Stochastic epidemiological models to test the origin and demographic impact of smallpox on Aboriginal Australians in the 18th Century
Nitschke, MC, AN Williams, SD Ingrey, B Griffiths, N Pitt, L Russell, S Ulm, K Beller, JR Hore, MI Bird, SH Fatima, IJ McNiven, F Saltré, A Bashford, C Wilson, CJA Bradshaw. Stochastic models indicate rapid smallpox spread and mass mortality of Indigenous Australians after colonial exposure. Nature Human Behaviour. In review
see also (now out-of-date) pre-print:
Nitschke, MC, AN Williams, SD Ingrey, B Griffiths, N Pitt, L Russell, S Ulm, K Beller, JR Hore, MI Bird, SH Fatima, IJ McNiven, F Saltré, A Bashford, C Wilson, CJA Bradshaw. 2025. Stochastic models indicate rapid smallpox spread and mass mortality of Indigenous Australians after colonial exposure. Research Square doi:10.21203/rs.3.rs-5683492/v1
The impact of smallpox (variola) on Aboriginal communities in Australia beginning in 1789 was catastrophic and continues to cause intergenerational trauma. Historically biased perspectives and contemporary misinformation of the disease’s introduction and spread impedemodern-day truth-telling and subsequent reconciliation and national healing. Understanding whether the disease entered and spread from pre-colonial Makassan (Indonesian) trade along the northern coast, or from the First Fleet’s arrival in south-eastern Australia in 1788, is necessary to estimate demographic impact. We developed stochastic, multi-patch epidemiological models supported by systematic evaluation of historical observations to test hypotheses regarding possible disease entry points, spread rate, and demographic impacts. Our models confirm that entry of the disease was in south-eastern Australia. Even under ideal conditions and with higher-than-probable infection rates, simulations show that smallpox was unlikely to reach Sydney from a northern entry. Due to a high mortality rate, the epidemic appears to have been limited to the south-eastern coastal margins of Australia and along major intersecting river systems such as the Murray and Lachlan Rivers. There is no evidence that the 1789 epidemic was Australia-wide. Assuming a 60% lethality based on global data, the loss of 40,000–240,000 people would have occurred in these regions. While catastrophic to traditional Indigenous lifeways in the southeast, the disease also provided the catalyst for population decline andmarginalisation of Indigenous people in the face of expanding European populations. It seems unlikely that other parts of Australia were affected by the initial epidemic, and we recommend revisiting previous assumptions of the subsequent impacts to Indigenous societies by other diseases and frontier violence. We warn readers that the content of this study is confronting and possibly distressing.
Scripts (Matlab)
- main script:
OneStage.m
Movement.m: simulates movement between neighbouring patchesSpread.m: simulates disease spread through contact within each patch (susceptible to exposed S → E)adaptive_tau.m: adaptively chooses tau for tau-leap that optimizes spread (within Spread.m function)RecoverDeath.m: simulates the other stages of the disease (exposed to infected E → I, and infected to recovered I → R)Middle.m: simplified global model (no age structure, 4 disease stages, 1 infectious stage)Full_Model.m: simplified global model (sets parameters and initial conditions)
Pox_patch_518A.f90: Fortran 90 version of Matlab code for faster processingAreas.txt: Areas of each patchCoordinates.txt: Coordinates of the center of each patchPatchProbs.txt: Probability assigned to each patch for non nearest-neighbour movementPopulations.txt: Population of each patch for initial movement probabilityQ.txt: an M×M matrix whose entry Q(i,j) represents the baseline probability of moving from patch i to patch j
We acknowledge the sovereign Traditional Owners and custodians (First Nations) of the unceded lands and seas where we live and work, including Kaurna in Tarndanya/Adelaide (M.C.N., C.J.A.B., F.S., S.H.F.), Bidiagal in Warrane/Sydney (A.N.W., N.P., J.R.H., A.B.), Dharawal in Kamay/Botany Bay (K.B., S.D.I.), Kulin Nation in Naarm/Melbourne (B.G., I.J.M., L.R.), Yirrganydji and Gimuy Walubara Yidinji in Gimuy/Cairns (S.U., M.I.B.), Dharawal in Woolungah/Wollongong (F.S.), Ngarrindjeri of Murrundi/lower Murray River, Kurangk/Coorong, and eastern Fleurieu Peninsula (C.W.), Palawa in Nipaluna/Hobart, lutruwitra/Tasmania (C.W.), and Peramangk in Bukatila/Mount Lofty Ranges (C.J.A.B.). We also recognise the deep historical and cultural harm our truth-telling exposes, and we commiserate with all First Nations peoples of Australia. We thank J. Llewelyn for initial discussions.
Funded jointly by the Australian Research Council Centre of Excellence for Indigenous and Environmental Histories and Futures (CE230100009), and the Australian Research Council Centre of Excellence for Australian Biodiversity and Heritage (CE17010001). L.R. supported by an Australian Research Council Laureate Fellowship (FL190100161). B.G. supported by an Australian Research Council Discovery Early Career Researcher Award (DE220100203). A.B. and N.P. supported by an Australian Research Council Laureate Fellowship (FL200100144).
