Enhancing surveillance of the northern Australian feral pig population for African swine fever and other high impact pests and diseases

With the ongoing global spread of African Swine Fever (ASF), particularly in the Asia-Pacific region, ASF has become an emerging biosecurity concern for northern Australia. The goal of this project was to take a science-based approach to feral pig surveillance in the context of enhancing preparedness for and early detection of ASF in Northern Australian landscapes. We specifically trialled the application of two modern techniques, genomic sequencing (Component 1) and aerial thermal imagery (Component 2) to assess their efficiency and effectiveness for enhancing feral pig surveillance.

Component 1
This research aimed to determine if it was possible to assess meta-population connectivity and landscape resistance of feral pigs across northern Australia. To achieve this, we obtained DNA from over 2,000 feral pig tissue samples collected by our network of partners from across northern Australia. A subset of 796 of these samples was genotyped using single nucleotide polymorphism (SNP) genetic markers from across the pig genome. Patterns of genetic diversity among individuals were mapped to illustrate feral pig population connectivity across northern Australia. We also applied landscape resistance modelling based on the genetic data to determine what geographic features may aid or hinder pig movement and thus gene flow. The results showed that feral pig population connectivity was far greater in north Queensland than in the Northern Territory, and connectivity was greater in the Northern Territory than in northern Western Australia. Genetic clustering analyses identified major geographic features contributing to feral pig population connectivity across the landscape. There were regional differences in the effects of environmental variation on feral pig population connectivity, as estimated from genetic data. Across northern Australia and in the Northern Territory, topographic roughness was the primary limitation on gene flow among feral pig populations, whereas watercourses and dry season habitat suitability were better predictors of feral pig population connectivity in north Queensland. The SNP genetic marker panel and analytical pipeline provide an efficient and powerful approach to extend the mapping of population connectivity of feral pigs nationwide.

The key outcomes from this research were:
• A blueprint upon how to create and sustain a network of partners who will freely provide feral pig tissue samples.
• A network of partner organisations across northern Australia from which feral pig tissue can be collected into the future.
• A large panel of informative SNP genetic markers mapped to the pig reference genome, customised for the Australian feral pig population based on samples from northern, southern, eastern and western Australia.
• An extensive feral pig genetic database for northern Australia, illustrating the spatial extent of feral pig meta-populations and locations of population discontinuities.
• Regional maps of modelled patterns of feral pig movement across the landscape, featuring predicted areas of high and low feral pig movement.
Component 2
This research aimed to assess the utility of remotely piloted aircraft fitted with a thermal camera to detect both alive and dead feral pigs in the northern Australian savanna landscape and measure their body temperature remotely. Previous studies have shown that aerial thermal photography can detect free-ranging feral pigs in southern landscapes. However, no trials have been undertaken in northern landscapes, and as far as we are aware the proportion of feral pigs within a population that may not be detected using aerial thermal survey has never been assessed.
Here, we selected a study area representative of a large proportion of northern Australian landscapes where feral pigs are found. Remotely piloted fixed wing aircraft were used as the survey platform. The Remotely Piloted Aircraft Systems (RPAS) surveys were conducted at the peak of the wet season (February) and surveyed a known number of feral pigs within a large, fenced paddock comprised of mixed woodland savanna. Each pig was fitted with a GPS-based tracking collar and body surface temperature monitor. Aerial thermal and optical images were collected every 4 hours over four days using a fixed-wing remotely piloted aircraft fitted with dual-rigged thermal (radiometric) and optical (red-green-blue) cameras. The study found that detection probabilities for feral pigs during daylight hours were less than 20% (i.e., only 1 in every 5 present feral pigs was detected due to obscuration by canopy cover). However, this rose to around 60% during surveys undertaken in the late afternoon (after 6 pm). Termite mounds and wallabies produced a heat signature very similar to that of a feral pig, and therefore we recommend that aerial thermal photography be accompanied by optical imagery. Aerial surveys during darkness are not recommended because it was not possible to obtain simultaneous optical imagery. The next stage is to provide the imagery from this study to software developers who could build upon our recommendations for northern landscapes. We also found that aerial thermal photography may prove challenging for identifying feral pigs with fever-induced elevated body temperatures, but further investigation is required.
The key outcomes from this research were:
• Documented methodology for using remotely piloted aircraft and aerial thermal imagery to complement existing feral pig surveillance activities in northern Australia.
• Guidelines to support image capture with the highest probability of autonomous feral pig detection within northern Australia woodland savanna during the wet season
• Extensive collections of simultaneous optical and thermal imagery from a known number of feral pigs with accurate geographic location data for the training and ground-truthing of Artificial Intelligence (AI) and Deep Learning routines.