Event Abstract

Rabies spread modelling within wild dog populations in northern Australia

  • 1 Sydney School of Veterinary Science, Faculty of Science, The University of Sydney, Australia
  • 2 School of Animal and Veterinary Sciences, Charles Sturt University, Australia
  • 3 Faculty of Veterinary Medicine, University of Montreal, Canada

Introduction Australia, a canine rabies free country, is threatened by the current spread of the disease across the Indonesian Archipelago. This spread has brought rabies to only 300 km from northern Australia. A high risk of incursion has been identified in specific areas of northern Australia (Sparkes et al., 2015), which are characterised by the presence of remote Indigenous communities. In addition, Australia is home to a large wild dog population (including dingoes, feral dogs and dingo-dog hybrids) which is distributed throughout most of the Australian continent. Hence, the incursion of rabies into wild dog populations in Australia would make disease control very challenging and would represent a threat to human health, domestic animals and wildlife populations in Australia. In this context, disease spread models can be used to predict the spatial spread of infectious diseases and evaluate mitigation strategies for responding to an incursion. However, reliable data on dingo density ─ a key ecological parameter for wildlife disease spread modelling ─ is lacking in northern Australia and has been identified as a major research gap in the literature (Gabriele-Rivet et al., 2019). The objectives of this project are to 1) generate data on wild dog ecology, including movement and density distributions, and 2) predict the spread of rabies within wild dog populations, based on parameters relevant to the wild dog populations in northern Australian. Materials and methods The study area selected is the Northern Peninsula Area region (NPA), located on the tip of Cape York Peninsula, Queensland. The NPA covers a land area of 1 072 sq. km, and includes 5 remote Indigenous communities located close to each other (see Figure 1). Using passive infrared motion detectors, a camera-trap study was conducted between 9th May 2016 and 15th May 2017. Twenty-eight camera traps were deployed within 21 road-based stations, systematically separated by 2 kilometer intervals. In addition, two focal point stations were established, located at the refuse dump and the local abattoir. All photographs collected were carefully examined and assigned to either ‘Domestic-like’, ‘Dingo-like’ or ‘Unsure’ categories based on information indicating domestication and dingo phenotypic features (for example, erect ears, tapered muzzle and bushy tail). Profiles of identifiable dingoes (i.e. Marked) were developed using combinations of distinctive natural characteristics such as color patterns, tail shape and size, body size, sex and scars. Individuals which did not display distinctive characteristics and therefore could not be linked to any of the ‘Marked’ profiles were categorised as ‘Unmarked’. A habitat mask shapefile was created which delimitated the ocean shore of Cape York Peninsula and excluded all habitats unsuitable for dingoes ─ such as water areas, mangroves and human populated areas comprising the five Indigenous communities. A maximum likelihood spatially explicit mark-resight model was used to estimate the density of dingoes across this habitat mask, while allowing population density to vary over space and between seasons (dry vs wet seasons). The effects of various covariates on the detection probability function were explored such as the type of camera stations (road-based vs focal point stations) and season. The best-fitting model was selected based on Akaike’s Information Criterion. Density estimates and information on dingo movement obtained from the camera-trap study are being used to parameterise a stochastic spatial model of rabies spread in wild dogs. This agent-based, discrete time model is implemented in Python, based on a daily time step, over a time period of 1 year or longer. Following the density distributions in the NPA obtained from the camera-trap study, the model randomly distributes centroid points across the study area (excluding areas unsuitable for dingoes). Each point represents the center of a home range for a pack of dingoes. Using a variety of home range estimates derived from the literature, a circular bivariate (2 dimensional) Gaussian distribution is fitted for each dingo to obtain the probability density function that the dingo is located at any specific location around the centroid of its home range. Probability of contact between each pair of dingoes simulated within the model is calculated at each time step based on the probability density function estimated for each dingo and the distance between the two home range centroid points. Disease spread is then simulated following a SEIIR (Susceptible – Exposed – Infectious Pre-clinical – Infectious Clinical – Dead) model. It is parameterised based on published canine rabies pathogenesis information and the probability of contact between each modelled individual. Outputs generated by the model ─ such as the number of individuals infected, the length of the epidemic and the spatial distribution of the spread over time ─ are assessed. Results The twenty-eight camera traps captured 1.374 million photographs over a total of 7 648 camera trap days. Dingoes were identified in 1.3% of these photographs, from which 10 030 photographs displayed marked dingoes compared to 510 photographs containing unmarked dingoes. The marked individuals consisted of a total of 66 unique dingo profiles, over half of which were sighted at multiple camera stations. Of these, the maximum distance separating cameras which detected the same individuals (i.e. maximum movement observed per dingo) ranged between 0.6 and 11.3 km (median 4.0 km). Figure 1 illustrates the layout of the camera stations across the study area, with the number of sightings (marked and unmarked combined) per camera trap day. Final density estimates and preliminary results in relation to the stochastic spatial rabies spread model will be presented. Conclusion The present study provides data on wild dog density in northern Australia, which currently represents a research gap in the literature on Australian wild dog ecology. Using this information, the model developed will help predict the spread of rabies in wild dogs and ultimately, provide the tools needed to better prepare for ─ and respond to ─ a potential rabies incursion in northern Australia. Overall, this project will help protect Indigenous communities, wildlife populations and the ecosystem in Australia.

