Shift in landscape use strategies during the transition from the Bronze age to Iron age in Northwest Serbia

2.1 Viewshed in archaeology

The noun viewshed implies the natural environment that is visible from one or more viewing points, as described in Merriam-Webster dictionary (Merriam-Webstern.d.), which is an aspect of great interest for research in landscape archaeology and understanding spatial relationships between sites. Viewshed analysis, also known as visibility analysis or LOS analysis, is a geospatial technique which determines visibility between a point of interest and its surroundings. Nowadays, this is achieved using GIS software by tracing the LOS from a viewpoint, across a DEM to every cell of the raster system. A simple Boolean operation then calculates whether the cell exceeds the height of the viewpoint, giving it a binary result of either visible (1) or not visible (0) [6–8]. The resulting map illustrates the unobstructed view of the observer from the selected vantage point which provides valuable insight to the researcher on site visibility, intervisibility between sites, and visual control of territory.

Observations made from these maps allow drawing conclusions on selection strategy in the past. Namely, for prehistoric strongholds – such as Bronze Age and Iron Age hillforts, fortified settlements, and other defensive or high-status enclosures in Europe – visibility was often a key strategic factor. Many of these sites occupy prominent hilltops or ridges, presumably to command extensive views of the surrounding terrain or to be highly visible themselves. Viewshed analysis allows archaeologists to quantitatively evaluate these assumptions. Modelling what could be seen from a stronghold allows assessment of its defensive surveillance capabilities (e.g. the ability to spot approaching threats) and its visual prominence in the landscape [9–11]. Through cumulative viewshed analysis, these factors point to locational criteria in hillfort construction, which were often built in the most visually prominent parts of their regions [12]. On the other hand, visibility analysis may also provide an outlook for a more symbolic purpose of a dominant location of a structure, as proposed in the case of Tholos A at Apesokari (south-central Crete, Greece), which figuratively demonstrates territorial rights on overlooked lands [13].

Availability of GIS software, DEM maps of large areas, and other landscape data, like Digital Terrain Models (DTMs) derived from LiDAR (Light Detection and Ranging) surveys, facilitated growing number of researcher topics associated with visibility, intervisibility, and spatial networking patterns in the past several decades. Such studies underscore the importance of viewshed analysis in testing hypotheses about why prehistoric communities chose certain locations for their building sites, most prominently strongholds. In short, visibility studies provide a window into past human experiences of landscape and contribute to debates on defence strategies, communication, and display of power in prehistoric Europe. The remainder of this review provides a historical overview of GIS-based viewshed analysis, surveys the principal methods and algorithms, and highlights key case studies – with a focus on European prehistoric strongholds – before discussing recent technological advancements.

2.2 Historical development of viewshed analysis in GIS

As one can assume, visibility studies predate widespread adoption of computer technology in archaeological research, since the notion of view from sites always attracted the attention of scholars. A comprehensive overview of the early visibility studies, both non- and GIS-based, was presented by Lake and Woodman [14] in an attempt to illustrate the progress and challenges of the method. Before the development of easily accessible GIS software, applications were either informal investigations, statistical and quantitative approaches close to processual archaeology, or humanistic studies of landscape phenomenology. First publications of GIS-based analysis also suffered from the lack of formalized inferential strategy, before developing both processual and cognitive-processual stances and a more finely defined analytical approach. Advancement was exemplified with David Wheatley’s attempt to rebut the hypothesis that Neolithic long barrows were located irrespective of the number of other barrows visible by utilizing the cumulative viewshed map and a common statistical procedure – the Kolmogorov–Smirnov test [6,15,16].

