D-PorCCA, a new tool to study the acoustic life of harbour porpoises

Despite decades of research, the social life of harbour porpoises in their natural environment is almost entirely unknown, although recent data suggest they are very social. Harbour porpoises produce only highly stereotyped clicks (narrow-band high frequency -NBHF-) that are ideal for monitoring purposes. These are emitted in click trains: a series of clicks with regular or gradually changing inter-click intervals. The patterns in repetition rates are indicative of the behaviour of the clicking porpoise, which can be broadly divided into three main categories: orientation or travelling, feeding, and socialising. This means passive acoustic monitoring (PAM) methods can be used to study the behaviour of harbour porpoises. In this project, a series of algorithms to study the behaviour of NBHF species from acoustic recordings were developed and put together in a standalone application, D-PorCCA. The algorithms were developed and tested using data from Scotland and Denmark collected using static and mobile PAM devices. D-PorCCA and all algorithms were developed in MATLAB. The application includes a user-friendly interface, a simple transient sound detector, a new porpoise click classifier (PorCC), and algorithms to identify behaviours. Additionally, the acoustic behaviour of harbour porpoises was studied while interacting with a solitary short-beaked common dolphin who inhabits the Firth of Clyde (Scotland). The performance the algorithms in D-PorCCA was tested against manual labelling, varying from 69.2% to 98.5% of accuracy. These steps are automated and provide researchers with pre-selected data and summary data such as ’positive porpoise minute’ and day/night activities. D-PorCCA has many functionalities and the user can easily inspect and verify the data. The main results of this project are four. First, evidence was found of interspecies communication between a solitary short-beaked common dolphin and harbour porpoises in the Firth of Clyde (Scotland). Second, a series of patterns consistent with social calls reported in the literature as well as new ones were found for the first time using PAM in the wild. The most striking patterns were those known as ’phrases’, which are a series of similar calls produced in a short period of time. Third, the feeding and socialising patterns overlap, limiting our ability to distinguish between them. Lastly, the algorithms can be used in other species that produce NBHF clicks. These results suggest D-PorCCA can potentially be used for behavioural studies of wild harbour porpoises as well as other NBHF species, and to fill knowledge gaps in our understanding of the behaviour of these elusive species. Moreover, it has potential for application in large monitoring project, such as the Static Acoustic Monitoring of the Baltic Sea Harbour Porpoise, known as the SAMBAH Project.Despite decades of research, the social life of harbour porpoises in their natural environment is almost entirely unknown, although recent data suggest they are very social. Harbour porpoises produce only highly stereotyped clicks (narrow-band high frequency -NBHF-) that are ideal for monitoring purposes. These are emitted in click trains: a series of clicks with regular or gradually changing inter-click intervals. The patterns in repetition rates are indicative of the behaviour of the clicking porpoise, which can be broadly divided into three main categories: orientation or travelling, feeding, and socialising. This means passive acoustic monitoring (PAM) methods can be used to study the behaviour of harbour porpoises. In this project, a series of algorithms to study the behaviour of NBHF species from acoustic recordings were developed and put together in a standalone application, D-PorCCA. The algorithms were developed and tested using data from Scotland and Denmark collected using static and mobile PAM devices. D-PorCCA and all algorithms were developed in MATLAB. The application includes a user-friendly interface, a simple transient sound detector, a new porpoise click classifier (PorCC), and algorithms to identify behaviours. Additionally, the acoustic behaviour of harbour porpoises was studied while interacting with a solitary short-beaked common dolphin who inhabits the Firth of Clyde (Scotland). The performance the algorithms in D-PorCCA was tested against manual labelling, varying from 69.2% to 98.5% of accuracy. These steps are automated and provide researchers with pre-selected data and summary data such as ’positive porpoise minute’ and day/night activities. D-PorCCA has many functionalities and the user can easily inspect and verify the data. The main results of this project are four. First, evidence was found of interspecies communication between a solitary short-beaked common dolphin and harbour porpoises in the Firth of Clyde (Scotland). Second, a series of patterns consistent with social calls reported in the literature as well as new ones were found for the first time using PAM in the wild. The most striking patterns were those known as ’phrases’, which are a series of similar calls produced in a short period of time. Third, the feeding and socialising patterns overlap, limiting our ability to distinguish between them. Lastly, the algorithms can be used in other species that produce NBHF clicks. These results suggest D-PorCCA can potentially be used for behavioural studies of wild harbour porpoises as well as other NBHF species, and to fill knowledge gaps in our understanding of the behaviour of these elusive species. Moreover, it has potential for application in large monitoring project, such as the Static Acoustic Monitoring of the Baltic Sea Harbour Porpoise, known as the SAMBAH Project