Comparative analysis of behavioural repertoires for Mahogany glider and Brushtail possum using accelerometer loggers and machine learning algorithms

Gliding has evolved independently as an isolated adaptive event within many vertebrate taxa. Yet, the underlying selection forces that led to these innovative adaptations remain ambiguous, especially in species that preclude direct observation. Our study utilized accelerometry and machine learning algorithms to compare the behavioural repertoires of two sympatric species, the Mahogany glider (Petaurus gracilis) and brushtail possum (Trichosaurus vulpecula), as to explore previously proposed selection pressures such as energy expenditure (VeBA), canopy use and ground avoidance measured by activity budgets. We found that mahogany gliders on average expend more activity-related energy than brushtail possums but at different stages throughout the day. Canopy use was observed to be greater amongst mahogany gliders than brushtail possums, and we observed frequent ground use in brushtail possums yet none in mahogany gliders. The study found strong evidence to support ground avoidance as a potential driver for gliding evolution. The implications of these findings are important when considering the lack of knowledge surrounding evolved gliding behaviours in marsupials. Furthermore, the use of accelerometers and machine learning algorithms in behavioural studies has proven to be a robust and informative method and should be incorporated into future studies to understand the evolution of gliding behaviour.