During the 1990s, visibility analysis was also a subject of criticism, ranging from amendable technical and procedural problems to the more complex questions of ontological, epistemological, methodological, and ethical aspects [14]. Most common technical issues were the uncertainty in viewshed results related to the binary visible/not-visible output being sensitive to errors in elevation data, as well as changing visibility of an object due to its size and distance from the observer. Efforts to address these issues included “fuzzy” and “probable” viewshed modelling, where visibility is expressed as a probability or degree (rather than a hard binary) to account for potential errors or environmental factors (e.g. haze) that could obscure distant targets [17–19]. Similarly, Ogburn [20] introduced a “target size-sensitive” approach: he computed the maximum distance at which an object of known dimensions (e.g. a 4.5 m wide Inca storehouse) could be recognized by the human eye, incorporating human visual acuity and atmospheric clarity into the viewshed calculation. The result is a probabilistic viewshed map indicating, for instance, that within a certain radius the object would be clearly visible (high confidence), whereas beyond that distance the probability of recognition drops off. These approaches produce more nuanced visibility maps but require more inputs (object size, error estimates, etc.) and computational effort.

Disapproval concerning the theoretical approach prompted two disparate understandings of “perception” regarding the natural environment, the first referring to environmental factors that affect object recognition, like contrast between the object and background, atmospheric conditions, and direction of illumination. The other is derived from human factors including physical (body – mechanical constraints on the field of view), but also cognitive in the sense of cultural preconceptions of environmental stimuli. This ambiguity in perception of environment has induced a more complex comprehension of visibility as a subjective result of human–environment interaction, rather than an objective trait of the environment [14], which is evident in some new studies [21].

Visibility is understood as a form of communication – one that could convey messages about control, cooperation, or sacred geography – but one that operated within a cultural system of meaning. One of the objectives of visibility analysis is to understand the connection between physical and mental perspectives of a society in the past, using quantitative visibility models to inform about how people may have conceptualized and given meaning to their environment [21]. Socially instructed visibility patterns were taken into account in some more recent articles. Analysis of large dolmens marking tombs of the Gor River valley (Granada, Spain) revealed no significant association between a monument’s size and its viewshed extent, rather hinting at a distributed signalling system, comprising a demarcated cultural landscape and a sign of community presence and territorial claim [22]. Similar social signalling is present in the case of Sardinian Bronze Age megalithic towers (nuraghi), which were places to achieve maximal visual control as symbols of authority [23].

In the last decade, developments in technology and methodology further enhanced viewshed analysis, increasing both its accuracy and the scope of questions it can address. One major advancement has been the improvement of input data quality. High-resolution DEMs derived from LiDAR surveys have become more widely available and are revolutionizing archaeological topographic analyses, including viewsheds. LiDAR can produce sub-metre resolution terrain models even under forest canopies, providing a much more detailed surface for visibility modelling than older 30 or 90 m grids. The use of LiDAR-derived DTMs has notably improved the fidelity of viewshed results – subtle terrain undulations, mounds, or saddles that might be missed in coarse data are now accounted for [24,25]. For example, a recent GIS analysis in the Orăştie Mountains (Romania) leveraged a LiDAR-based DTM to compute total viewsheds around Dacian and Roman sites. The fine-grained data allowed researchers to identify small “dead ground” areas that would have been hidden from view and to calculate more realistic visual coverage statistics (e.g. percentage of landscape seen within various distance bands) [26]. With LiDAR, it also becomes feasible to incorporate vegetation modeling: by using the point cloud to reconstruct the height of forest canopies or hedgerows, one can simulate past views both with and without woodland cover. While reliably reconstructing prehistoric vegetation remains challenging, LiDAR data coupled with palaeoenvironmental evidence enable sensitivity analyses – e.g. running a viewshed with an assumed tree height of 5 m across certain areas to see how that affects visibility [27].

