Supplementary Information for: Points, patterns, and predictions in archaeological settlement data: Site-environment relationships of Paracas and Nasca communities in the Peruvian Andes

This repository contains the necessary instructions to carry out the analyses described in the paper titled: Points, patterns, and predictions in archaeological settlement data: Exploring site-environment relationships of Paracas and Nasca communities in the Peruvian Andes through point pattern analysis and predictive modeling Authors: Giacomo Bilotti(a,b), Markus Reindel(c), Johny Isla(d), Christian Mader(e) a Institute of Pre- and Protohistoric Archaeology, Kiel University, Kiel, Germany b Social Resilience Lab, Center for Humanities Computing, Aarhus University, Aarhus, Denmark c Commission for Archaeology of Non-European Cultures, German Archaeological Institute, Bonn, Germany d Nasca-Palpa Management Plan, Peruvian Ministry of Culture, Nasca, Peru e ArchDepth Research Group, Bonn Center for Dependency and Slavery Studies, University of Bonn, Bonn, Germany Abstract This paper provides framework for studying settlement patterns during pre-Inka times in the western Andes, using Point Pattern Analysis (PPA). PPA is a very flexible statistical method widely used to investigate spatial relationships between archaeological sites, offering insights into settlement dynamics and human-environment interactions. Nevertheless, the technique has found limited application in the Andean region, partly due to data availability and historical research trajectories. This study aims to fill this gap and employs PPA to analyze settlement patterns and site-environment relationships in the Palpa valleys of the western Andes during the Paracas (800–160 BCE) and Nasca (160 BCE–620 CE) periods. Our analysis examines the spatial structure of prehispanic settlements in relation to landscape features, identifying factors influencing site location choices and their evolution over time. Furthermore, we use the results from PPA to predict site intensity in nearby regions that were only marginally investigated archaeologically in order to identify the most relevant areas for future research activity. The performance of the model proposed here is tested at different levels in order to improve our knowledge and increase the fit of the final model. The results show changes over time in the occupation of the landscape, most of which were directed towards optimizing agricultural production. However, we have also detected a strong impact of mobility during certain periods and inter-site interaction. Structure of the repository The repository folder is structured as follows: README.md: This file (repository overview). analyses.qmd: Quarto file performing all the analyses needed for the paper. scripts/: Contains all R scripts for data preparation, analysis, and visualisation. covariates/: Contains the R scripts to produce most of the covariates required for the analyses. 00_area_creation.R: Defines the area of the study. 