Figure 1


This project is funded by the Australian Government Department of Agriculture and Water Resources. We acknowledge the Australian Government, The University of Sydney and the Natural Sciences and Engineering Research Council of Canada for student grants. We would like to thank Rahel Sollman and Peter Fleming for their assistance in the camera-trap study design and analysis and Charlotte Nury for her help in the individual recognition of the photographs.


Sparkes J, Fleming P, Ballard G, Scott‐Orr H, Dürr S, Ward MP. Canine rabies in Australia: a review of preparedness and research needs. Zoonoses Public Health (2015) 62(4):237-53. Gabriele-Rivet V, Arsenault J, Wilhelm B, Brookes VJ, Newsome TM, Ward MP. A scoping review of dingo and wild-living dog ecology and biology in Australia to inform parameterisation for disease spread modelling. Frontiers in Veterinary Science (2019) 6:47. doi: 10.3389/fvets.2019.00047

Keywords: disease modelling, wild dogs, dingoes, Australia, Rabies

Conference: GeoVet 2019. Novel spatio-temporal approaches in the era of Big Data, Davis, United States, 8 Oct - 10 Oct, 2019.

Presentation Type: Student oral presentation

Topic: Spatio-temporal surveillance and modeling approaches

Citation: Gabriele-Rivet V, Brookes VJ, Arsenault J and Ward MP (2019). Rabies spread modelling within wild dog populations in northern Australia. Front. Vet. Sci. Conference Abstract: GeoVet 2019. Novel spatio-temporal approaches in the era of Big Data. doi: 10.3389/conf.fvets.2019.05.00074

Copyright: The abstracts in this collection have not been subject to any Frontiers peer review or checks, and are not endorsed by Frontiers. They are made available through the Frontiers publishing platform as a service to conference organizers and presenters.

The copyright in the individual abstracts is owned by the author of each abstract or his/her employer unless otherwise stated.

Each abstract, as well as the collection of abstracts, are published under a Creative Commons CC-BY 4.0 (attribution) licence (https://creativecommons.org/licenses/by/4.0/) and may thus be reproduced, translated, adapted and be the subject of derivative works provided the authors and Frontiers are attributed.

For Frontiers’ terms and conditions please see https://www.frontiersin.org/legal/terms-and-conditions.

Received: 31 May 2019; Published Online: 27 Sep 2019.

* Correspondence: Mx. Vanessa Gabriele-Rivet, Sydney School of Veterinary Science, Faculty of Science, The University of Sydney, Sydney, Australia, vgab5928@uni.sydney.edu.au