Another area of advancement is computational efficiency and scale. The increasing ubiquity of multi-core processors and cloud computing has opened the door to extremely large or intensive visibility analyses. Researchers are now computing extensive total viewsheds that cover entire regions or countries, something previously impractical [28]. High-performance computing can handle thousands of viewpoints or very high-resolution grids by parallelizing the LOS calculations. For instance, dedicated algorithms have been developed for GPU (graphics processing unit) acceleration of viewshed calculations, and “big data” spatial frameworks can distribute the computation of cumulative viewsheds for hundreds of points across computing clusters. These technical strides mean that archaeologists can attempt ambitious analyses – such as iterative visibility simulations or real-time interactive visibility in virtual landscapes – that push beyond the static, single-viewshed maps of the past [29–31].

Integration with other analytical techniques is also advancing the field. One notable trend is combining viewshed analysis with network analysis and graph theory, as described in the chapter Visibility networks from the comprehensive The Oxford Handbook of Archaeological Network Research [32]. Instead of treating each site’s viewshed in isolation, scholars model sites as nodes in a visibility network, with edges between nodes that can see one another. Network metrics (e.g. node degree, betweenness centrality, etc.) can then identify which sites are visual “hubs” or intermediaries in communication webs [21]. Brughmans and Brandes [33] introduced formal network patterns for visibility, proposing first- and second-order visibility graphs to represent cumulative viewshed relationships. This approach moves beyond mapping to statistically comparing observed intervisibility networks against hypothetical models (for example, to test if a network was optimized for certain signal paths or if it was a byproduct of other factors) [33]. Such integration of viewshed data with network science allows more rigorous exploration of how visual connectivity might reflect social, military, or political networks in prehistory.

Furthermore, machine learning and advanced spatial statistics are beginning to incorporate viewshed-derived variables. In predictive modelling of site locations, for example, visibility indices can be used alongside environmental factors (distance to water, soil fertility, etc.) to train models that predict where undiscovered sites might be. Attempts in preventive archaeology have included viewshed as one layer in logistic regression and machine-learning classifiers to identify areas with high potential for past settlement [34].

To sum up, by the 2010s, GIS-based viewshed analysis was regularly employed in high-impact archaeological research, often in combination with other spatial analyses. Its historical development can thus be characterized as a progression from basic LOS mapping towards more complex, realistic, and hypothesis-driven visibility models. Key milestones include the introduction of cumulative viewsheds in the mid-90s, the adoption of fuzzy/probabilistic methods by the late 90s, and continual improvements in computation and data integration in the ensuing decades. This trajectory sets the stage for the diverse methods and applications in use today.

Recent advancements have made viewshed analysis more accurate (through better data and error modelling) and more powerful (through better algorithms and integration with complementary methods). The ability to factor in vegetation, atmospheric conditions, or human visual acuity means modern visibility studies can approximate past realities more closely than ever before. Concurrently, the fusion of visibility analysis with network models, predictive algorithms, and other cutting-edge techniques is expanding the range of archaeological questions that can be tackled – from understanding regional defence systems to modelling how landscapes were perceived in a cognitive sense. These developments ensure that viewshed analysis remains at the forefront of landscape archaeology research, continually refining our understanding of the prehistoric world.

2.3 Practiced method

For our visibility analysis, we used the freeware open-source Quantum GIS (QGIS)programme. The DEM map of the researched area was downloaded from the free service of NASA Earthdata SRMT (Shuttle Radar Topography Mission) online database with a resolution of 3 arc-seconds (approximately 90 m), through the SRTM Downloader plug-in in QGIS. Visibility analysis was executed with the Visibility analysis plugin, version 1.9.1, created by the archaeologist Zoran Čučković, a research associate at the University of Clermont-Auvergne (France). Observer’s viewpoints were created using precise coordinates collected by the authors on the field using a South Galaxy G6 GNSS Rover. The radius of analysis was set to 70+ km, even though achieving such high ceiling of LOS is arguable and only plausible under ideal conditions. Observer height was set to 10 m to account for observer’s body height as well as possible structure that likely could have been used for watching over the surrounding area. The aforementioned structure, similar to watchtower, could have aided in observing over common sized vegetation and other objects that would otherwise obstruct the view. Viewshed analysis was set to binary calculation, meaning that targets are either visible or not visible. Earth curvature was taken into account, and atmospheric refraction was set to a default value (0.130).