01_DEM.R: (down)loads the DEM and processes it. 01_rivers.R: Calculates distance from rivers. 02_FETE.R: Computes the FETE LCPs following Bilotti et al. 2024 (it can be iterated to avoid crashes). 03_sum_LCPs.R: Sums the outputs of the previous scripts to get a single raster (only necessary if the previous step was iterated). 04_covariates.R: Computes slope, TPIs and distance from LCPs. 05_plot_covs.R: Creates and saves the plots of the covariates. test_correlation_covariates.R: Tests the autocorrelation of the different covariates. sites_plots.R: Creates and saves the plots for sites divided by phase. SI_sec_5.R: Qualitatively compares Silverman 2002 results with the ones presented in the paper. SI_sec_6.R: Generates the figure included in section 6 of the SI. plot_predictive_models.R: Generates Fig. 6 of the paper. Kres_intervals.R: Automatically detects the interval of the Kres function that lie outside the CI. This script is called directly from the QMD when needed. data/: Raw and derived data necessary for the paper. raw_data: Where raw input files should be stored. derived_data: Contains processed or derived datasets. csv/: Contains the predictive model results for each site and period. Supplementary Information.pdf: Contains the SI in PDF format. In GitLab only the scripts are available.本仓库包含执行下述论文中所述分析所需的全部操作指南： 《考古聚落数据中的点位、模式与预测：通过点位模式分析（Point Pattern Analysis, PPA）与预测建模探究秘鲁安第斯山脉帕拉卡斯（Paracas）与纳斯卡（Nasca）社群的聚落-环境关联》 作者：Giacomo Bilotti(a,b)、Markus Reindel(c)、Johny Isla(d)、Christian Mader(e) a 德国基尔大学史前与原史考古研究所，基尔，德国 b 丹麦奥胡斯大学人文计算中心社会韧性实验室，奥胡斯，丹麦 c 德国考古研究所非欧洲文化考古委员会，波恩，德国 d 秘鲁文化部纳斯卡-帕尔帕管理计划，纳斯卡，秘鲁 e 波恩大学波恩依赖与奴隶制研究中心ArchDepth研究组，波恩，德国 摘要 本文以点位模式分析（Point Pattern Analysis, PPA）为工具，构建了一套研究安第斯山脉西部印加前时期聚落模式的分析框架。点位模式分析是一种极具灵活性的统计学方法，被广泛用于探究考古遗址间的空间关联，可为聚落动态与人地互动提供洞见。然而，该技术在安第斯地区的应用仍较为有限，部分原因在于数据可得性与历史研究轨迹的限制。本研究旨在填补这一空白，采用点位模式分析方法，对安第斯山脉西部帕尔帕河谷在帕拉卡斯时期（公元前800年—公元前160年）与纳斯卡时期（公元前160年—公元620年）的聚落模式及聚落-环境关联展开分析。本研究考察了前西班牙殖民时期聚落的空间结构与景观特征之间的关联，识别出影响遗址选址决策及其随时间演变的关键因素。此外，我们利用点位模式分析的结果，对考古学研究程度较低的邻近区域进行遗址强度预测，以确定未来研究的重点区域。为提升模型性能并优化最终拟合效果，本研究在多个层级对所提出模型的性能进行了测试。研究结果表明，景观占用随时间呈现出显著变化，其中多数变化以优化农业生产为导向；同时，我们也发现在特定时期内流动性与遗址间互动的强烈影响。 仓库结构 本仓库文件夹结构如下： README.md：本文件（仓库概览）。 analyses.qmd：用于执行论文所需全部分析的Quarto文件。 scripts/：包含用于数据准备、分析与可视化的全部R脚本。 covariates/：包含用于生成本分析所需绝大多数协变量的R脚本。 00_area_creation.R：定义研究区域范围。 01_DEM.R：（下载并）处理数字高程模型（Digital Elevation Model, DEM）。 01_rivers.R：计算至河流的距离。 02_FETE.R：参照Bilotti等人2024年的研究计算FETE最小累积成本路径（Least Cost Paths, LCPs）（可通过迭代运行以避免崩溃）。 03_sum_LCPs.R：将前述脚本的输出结果进行求和，以生成单一栅格文件（仅当前述步骤被迭代运行时需要）。 04_covariates.R：计算坡度、地形位置指数（Topographic Position Index, TPI）以及至累积成本路径的距离。 05_plot_covs.R：生成并保存协变量的可视化图表。 test_correlation_covariates.R：测试不同协变量的自相关性。 sites_plots.R：生成并保存按时期划分的遗址可视化图表。 SI_sec_5.R：将Silverman 2002年的研究结果与本文所述结果进行定性对比。 SI_sec_6.R：生成补充材料第6节中的图表。 plot_predictive_models.R：生成论文中的图6。 Kres_intervals.R：自动检测位于置信区间（Confidence Interval, CI）之外的Kres函数区间，该脚本将在需要时由Qmd文件直接调用。 data/：论文所需的原始数据与衍生数据。 raw_data：原始输入文件的存储目录。 derived_data：包含经过处理或衍生得到的数据集。 csv/：包含各遗址与时期的预测模型结果。 Supplementary Information.pdf：PDF格式的补充材料。 注：GitLab平台上仅开放脚本文件。