29 July 2024
In certain areas of Australia, millions of sterile male fruit flies rain from the skies every two weeks. These Queensland fruit flies are reared to the peak of health in a special facility, then sterilised through irradiation, before being loaded into an aeroplane and dropped from the air. When the sterile males mate with local females, the females are unable to lay viable eggs.
This method effectively suppresses fruit fly populations, which cost Australian growers hundreds of millions of dollars a year in damaged fruit, pest control and lost market access opportunities. But is this too many sterile fruit flies to drop from the skies? TPM Whānau member Dr Tom Moore wants to know.
The Queensland fruit fly is an important pest of concern for Aotearoa New Zealand. Although there have been multiple detections in Aotearoa, the fly has not yet established a foothold. But this comes at a cost. In the most recent incursion, 11 male flies were caught on Auckland’s North Shore – at a cost of $18 million.
Tom is a quantitative ecologist who specialises in integrating scientific hypotheses into statistical models to address causal questions. He is particularly interested in understanding how invasive species establish themselves through data that spans across time and space. Aotearoa relies heavily on its biological resources, so this sort of agricultural research plays a pivotal role in our economic prosperity and shaping a sustainable future.
“Current methods of controlling fruit fly populations are effective,” says Tom, “but without a deeper understanding of the causal processes at work these could become less efficient due to future change, resulting in unnecessary costs.”
Tom has seed funding from Te Pūnaha Matatini to explore new methods of causal inference, which is the process of determining the independent effects of the individual parts that make up larger systems. Causal inference techniques drawing on interdisciplinary methods linking mathematics, computer science and statistics have become popular in other parts of the world, but are not yet widely used in Aotearoa.
Current statistical approaches can be limited in their ability to identify causal relationships. “It’s very common to collect a lot of data without a targeted question, chuck it in all in a model, and see what comes out the other side,” says Tom. “Richard McElreath calls this a ‘causal salad’. So while a certain model might make good predictions, it may be misleading in terms of causation, and unhelpful in planning interventions.”
Causal inference is a technique that considers how variables are related. This approach has been applied in other systems like economics, but is a new and developing technique in ecology. Developing causal models to support agricultural management decisions like the suppression of fruit flies has the potential to significantly improve our understanding of how to effectively intervene in these complex systems to improve food security in an uncertain future.
“By explicitly modelling cause and effect relationships in complex systems based on ecological theory,” Tom explains, “these techniques can evaluate if, and under what conditions, cause and effect relationships can be identified.” This project is an exciting opportunity to both develop complex systems theory, and to apply it to make a real difference to agriculture in Aotearoa.
Tom is working with a team including Te Pūnaha Matatini Principal Investigators Dr Will Godsoe and Dr Giulio Valentino Dalla Riva to develop a non-linear model of population dynamics using three years of weekly fruit fly trapping data.* They are also collaborating with entomologist Dr Lloyd Stringer from Plant and Food Research, and three colleagues from the from the University of Canterbury: statisticians Professor Elena Moltchanova and Dr Phillipp Wacker, and doctoral student in computational and applied mathematical sciences, Pooja Baburaj.
Tom and his team will then use this modelling approach to test out different biocontrol scenarios – exploring whether less frequent deployment of biocontrol measures can achieve the same population suppression. “If the model shows that you can reduce the frequency of biocontrol to monthly and the data looks the same, that’s a great outcome,” says Tom, “and would save a bit of money.”
The team is motivated to explore and showcase how causal inference can be used within the scientific process when defining hypotheses to generate more meaningful insights. Tom is especially excited about sharing this approach with his colleagues at Plant and Food Research and the broader research community, and hosting a workshop with Te Pūnaha Matatini colleagues to increase the capacity of the causal modelling of complex systems in Aotearoa. He hopes that this work will inspire interdisciplinary collaboration between agriculture, ecology and statistics.
Tom says that “we can ultimately use causal inference to predict where invasive species might spread, and how their population dynamics will manifest in a system, so that we can make evidence-based decisions on how to respond.”
“This is a way that we can solve real-world problems with innovative and evidence-based research.”
Illustrated by Chelsea Sullivan.
*This data was collected as part of the post factory pilot of SITplus fly production project (FF17001) that was funded by the Hort Frontiers Fruit Fly Fund, part of the Hort Frontiers strategic partnership initiative developed by Hort Innovation Australia, with co-investment from Macquarie University, New South Wales Department of Primary Industries, Plant and Food Research New Zealand, and contributions from the Australian government.