3.1 Archaeological excavations and remote sensing surveys

Archaeological excavations conducted in the last decade on some of the mentioned hillforts have uncovered similar prehistoric architecture and material culture that is not noticed before in the region of northwestern Serbia [5]. For example, at the Trojanov grad, during campaigns 2013 and 2014, a structure made of dry-wall stone technique (Figure 2) was discovered. This structure is located near a polygonal Roman tower, containing pottery from the Transitional period from the Late Bronze Age to Iron Age, i.e. Kalakača cultural group.

Figure 2

Photo of the prehistoric stone dry-wall at the Trojanov grad hillfort (photo: Institute of Archaeology, Belgrade).

Outside this prehistoric rampart was a shallow but wide artificial ditch, likely facilitating access to the western side. The LiDAR survey of the site, carried out in 2024, proved to be a transformative tool for documenting Trojanov Grad, which had long remained concealed beneath a dense forest canopy. Multiple drone flights produced centimetre-resolution point clouds and digital terrain models [35]. By varying flight parameters and integrating overlapping surveys, we achieved a remarkably clear topographic reconstruction of the site (Figure 3), with visible features such as the interior and exterior edges of the rampart, the prehistoric defensive ditch (Figure 3/1), prehistoric dry wall (Figure 3/2), and the polygonal base of the Roman tower (Figure 3/3).

Figure 3

LiDAR picture of the Trojanov Grad (graphic: Institute of Archaeology, Belgrade).

Similarly, on the highest peak of the Vidojevica ridge, a prehistoric hillfort with medieval fortification was identified, with finds from the Transitional period as well. Further, since 2014, excavations at the Cikote hillfort have uncovered remains of a stone rampart (Figure 4) constructed using dry-wall techniques [5], with wooden posts forming the wall’s outer section. The pottery found here also belonged to the Kalakača cultural group. The wooden post sample dated back to 973–840 BCE (sigma 1 range).

Figure 4

Remains of the stone rampart at the Cikote hillfort constructed using dry-wall techniques. Red colour is the result of burned wooden cassettes that was filled with crushed stones (photo: Institute of Archaeology, Belgrade).

A new LiDAR survey (Figure 5) further revealed a massive stronghold (250 m × 150 m) at the hilltop, with a surrounding wall (Figure 5/1), a wide defensive platform at the entrance (Figure 5/2), and a shallow artificial ditch in front of it (Figure 5/3). This stronghold, with the material culture only from the Transitional period covered the impressive surface about 4 ha. Our excavations did not confirm other archaeological features on Cikote except the stone wall, and a geomagnetic survey confirmed our excavation results that the whole southern wall was burned in one moment [36].

Figure 5

LiDAR picture of the Cikote hillfort (graphic: Rio Sava DOO, Belgrade).

3.2 Kalakača culture – relative and absolute chronology

Kalakača culture, as a distinct cultural phenomenon of the Early Iron Age in southern Pannonia, was identified several decades ago by Medović [37] as a pre-Basarabi layer, i.e. the Bosut IIIa horizon. Although some early authors questioned the existence of this phase or horizon of the Early Iron Age, today its classification appears much clearer and scientifically well-founded. It is now more appropriately recognized as a cultural group, which occupied a significantly larger territory than initially assumed. In addition to Srem and southern Bačka, the bearers of this group also inhabited the peri-Pannonian border zone, and stylistic typological features characteristic of Kalakača pottery have been identified in central Pomoravlje [38], as well as in southern Pomoravlje and the Vardar valley [39].

