Data, MATLAB and R-scripts for the analysis of archaeological C14 and paleoclimate proxy data

Code and data availability on this repository: https://github.com/kaiwirtz3/holocene, also archived on Zenodo as https://zenodo.org/doi/10.5281/zenodo.10467363.

The MATLAB and R code enables standalone or parallel processing. On High Performance Computers (HPC) with the SLURM scheduler, parallel processing of batch jobs can be invoked on HPC by sbatch --array=1-64 slurm.sh.

Requirements

MATLAB

• needs packages: m_map and pca_toolbox
• you can get the pca_toolbox from https://www.mathworks.com/matlabcentral/fileexchange/134751-pca-toolbox-for-matlab
• you can obtain m_map from https://www.eoas.ubc.ca/~rich/map.html
• insert/correct installation path in addpath('~/tools/m_map') found in the header of some MATLAB-scripts (foremost load_par.m), or addpath('~/tools/pca_toolbox') in proxy_dtw.m

R

• needs libraries: rworldmap, rcarbon, R.matlab, RColorBrewer, cluster, sp/sf for rcarbon version <1.5 see comments in grid_growth.r ("spweights")
• we provide a conda environment that takes care of most of these dependencies, enumerated in the file environment.yaml and some further R-specific dependencies in environment.r. The below commands should get you a reproducible environment
• if conda is not available or broken, then installing R manually in your operating system, followed by Rscript environment.r should also work):

conda env create -n holocene -f environment.yaml conda activate holocene Rscript environment.r `

Input data

(in directories c14mat/, paleoclim/, and data/)

Note that the input data are already available in pre-processed binary (MATLAB) format.

Archaeological data (C14)

The C14 dates from the p3k14c dataset were downloaded in March 2023 from https://www.nature.com/articles/s41597-022-01118-7/tables/4. Updates can be found at https://www.p3k14c.org The p3k14c csv/xls file can be converted by read_dASIS.m (see outcommented settings) to c14mat/p3k14c.mat (needed by merge_C14data.m)

The new compilation refers to the database dASIS which in turn collects previous databases and adds own dates compiled by Gronenborn/Antunes at LEIZA Mainz. The xls file c14mat/C14_dASIS.xlsx was converted by runing read_dASIS.mto c14mat/C14_dASIS.mat.

Using the script merge_C14data.m, all binary files for pooled and gridded were produced

• c14mat/C14europe[1-64].mat: radiocarbon dates for Europe on a coarse grid
• c14mat/C14_europe0.mat: radiocarbon dates for Europe pooled
• c14mat/C14_[cont].mat: radiocarbon dates for continents
• c14mat/C14_NIreland.mat: radiocarbon dates for North Ireland
• c14mat/C14_dASIS.xlsx: compilation of (dASIS) C14v dates in XLS format

All of the collected data were available under open access licences. These are RADON and RADONb (Rinne et al. 2024, Hinz et al. 2012); P3K14C under CC0-1.0 (Bird et al. 2022); AIDA under CC-BY-4.0 (Palmisano et al. 2022); MedAfriCarbon under CC-BY-4.0 (Lucarini et al. 2020); Aegean History under CC-BY-4.0 (Ktasanis et al. 2020); Chapple (2018) under CC-BY-4.0; EUBAR (Capuzzo 2014); and NAKALA under CC-BY-4.0 (Perrin 2021)

Paleoclimate data

Collected paleoclimate proxy time-series and meta-info in

• paleoclim/InEur_27_920.mat: index ('InEur') to proxies in EU domain
• paleoclim/proxydescription_488_0_11.00_1.70.mat: MATLAB structure ("evinfo") with paleoclimate info for 209 global records. the 98 used indices (for Europe) are given by 'InEur'

meta-info: e.g., "Proxy" for proxy type (e.g. d18O), "Plotname" for site name, "Latitude", "Longitude" or "Source" for bibtex-key of original reference

• evinfo.time: cells with vectors containing the time values of a record
• evinfo.value: cells with vectors containing the normalized values

Note that the MATLAB structure 'evinfo' and its fields can be converted into an R dataframe using the R.matlab package

misc time-series or fields

• SA_spd_rgr.mat: RGR South America from digitized and calculated from Riris et al 2019
• Steinhilber2012_Solar.dat: Total Solar Irradiance (TSI) from Steinhilber et al 2012
• bog_std.mat: Northern Irish bog data by Rowan McLaughlin
• seamask_norm_0.05.mat: landmask Europe at 0.05 degree resolution

The paleoclimate data was obtained from the PANGAEA and NOAA archives. Many of the data not available in these archives was digitized from the graphics in the original publication. All data sources are cited in the supplementary material.

If you use the combined dataset provided here, we ask you to cite this publication, and -- if highlighting a particular source -- also the original dataset.

Execution

All scripts can be invoked by a single master shell command: ./master.sh this serial processing may take some time depending on computational resources; time consuming script processing such as of 'spd_growth.r' or 'cluster.r' can be run at a HPC in parallel mode (see also slurm.sh)

• for running R-scripts (e.g."r-example.r"): start 'R' terminal and type 'source("r-example.r")' or from (bash) terminal, type 'Rscript r-example.r'
• for running MATLAB-scripts (e.g."m-example.m"): start MATLAB terminal and type 'm-example' or from (bash) terminal, type 'matlab -nodesktop -r "try; m-example; catch; end; quit" '

Sequence of script execution and dependencies (see also master.sh)

1. grid_growth.r # calculates spatial statistics of SPDs on a 4°x4° grid
2. collect.r # collect/merge grid cells and prepares clustering
3. cluster.r # create region clusters of C14 sites
4. makegrid.m # spatial kriging on a grid based on cluster points (uses cldistance.m, makegridregions.m)
5. spdgrowth.r # calculates Summed Probability Density (SPD) and related growth (RGR) for each region and time slice (uses movavg.r) # reads PrePop clusti writes AllPop for time slices
6. spdpooled.r # calculates SPD and RGR for pooled method, dreads C14europe0 (or C14EA, C14NIreland), writes AllPop_all
7. plotRGR.m # process and plot calculated growth rates (RGR) (uses movavg.m, calcaravgrgr.m); reads AllPop writes AllPoptagall avgrgr RGR_Comp.png

Process climate proxy data

1. proxy_dtw.m # applies DTW writes dtw_proxydata (uses dtw.m by T. Felty, movweighavg.m)
2. collect_ts.m # integrate and smooth time-series, merges DTW time segments data nd includes other times series (e.g. RGR, solar forcing) (uses movavg.m, movweighavg.m); reads AllPopEAall avgrgr data/* writes targetts0

collects RGR from time slices and plot_varmap_slice %reads AllPop_i

1. glmloop.r # run GLM model for different sets of input variables (uses dolgm.r) # reads targetts_0, writes glmres*

[ 4. addlogitres.m # add logit model results to matrix ] overlapts.m # calc and plot overlap between time-series (uses addlogitres.m and calcoverlap.m)

License

If not stated otherwise, the entire analysis software is licensed under the GNU Public License version 3 or later. See http://www.gnu.org/licenses/gpl-3.0.txt for the complete terms.

The collection of archaeological data and the collection of paleoclimate data is licensed under the Creative Commons Attribution (CC-BY-4.0) license.

Documentation

See text and equations in Material and Methods of Wirtz et al, Multicentennial cycles in continental demography synchronous with solar activity and climate stability, subm.