New research uses AI to tune into nature’s pest controllers

Photo caption: UC Professor Stuart Parsons and UniSC research assistant Maggie Campbell-Jones in front of a cotton bale.

Led by Professor Stuart Parsons, Executive Dean of Science at Te Whare Wānanaga o Waitaha | University of Canterbury (UC) in collaboration with colleagues at the University of the Sunshine Coast and Queensland University of Technology in Queensland, the research is measuring how integrated pest management strategies and biodiversity corridors influence the diversity and activity of birds and bats known to support natural pest control in cotton farming.

“Our goal is to support farmers in their work to grow and maintain biodiversity,” Professor Parsons says. “This can be achieved through aspects of land management that will allows species of bats and birds that eat insect pests of cottons to access farms.”

Using sensors with wireless acoustic monitors and AI integration, the project aims to show how on-farm biodiversity can be measured at scale and used to help growers make more informed decisions about sustainable pest management.

“The farmers get recommendations on how to manage their landscape, to draw in beneficial birds and bats that provide natural pest control.

“Previous research has shown the economic benefit of natural pest suppression can be substantial. In the US, it’s estimated to be worth about $2 billion a year, and in Australia it varies, but it’s in the tens to hundreds of millions of dollars.”

He says attracting beneficial birds and bats could reduce pesticide use and strengthen farmer’s sustainability credentials and social licence.

Professor Parsons says birds such as Cisticolas (small insectivorous birds) appear to move into cotton crops as the plants grow taller, suggesting connected landscapes such as trees or river corridors may help beneficial species move from nearby woodland or river areas into crops where they can support natural pest control.

“They don’t like to cross a big open field but if they’ve got a connection, which could be the trees of a stock route, or the trees along a river, they will be there,” Professor Parsons says.

“The solar-powered sensors listen for sounds at a certain time of the day and night, and they record those sounds. Then it goes through several AI algorithms which determine what species of bird and bat are there. We then validate the results of the AI against human observation.

“The sensors are unique because they combine embedded AI, audible and ultrasonic recording, internet connectivity, and long-term autonomous field deployment. This allows us to identify birds and bats and send data remotely without having to regularly retrieve memory recording cards from the field.

“Birds are easier because their calls are designed for communication, so they tend to be quite species-specific. Bats are more complicated because echolocation is about locating an object in three-dimensional space, not social communication. Different bat species trying to achieve the same thing can produce similar calls,” Professor Parsons says.

“Bats also produce ultrasound, which is beyond human hearing and very directional. If the bat is pointing away from the microphone, or is further from the sensor, the call can look different. That’s why we validate the AI results against expert human analysis, so we know how accurate it is.”

Professor Parsons says working closely with growers has been central to the project, with farmers actively supporting the research and showing a strong commitment to understanding and improving biodiversity on their land.

While this project focussed on cotton farms in Australia, Professor Parsons says the approach could have future relevance for New Zealand horticulture, particularly in sectors where biodiversity, pest control and sustainable production are priorities.

He says the work points to the value of understanding how features such as riparian planting, biodiversity corridors and connected habitats can support beneficial species in agricultural landscapes.

The research project is funded by the Cotton Research and Development Corporation and the National Landcare Program, with the funding administered through the University of the Sunshine Coast (UniSC), where Professor Stuart Parsons is an adjunct, in collaboration with Queensland University of Technology (QUT).

/Public Release. View in full here.