According to Medović, the Kalakača horizon in southern Pannonia was dated to the Ha B period of Central European chronology, i.e. between 900 and 750 BCE, or even earlier [39]. Stojić places it in the Iron Age Ic in the Pomoravlje region, which chronologically corresponds to the aforementioned dating [38]. Later, thanks to radiocarbon dating, this group was dated in Srem to the tenth/ninth–eighth centuries BCE (Table 1) [40,41]. As for relative chronology, most scholars working on this topic agree that the Kalakača horizon was preceded by the channelled pottery horizon (Urnfield culture or the later phase of the Belegiš group, or the Vojvodina group of the transitional period), and that it was followed by the Basarabi group [37].

Characteristic pottery forms of the Kalakača group include faceted or channelled-rim bowls, hemispherical bowls with thinned rims, high-handled cups with button-like protrusions, pear-shaped two handled beakers, and amphorae with everted rims and conical necks. Common ornaments feature diagonal channels, incised lines, stamped dimples, zigzags, and garland motifs. Typical decorative patterns include fir-branch designs, hatched triangles, wavy lines, and linked S-motifs.

As we mentioned, in recent years, Kalakača group pottery has been documented not only in the Pomoravlje and Danube regions but also at numerous sites along the peri-Pannonian rim, particularly in Mačva, Jadar, and Podgorina [5,45]. The main characteristics of this pottery do not differ from those found in Srem. A wide array of forms are represented, including bowls with inset channelled or faceted rims (Paulje, Cikote, Brangović, Trojanov Grad, and Vito); hemispherical bowls with unprofiled, thinned rims (Cikote, Brangović, and Vito); pear-shaped beakers with two handles (Brangović); cups with handles extending above the rim, often featuring one or two button-like protrusions at the top (Paulje, Cikote, Trojanov Grad, and Vito); and amphorae with long conical necks (Paulje, Cikote, Brangović, Trojanov Grad, and Vito). Common decorative elements include channels in the beaker belly (Paulje, Brangović, Trojanov Grad, and Vito); incised zigzag lines (Paulje, Cikote, and Brangović); hatched triangles (Cikote, Brangović, and Trojanov Grad); multiple incised lines (Paulje, Cikote, Brangović, Trojanov Grad, and Vito); and fir-branch motifs (Brangović), among others (Figure 6).

Figure 6

Plate with pottery from various archaeological settlements analysed in this article (graphic: Institute of Archaeology, Belgrade).

Given these identical stylistic and typological features, it appears that the material culture of the Kalakača group in Srem and the peri-Pannonian rim did not differ. This conclusion is supported by radiocarbon dates for the Kalakača group, with the earliest being contemporaneous and originating from Gomolava (Srem) and Cikote (Jadar) (Figure 6). Based on the available absolute dates, it seems that the group not only emerged simultaneously but also developed in parallel across both regions.

Regarding the relationship between the absolute and relative chronology of the Kalakača group, an important finding comes from Ratača near Bogatić. In one pit, fragments of pottery were discovered that correspond to the Gava–Belegiš II group, which is believed to have preceded the Kalakača group. This pottery, decorated exclusively with wide channels and lacking any features associated with the Kalakača group, was dated to the eleventh and the first half of the tenth century BCE (Figure 6). Thus, the absolute chronology confirmed not only the previously accepted relative chronology, most clearly illustrated by the stratigraphy at Gomolava, but also the other sites, where the channelled pottery of the Gava–Belegiš II type directly preceded the Kalakača group. A similar situation was observed in a pit at the site of Paulje in Brezjak near Loznica (Figure 6).

