Using Landscape Learning to Explore Diachronic Change: A Quantitative Model and Western Stemmed Tradition Case Study

Abstract

University of Washington Abstract Using Landscape Learning to Explore Diachronic Change: A Quantitative Model and Western Stemmed Tradition Case Study David B. HuntChair of the Supervisory Committee Professor Emeritus Donald K. Grayson Anthropology In this research, I propose new methodologies for measuring landscape learning and gauging residence time on a landscape. I use a landscape learning model to set expectations and propose testable hypotheses utilizing these methods. The model and methodologies are then tested against data in the context of a Paleoindian colonizing event within the Old River Bed (ORB) delta in Utah.I develop what I refer to as the Discoverability model to predict the order in which a random walker will discover patchy resources found on a neutral landscape, dependent only on distance and patch size. The simulation results for the model support my hypothesis that patch size affects encounter rate and that the model could be used to create a deterministic baseline for patch discovery against which to measure the accumulation of landscape knowledge. I also present an original methodology to quantitatively determine toolstone patch sizes, or exposure extents, using hydrographic algorithms along with known primary source locations. These methods are tested on toolstone sources used by Paleoindians residing in the ORB delta. The results demonstrate that, on average, the methodologies predicted 66% of the actual downslope flow of obsidian sediments and successfully returned an average scaled prediction of 89% of the area of the actual surveyed flow extents. To test the Discoverability model, the ORB Paleoindian assemblages are divided into temporal groups. For each assemblage, the Discoverability values were calculated using the exposure and distance values for each toolstone source, and Discoverability lists (Dlists) of expected rank-order usage of toolstone sources are created. The corresponding Observed lists (Olists) were created using the observed toolstone proportions in each assemblage. The Dlists and Olists were then compared using Spearman’s rank order correlation. From these results, the landscape learning variable (%LL) was calculated for each temporal group/assemblage. The oldest temporal group’s Olist returned a very strong correlation (rs = 0.777) with its expected Dlist. This, in turn, returned the lowest level of landscape learning of any of the temporal groups (%LL=39.7%), as my model predicts. Importantly, the magnitude of difference in %LL (δ=35.1%) between the oldest and next oldest assemblage (~1096 cal years later) is significantly greater than any differences between any other subsequent temporal steps between the assemblages. These results indicate a significant step in landscape learning occurred between the earliest assemblage and the next temporally discrete assemblages. Overall, the results suggest that up to 48% of the variance in landscape learning over time at the ORB delta is explained by my Discoverability model. With limitations of scale and archaeological resolution, the model and resultant methods show promise as a means to quantify and rank the level of landscape learning within an assemblage.