additionally of the land-use strategies) from the Iberian to the Roman period. It is vital to note that a homogeneous situation for the entire territory of the colony cannot be assumed, and it is likely that diverse settlement patterns may have occurred in the zones of the ager that are not considered in this paper (Prevosti and Guitart, Reference Prevosti and Guitart2011; Palet Martínez and Orengo, Reference Palet Martínez, Orengo, Prevosti and Duran2010, Reference Palet Martínez and Orengo2011).
4. CONCLUSIONS
This paper has shown that settlers during the conquest and early colonial period did not randomly scatter across the landscape, but rather adapted their choices to the pre-existing organization of land use adopted by the native community that inhabited the area. This resulted in a settlement pattern in which large, nucleated settlements prevailed. More analysis is necessary to better understand the actual character of these settlements. Precise dating through excavation, coring and other field techniques of the parcellation and drainage systems that are identified in colonial territories could significantly add to our understanding of land-use systems and their influences on settlement choices over time. Additionally, geophysical survey and excavation at the clustered settlements described in this article are limited or lacking: only excavation or targeted invasive probing of some of these clustered, nucleated settlements (e.g. the possible villages) can confirm whether they were villages, hamlets, villas, or other types of settlement agglomerations, and shed light on the identities and demographics of the people dwelling in these places (Curchin, Reference Curchin1991: 123–5). As it stands, however, according to the analysis conducted in this paper, the adaptive and integrative (but opportunistic) model presented for the Roman conquest and early colonization of the western Mediterranean is plausible: the effects of native land-use legacies may have significantly influenced later Roman period settlement patterns.
Colonial settlers may have conditioned their choices to follow and adapt to pre-Roman land-use organization. The decision to cluster only within certain zones could be explained by the possibility that the land these colonial settlers encountered was not an untouched wilderness offering ample settlement options, but a landscape that was already densely settled (Terrenato, Reference Terrenato, Alcock and Osborne2007; Stek, Reference Stek and DeRose Evans2013) and shaped by native populations for multiple generations, thus constricting their choices. This structured and engineered native landscape, and thus the embedded legacies of previous land-use investments, probably functioned as an underlying footprint that constrained, guided or influenced the design of settlement developments during the later conquest and colonial periods. Based on the observations drawn from this analysis, it is possible to argue that the process of Italic and Iberian settlement in the rural hinterlands of the future colonial towns of Cosa, Tarraco and elsewhere (e.g. the Venusia hinterland in Casarotto, Pelgrom, Stek, Reference Casarotto, Pelgrom and Stek2019) established historical land-use and settlement patterns that influenced the trajectories of socio-ecological landscapes during the conquest and colonial periods. This paper put forward the possibility that Iron-Age and pre-Roman land-use legacies and settlement systems, such as villages and their surrounding (engineered and/or farmed) landscapes, probably served as key anchors in the development of subsequent colonial period settlement processes (for the general idea of Roman colonization targeting pre-existing economic and cultural hotspots, see Stek, Reference Stek, Farney and Bradley2017; e.g. sanctuaries: Stek and Burgers, Reference Stek and Burgers2015).
Previous settlement occupation represented both an ecological opportunity and constraint for successive settlement systems, and thus was one of the main anthropogenic factors influencing Roman conquest and colonization. If this interpretation is valid, the severity of the initial interaction between native and colonial communities should be downplayed: previously built environments were not always immediately destroyed by Rome but rather opportunistically exploited, maintained, adapted and developed during the conquest and colonial periods. The legacy landscapes constructed over centuries by native communities populating the western Mediterranean constituted an important socio-economic asset for the expansion of the Roman empire, offering rural resources, services and infrastructures (e.g. settlements, roads and land parcellations) that may well have been opportunistically reused by the Roman agrarian economy both in Italy and in the provinces.
Acknowledgments
This work was supported by the Cultuurfonds (Regato Fund) and was conducted in the framework of the project The Impact of Roman imperialism in the West (University of Groningen, KNIR-Royal Netherlands Institute in Rome).
Open access