3.3 Topography and visibility of the settlements and hillforts

When considering all sites associated with the material culture of the Kalakača group in northwestern Serbia, we can identify 11 hillforts Vidojevica, Trojanov Grad, Koviljača, Orlovine, Ostenjak, Cikote, Plužac, Brangović, Vito, Taor, and Krčmar and two lowland settlements: Paulje, and Radalj. In addition, our analysis included two large contemporary proto-urban centres located along the northern boundary of the region: Gradina on Bosut and Gomolava. These settlements are situated in the Srem region of southern Pannonia, the cultural heartland of the Kalakača group. Notably, no graves have been discovered within this cultural horizon in the northwestern Serbia [5]. In the core territory, in the Kalakača group, burials have not yet been sufficiently researched, with scarce grave finds. Based on the existing data, not only inhumation [46–48] but also cremation as a treatment of the deceased was recorded [49]. This distribution presents a striking contrast to earlier periods while numerous fortified hilltop settlements are present, there are only three known open settlements and no burial sites, which are typically interpreted as a reflection of unstable conditions and a hostile environment, as well as possible partial transformation in population dynamics.

Significantly, the settlements of Gradina on Bosut and Gomolava fall within the visibility range of two major hillforts in the Cer area Trojanov Grad and Vidojevica as well as the Gradac–Koviljača hillfort, which strategically controlled a key road running upstream along the Drina River, as well as an important river crossing. Also, on the cumulative viewshed map (Figure 7), it seems that visibility range with Trojanov Grad also exists, which gives additional importance to Koviljača stronghold. One of the central questions in this type of analysis concerns the maximum distance of visibility under clear atmospheric conditions, particularly in the mountainous terrain. Based on available data and visibility modelling, optimal conditions likely enabled visual communication over distances of up to 50–100 km. In our specific case, the distance from Trojanov Grad to Gomolava and Bosut was 38 and 61 km, respectively, while the distance from Vidojevica to Gomolava and to Bosut measured 43 and 51 km. While there is no direct archaeological evidence for a communication system between these sites, it is highly plausible that some form of signalling was employed. It is also possible that visual communication was more effective at night due to fire or light-based signalling, although the content, complexity, and reliability of such messages remain unknown. In one field photo, under moderate haze and elevated humidity, the Cikote hillfort was clearly visible from Trojanov Grad a distance of 14 km (Figure 8). Behind it, other hills and low mountain ridges up to 30 km away were distinguishable, demonstrating the potential extent of intervisibility. Undoubtedly, the Cikote stronghold, together with Koviljača, Vidojevica, and Trojanov Grad, represents one of the most strategically significant positions, primarily because of its commanding oversight of the Jadar River valley and visual connectivity with several fortifications, otherwise positioned less optimally in relation to key regional sites.

Figure 7

(a) Cumulative viewshed map of the analysed settlements and strongholds (graphic: Institute of Archaeology, Belgrade). (b) Cumulative viewshed map of the analysed settlements and strongholds (graphic: Institute of Archaeology, Belgrade).

Figure 8

Photo of the Cikote hillfort from the Trojanov Grad (photo: Institute of Archaeology, Belgrade).

4.1 Strategic placement and topographic control

The location of hillforts in elevated and naturally defensible positions, as revealed by DEM and terrain analyses, suggests deliberate choices driven by strategic considerations. Most fortified sites dominate their immediate surroundings, offering commanding views of river valleys, access routes, and arable land. This pattern aligns with a broader trend across prehistoric Europe where hilltop positions were exploited not only for defence but also for territorial control and surveillance. Vidojevica hillfort was positioned at the farthermost western end of the Cer Mountain massif and had good visual communication with whole Mačva, Semberija, Bosut, and Gomolava settlements in Srem (Figure 7), but also Koviljača at the start of the Drina canyon valley that has clear contact with Vidojevica (Figure 9). The fortified site at Cikote, for instance, holds a key position over the Jadar River valley, while others such as Koviljača and Trojanov Grad appear to flank major corridors of movement. The clustering and distribution of these sites imply a shared logic of occupation that responds to both natural defensibility and visibility.

Figure 9

Photo of the Vidojevica hillfort from the Koviljača hillfort (photo: Institute of Archaeology, Belgrade).

