Estimating pre-European population size for Australia

R translation of the Australia-wide 'hunter-gatherer' (non-agropastoralist) population estimate pipeline from Zhu et al. 2021 (Global hunter-gatherer population densities constrained by influence of seasonality on diet composition. Nature Ecology and Evolution 5:1536-1545) supplementary source code in Python.

This repository is embedded within a larger analysis described in a separate repository: AusIndigN

Focal manuscript

Williams, AN, R Tobler, B Griffiths, S Ulm, MC Nitschke, MI Bird, S Ingrey, F Saltré, K Beller, IJ McNiven, N Pitt, L Russell, A Paterson, C Wilson, CJA Bradshaw. Large size of the Australian Indigenous population prior to its massive decline following colonial invasion. Nature Human Behaviour In review

Contents

• scripts/estimate_australia.R: R implementation of the Australia-wide estimate
• INPUT/AUS.geojson: Australia polygon used for area-weighted intersections
• OUTPUT/globe_S0.nc, OUTPUT/globe_S1.nc, OUTPUT/globe_S2.nc: bundled FORGE global outputs copied from Zhu et al. (2021)
• OUTPUT/Australia_population_estimate.txt: generated summary report

Required R packages

lwgeom, ncdf4, sf

Optional arguments:

• Rscript Source/estimate_australia.R --scenario S2 (most likely scenario from Zhu et al. 2021)
• Rscript Source/estimate_australia.R --ci-level 0.95 --density-cv 0.7 (an assumed proportional standard deviation on density; not measured)
• Rscript Source/estimate_australia.R --output-dir OUTPUT --country-geojson INPUT/AUS.geojson (Australia polygon)

Current default assumptions

• preferred run is S2.
• 95% confidence interval based on assumed density uncertainty, not scenario spread
• default density uncertainty assumption is SD = 0.7 × mean density, following Bradshaw et al. (2021) (doi:10.1038/s41467-021-21551-3)
• confidence interval computed with log-Normal distribution to ensure population total interval remains positive