Settlement Clustering and Site Nucleation
The deliberate occupation of the Plateau well before the arrival of the Inka is crucial for understanding the regional dynamics during the Inka era. This occupation was characterized by settlement clustering, a tendency that continued after the state arrival. In order to evaluate this trend, a k- means clustering analysis was performed. Changes in the average distance between sites were also examined as a way to understand processes of site nucleation and settlement interaction. The purpose was to see if the site congregation revealed the presence of political units, and if the changes in the distance between settlements could indicate social or economic factors at play.
The k-means analysis is a nonhierarchical, divisive clustering method that groups sites based on spatial proximity. For each group of sites, a hypothetical centroid is calculated (Kintigh and Ammerman 1982).1 Of course, there are limitations in this analysis when political or natural
Figure 4.15. Distance of 200 m between settlements in the Early Yampara period.
boundaries of a region are not considered. Therefore, the Valley and the Plateau were separated since they constitute separate ecological and topographic units. Circles around the sites of each period were also drawn in GIS until they aggregated next to each other, in order to assess the average distance between settlements. As an independent test, this strategy also helped to visualize the clusters already identified by the k-means analysis. For the Early Yampara period (A.D. 400-800), we only documented a few associated sites. As a result, the sites did not congregate into visible clustering patterns. Most sites were on the Valley floor, and the space between them was wide (0.5-2.5 km apart) (Figure 4.15).