Vidojevica, however, easily controlled the alluvial valley of the lower Drina and was positioned ideally for the defensive and scouting purposes. Furthermore, according to our photos, Vidojevica definitely could visually reach the Cikote hillfort (Figure 10). Even the strategic position of the Likodra hillfort, situated within the seemingly secluded Likodra River valley, should not be underestimated. First and foremost, the valley itself is narrow, flanked by steep, nearly impassable hills, and the surrounding area is rich in ore deposits and ancient mining activity. Additionally, the hillfort offers a direct LOS to Trojanov Grad (Figure 11), which likely explains why communities of the Transitional period chose to establish their stronghold on this distinctive rocky outcrop. Moreover, the visual connection between Likodra and the Plužac hillfort should be taken into account, as it appears that only a limited number of hillforts were able to maintain visual communication along the west–east axis (Figure 7). Unfortunately, although Orlovine – a medieval fortress that also yielded pottery from the Transitional period was situated in a highly advantageous strategic location, effectively controlling communication routes upstream from the Koviljača hillfort, its visual position was relatively poor (Figure 7), primarily due to the high hills that surround the site and obstruct long-range visibility.

Figure 10

Photo of the Trojanov Grad (1) and Vidojevica (2) from the Cikote hillfort (photo: Institute of Archaeology, Belgrade).

Figure 11

Photo of the Trojanov Grad from the Likodra hillfort (photo: Institute of Archaeology, Belgrade).

Trojanov Grad appears to have been the most important stronghold, as it controlled an extensive area (Figure 7), with visual coverage extending from the Bosut settlement in Srem to the Vito hillfort on Mount Povlen – a total of 11 sites. In addition to its broad viewshed, Trojanov Grad was also the closest hillfort to the Milinska and Cernica rivers, suggesting that the guards stationed there could monitor the tin-bearing river deposits as well as, potentially, copper sources in the nearby Srebrne Rupe mining area. Although there is no direct LOS to the Vidojevica hillfort, the construction of simple wooden watchtowers at several key locations on Cer could have easily bridged this gap in visual communication. The four eastern hillforts Brangović [50], Vito [45], and Krčmar each appear to have controlled their own local territories while also maintaining visual communication with at least one western hillfort. In contrast, the Taor hillfort was located on the opposite side of the mountain and had no visual connection with any of the sites included in our analysis. Notably, Trojanov Grad maintained visual contact with all three of the aforementioned eastern hillforts.

4.2 Visual networks and intervisibility

The viewshed analysis demonstrates that several hillforts were intervisible, forming a loosely connected network that spans large portions of the region. These visual links may have supported systems of communication, coordination, or even symbolic alignment. For example, intervisibility between sites like Cikote and the surrounding minor fortifications could imply signalling routes or zones of mutual awareness. This capacity for long-distance visibility may have been as much about asserting presence and territoriality as about practical signalling. In this sense, visibility could have served to display social power and community identity within a shared cultural framework.

4.3 Cultural attribution and symbolic landscape

The hillforts discussed here likely represent not only physical strongholds but also culturally charged locations. Their placement in dominant visual positions may have contributed to their symbolic significance. Drawing on perspectives from sensory archaeology and landscape phenomenology, these sites may have structured human experience, reinforcing notions of community, identity, and control. The absence of graves, combined with the strong emphasis on settlement hierarchy and visibility, points to a landscape structured more by spatial presence than by mortuary expression.

4.4 Towards a more interpretative framework

While the technical results provide a strong foundation, future work should more deeply explore how these landscapes were perceived, experienced, and remembered. Visibility was likely not just functional but embedded in a web of meanings: political, social, and possibly ritual. Ethnographic analogies and cross-regional comparisons could contextualize the spatial strategies observed here. In conclusion, this discussion has aimed to move beyond a descriptive summary of spatial data and towards an interpretive synthesis that recognizes the complexity of prehistoric landscape organization. The hillforts of northwestern Serbia emerge not only as defensive structures but also as significant elements in the construction of social landscapes shaped by both strategic logic and symbolic meaning.