Thesis topics

Investigate humpback whale feeding ecology based on visual observations, passive acoustic recordings, and biotelemetry. All data exist; no additional fieldwork is needed. In January 2014 the WAVES expedition (Centre for Whale Research, Curtin University and University of Alaska) deployed a high resolution “Lander” tag on a feeding humpback whale and followed it for 12 hours with a Biosonics DTX single-beam sonar running. The tag was recovered and contained good sensor data from 3-axis accelerometers, magnetic heading orientation, depth, temperature and GPS location, combined with sonar data of the krill patches the whale was feeding on. The tag data will allow the student to calculate feeding rates and to correlate these with biomass of the krill patch the whale passed through. The krill patch was remarkably fluid with time, the dynamics of this can be investigated and linked to how the feeding humpback exploited the krill patch.

This project investigates how urbanisation affects bat biodiversity in Perth by combining passive acoustic monitoring with environmental data to assess the influence of greenspace and urban noise on local bat species. By deploying acoustic recorders across a range of urban, peri-urban, and semi-natural sites, the study aims to identify which bat species are present, how their activity and diversity are shaped by habitat features and noise, and to generate new insights for urban conservation. The research will build on previous findings, refine detection methods, and produce a publicly accessible dataset and high-impact publication, offering students the chance to contribute to real-world biodiversity and conservation outcomes in urban environments.

This project investigates how fish detect and process underwater sound by modelling and measuring hearing sensitivity and directionality across four key fish hearing groups. Combining CT and micro-CT imaging, finite element modelling, scanning electron microscopy, and behavioural audiograms (with trained fish!), this research aims to understand how sound interacts with otoliths and hair cells, and how this varies by species, anatomy, age, and environmental conditions like water depth.

The project will test model predictions through controlled behavioural experiments using reward-based conditioning in a purpose-built tank. Findings will help develop transfer functions that link otolith motion to hearing thresholds, a key step toward predicting the hearing abilities of species that are difficult to test directly.

This research will deliver new insights to inform environmental assessments and improve management of noise impacts on fish in Australian and global waters.

CMST has a large archive of recorded ocean sound (30 years from various sites around the Australian continent) which shows seasonal visitation by dwarf minke, Bryde’s, sei, and fin whales. None of this acoustic data has been analysed in any depth for the migratory patterns or calling behaviour of the respective species, which remain poorly understood in the Australian context. The student would choose one of these species. Chapters may include song structure, geospatial habitat, seasonal/lunar/diel patterns of calling, song structure, non-song sounds, species distribution model, and the effects of noise.

In January 2014 the WAVES expedition (Centre for Whale Research and Curtin) collected almost 30 days of sonobuoy data from the Southern Ocean and east Antarctic ice edge.  The data sets are rich in biological signals (whales and seals mostly), physical sea noise (storm and ice asssociated) and man-made noise (seismic survey from Australian shelf break at 1200 n mile).  Several of the sonobuoys were deployed in broadband mode and have some surprising signals in the high frequencies. Most sonobuoy deployments allow bearing estimation of the sources detected. The data can be analysed for biological signals or physical sea noise sources and offers an opportunity for a student to become involved in bioacoustic analysis while studying the Antarctic environment.

Each summer a large number of mulloway enter the Swan River to spawn, forming an aggregation each night around the Mosman Bay area. CMST deploys an underwater noise logger in Mosman Bay, recording advertisement calls as males attempt to attract females. This aggregation moves up and down the Swan River in the late afternoon and around sunset, varying with the season, lunar phase and tide times.  A number of projects may be conducted at this location from analysing mulloway calls or monitoring aggregation movement patterns, to assessing the level of call discrimination by attracting female mulloway with Acoustically Baited Remote Underwater Videos (ABRUVS).  This project will require analytical skills and Matlab programming. Boating experience preferential.

The marine soundscape can be split into its biophony (the sounds of whales, dolphins, fish, crustaceans etc.), geophony (the sounds of wind, rain, waves, ice etc.) and anthropophony (the sounds of human/industrial operations). Ship traffic is the most persistent source of man-made noise in the marine environment—with potentially significant bioacoustic impacts on marine fauna, most of which rely heavily on acoustics for their critical life functions. CMST has recorded the marine soundscape around Australia for 25 years at various sites. Using publicly available position logs of large vessels, we can 1) compute received levels of individual ships, 2) calculate source levels of individual ships by sound propagation modeling, and 3) determine the contribution of shipping to the local noise budgets. We may then map ship noise in specific regions (eg close to marine parks or critical habitat), we can model various scenarios for noise reduction in critical habitat (eg ship slow-down, rerouting). We can also model the risk of ship strike. This project will suit a mathematically skilled student with some experience in scientific software development, data analysis and numerical modelling. An acoustic background is not necessary.

Endangered black cockatoo species of Western Australia rely on large and acoustically complex habitats, making long-term monitoring both essential and challenging. Being “noisy”, they produce sounds that differ by species, age, gender and behaviour. We aim to explore whether passive acoustic listening can provide additional data on population size, distribution and demographics.

One of the objectives of the project is to to develop and evaluate machine-learning methods for the automated detection of Baudin and Carnaby cockatoo vocalisations from long-duration environmental audio recordings. The student will work with real-world acoustic datasets collected in Western Australian soundscapes and investigate how different signal processing and classification approaches perform under varying environmental conditions (e.g. background noise, seasonal variation, and overlapping biological sounds).

We have preliminary recordings of cockatoos near the Curtin University Bentley campus, and in John Forrest National Park. The Honours students will be involved in additional field work, including recordings and visual observations, establish a call repertoire of these two species, correlate calls with behaviour and demographic parameters, and potentially look at changes in calling behaviour as a function of human disturbance.

The project will involve data exploration, annotation analysis, feature extraction from audio, model training and evaluation, and interpretation of detection performance. The outcomes will contribute to more efficient and scalable tools for ecological monitoring and conservation, while providing the student with hands-on experience in applied machine learning and ecoacoustics.