Testing Circumscription in Iron Age Galicia, Spain: Computational Approaches to Conflict and Subsistence in the Castro Landscapes

Technical Summary: Testing Circumscription in Iron Age Galicia, Spain

Problem Statement
This research addresses a critical gap in archaeological theory: the widespread acceptance of Carneiro's circumscription theory as an explanation for the emergence of social hierarchy, contrasted with a lack of formal, quantitative testing of this hypothesis in specific regional contexts. Circumscription theory posits that environmental or social constraints on land, combined with population growth, inevitably lead to resource scarcity, warfare, and the forced incorporation of defeated groups into hierarchical political units.

In the context of Iron Age NW Iberia (Galicia), this theory has been frequently invoked to explain the transition from egalitarian segmentary societies to hierarchical structures during the Late Iron Age (post-4th century BC). Proponents of hierarchical models argue that demographic pressure and environmental circumscription (e.g., limited arable land on peninsulas) created an "agrarian trap," forcing the emergence of warrior elites. Conversely, the segmentary model argues that these societies actively controlled settlement growth and surplus to maintain egalitarianism. The paper seeks to resolve this debate by formally testing whether the balance between population and resources in Iron Age castros (hillforts) actually supported the conditions necessary for circumscription-driven hierarchisation.

Methodology
The authors present a computational framework designed to model the population-resource balance under a "worst-case scenario" to rigorously test the null hypothesis: For any given castro, the catchment area required to ensure the survival of its inhabitants does not overlap with those of any other castro. If the null hypothesis holds (i.e., most sites have sufficient non-overlapping resources), circumscription is rejected as a driver of hierarchy.

The methodology involves four integrated steps:

1. Demographic Estimation:

   • Data Source: Settlement sizes were derived from remote sensing (LiDAR, historical 1956 aerial photography, satellite imagery, and slope maps) to calculate habitable area within enclosure walls.
   • Proxy: Population was estimated using a density range of 150–200 inhabitants per hectare, a standard accepted across various interpretative models for Iron Age castros.
   • Strategy: To maximize stress on the model, the analysis treats all known castros as contemporaneous, creating an artificial state of maximum demographic pressure and over-occupation.

2. Agrarian Resource Modelling:

   • Four distinct agricultural production models were developed, ranging from least to most intensive, to test sensitivity:
     • Large Cycle (Slash-and-burn): 3.2 ha/person (1 harvest, long fallow).
     • Extensive Short Cycle: 1.4 ha/person (annual/biannual rest).
     • Intensive Short Cycle: 1.1 ha/person (higher yields, complex rotation).
     • Ternary Rotation: 0.56 ha/person (1/3 winter cereal, 1/3 summer cereal, 1/3 fallow; 2 harvests/year).
   • Parameters: The models assume a cereal-based diet (70% of intake) requiring 188 kg of cereals per person/year. Productivity values were deliberately set lower than experimental archaeology data from the British Iron Age to ensure a conservative, stress-test approach.

3. Agricultural Potential Mapping:

   • A new computational tool (QGIS plug-in) was utilized to generate maps of agricultural potential based on five proxy layers: elevation, slope, soil erosion, wetness, and rocky soils.
   • Land was classified into three categories: Intensive potential (heavy, deep soils), Extensive potential (light, shallow soils), and Null potential (unsuitable for cultivation).
   • Weighting: In the analysis, intensive land was weighted at 1.25, extensive at 0.75, and null at 0.

4. Site Catchment Analysis (SCA):

   • SCA1 (Fixed Time): Calculated catchment areas based on fixed travel times (60, 90, and 120 minutes) using the r.walk algorithm in GRASS GIS. Territories were truncated using Thiessen polygons to ensure mutual exclusivity.
   • SCA2 (Dynamic Expansion): A more sophisticated approach where catchment areas expanded dynamically until they met the specific resource requirements of each site's population. This method avoids fixed time limits and Thiessen polygons, instead creating exclusive territories separated by cost-based borders. This allows larger settlements to naturally acquire larger territories if necessary, while accounting for the spatial distribution of land quality.

Key Results
The analysis was conducted across the entire region of Galicia, testing the balance under the most restrictive demographic and agrarian parameters.

• SCA1 Results (Fixed Time):

  • Under the Ternary Rotation model (most intensive), 97.4% to 98.3% of castros maintained a positive population-resource balance across all travel times (60–120 mins).
  • Even under the Large Cycle model (least intensive, worst-case), 64.3% to 74.4% of castros remained self-sufficient.
  • Surplus land was significant; for the Ternary model, sites had on average 10.2 to 14.9 times more land than required.

• SCA2 Results (Dynamic Expansion):

  • This method showed even higher rates of self-sufficiency.
  • 99.5% of castros achieved a positive balance under the Ternary Rotation model.
  • 99.0% under the Intensive Short Cycle.
  • 98.5% under the Extensive Short Cycle.
  • 91.0% under the Large Cycle model.
  • Only 12 castros failed to achieve balance in the Ternary model, primarily due to their location on small islands where the model restricted movement to terrestrial routes only.

Significance and Claims
The paper claims that these results allow for the rejection of the null hypothesis regarding resource scarcity. The findings demonstrate that:

1. Rejection of Circumscription: Demographic pressure and environmental constraints were not structural drivers of social hierarchisation in Iron Age Galicia. The vast majority of settlements possessed sufficient local resources to sustain their populations without entering into conflict over land, even under extreme, worst-case assumptions.
2. Validation of the Segmentary Model: The results align with the interpretation of Iron Age Galicia as an agrarian, segmented landscape. The ability of communities to maintain self-sufficiency supports the argument that these societies actively controlled settlement size and surplus to prevent the emergence of hierarchy.
3. Nature of Conflict: The dispersed settlement patterns and defensive structures are reinterpreted not as evidence of open warfare driven by resource scarcity, but as a social construction of "complementary opposition" and isolationism characteristic of egalitarian segmentary societies.
4. Role of Rome: The paper posits that the dismantling of this egalitarian ethos and the emergence of hierarchical structures (evidenced by nucleated settlements >3 hectares) occurred only in the last two centuries of the first millennium BC, coinciding with Roman influence and the disruption of traditional segmentary organization.

Conclusion
The study provides a reproducible computational framework for testing circumscription theory in other late prehistoric contexts. By demonstrating that Iron Age Galicia was not environmentally or demographically circumscribed, the paper challenges the assumption that hierarchy is an inevitable outcome of agricultural intensification and population growth, suggesting instead that social organization is a historically contingent choice.