The distribution of pā sites in the central New Zealand region of Tōtaranui/Queen Charlotte Sound is investigated to determine the relationship between pā and other Māori archaeological sites, and the influence of maritime and introduced terrestrial resources. Particular aims of research are to investigate the role of visibility as a measure of defensibility in the distribution of Tōtaranui pā, and whether this distribution is influenced by the distribution of garden sites and karaka stands, two important introduced resources often considered to influence pā distribution at national levels. Additionally, evidence for gardening activity is ambiguous in Tōtaranui, for which there is a number of archaeologically recorded garden sites but a noted absence of gardening activity in the earliest historical records.
Investigation into the distribution of pā is done by comparing proximity and intervisibility of pā sites to a range of Māori archaeological site types (garden sites, karaka stands, midden sites, oven sites, other pā sites, pit sites, and terrace sites). Proximity of pā to other archaeological sites is quantitatively measured by way of cost distance analysis, and intervisibility of sites measured by way of viewshed analysis within Geographical Information Systems (GIS). Cost distances and frequency of intervisibility of pā to other archaeological sites is recorded and compared against a dataset of random points.
Based on the methodology used in the analysis, neither maritime nor introduced terrestrial resources had a significant influence on the distribution of Tōtaranui pā. A desire for high levels of visibility as a measure of defence was not found among these pā, as they did not have significantly larger viewsheds than random points. Pā were however, found to occupy spatially and visually central positions in Tōtaranui settlement systems, as cost distances from pā to other archaeological site types were significantly lower than cost distances from random points, and the frequency of intervisibility with other archaeological sites was higher among pā than among random points. Cost distance and viewshed analysis are shown to produce meaningful results in New Zealand archaeological contexts, and concerns are raised regarding the application of models based on distribution at national scales to individual regions, particularly those in areas considered marginal.
