The data and code provided in this repository is associated with the analysis performed in "Contributions of Bayesian Phylogenetics to Exploring Patterns of Macroevolution in Archaeological Data" written by Erik Gjesfjeld and Peter Jordan. This chapter was submitted for review in February 2018 to the "Handbook of Evolutionary Research in Archaeology" edited by Anna Prentiss and published by Springer Press. Please do not cite any of the data or code contained in the repository without first consulting the primary author (Erik Gjesfjeld).
The primary data sources used in the analysis include the following files: plains_houses_rev.csv - This data was compiled by Erik Gjesfjeld (Gjesfjeld 2003) and contains characters and characters states for 40 house features from archaeological sites throughout the Great Plains of North America. A key for the characters and character states can found in the chapter text.
plains_house_dates.csv - This data is a series of dates used to time-calibrate the trees. These dates were compiled using a variety of sources that can be found in the book chapter. If radiocarbon dates associated with each site (or even each house feature) could be identified in the literature these dates were used. Otherwise, the age range for the cultural period typically assigned to the site or house feature was used. This is admittedly an imprecise way to provide tip dates but was one of the only options available.
plains_houses_lifespans.txt - This data also derives from the series of dates described above. The format of the file is that the clade is equal to the cultural tradition of each house feature (1-Central Plains, 2-Middle Missouri, 3-Coalesecent). The species number refers to the site of each house feature (see plains_houses_rev.csv), ts is the earliest age associated with the site/house, te is the latest age associaed with site/house. Age ranges are given in time before present.
BEAST2 Analysis
BEAST2 was the preferred software for constructing the phylogenetic trees used in this research. The analysis can be replicated in BEAST2 by using the XML control file provided here and loading it into BEAUTi. The control file provided is for the time calibrated tree with tip dates provided in Figure 3. The naming of the XML file provides the different choices made by the analyst (Erik Gjesfjeld) and described in the chapter text.
plains_houses - the raw data used in the analysis Mk - Markov model 4 - Number of gamma rate categories strict - Clock model used Sky - Prior used (Bayesian Skyline Prior) subs - substitution rate was also logged by the MCMC 50 - Number of trees used in the posterior distribution (50,000) Calib - Nodes for cultural traditions were calibrated on age BP_TipDates - Tip Dates were provided in BP ages
Tree Files Files with the .trees or .tre suffix are trees that are resulted from the BEAST2 analysis. Files that have the MCCT.trees suffix are a single maximum clade credibility tree that were produced using the TreeAnnotator program associated with BEAST2. Tree files like these can be viewed directly in programs such as FigTree
PyRate Analysis
The analysis of origination and extinction rates was performed using the PyRate program developed by Daniele Silvestro and colleagues (Silvestro et al. 2014). A draft tutorial on using PyRate with cultural evolutionary data can be found at github.com/erikgjes/PyRate_Cultural_Evolution and the PyRate program can be downloaded at https://github.com/dsilvestro/PyRate.
Raw data for this analysis is found in the plains_houses_lifespans.txt file and marginal rate estimates from the analysis are found in the plains_houses_0_BD_marginal_rates_RTT.r file which can be loaded into R and plotted using the code below.
R code for Figures
Code for figures 9.1, 9.2, 9.3 and 9.4 are provided as separate R files. These largely contain code for graphing and plotting using the results contained in the above files.
Any questions, comments or constructive criticism about the analysis performed here should be directed to the primary author (Erik Gjesfjeld) at erik.gjesfjeld@gmail.com.