Many seabird species breed in colonies counting up to hundreds of thousands of individuals. Life in such crowded colonies might require special coding–decoding systems to reliably convey information through acoustic cues. This can include, for example, developing complex vocal repertoires and adjusting the properties of their vocal signals to communicate behavioural contexts, and thus regulate social interactions with their conspecifics. We studied vocalisations produced by the little auk (Alle alle)—a highly vocal, colonial seabird—over mating and incubation periods on the SW coast of Svalbard. Using passive acoustic recordings registered in a breeding colony, we extracted eight vocalisation types: single call, clucking, classic call, low trill, short call, short-trill, terror, and handling vocalisation. Calls were grouped by production context (based on the typically associated behaviour), to which a valence (positive vs negative) was later attributed, when possible, according to fitness threats, i.e. predator or human presence (negative) and promoters, i.e. interaction with a partner (positive). The effect of the putative valence on eight selected frequency and duration variables was then investigated. The putative contextual valence significantly affected the acoustic properties of the calls. Calls assigned positive valence had higher fundamental frequency and spectral centre of gravity as well as shorter sound duration than these assigned negative valence. These results indicate that the little auk’s vocal communication system may facilitate expression of complex behavioural contexts, and seems to include vocal plasticity within vocalisation types—however, more data are necessary to better understand this effect and possible interplays of other factors.

Bioacoustic research spans a wide range of biological questions and applications, relying on identification of target species or smaller acoustic units, such as distinct call types. However, manually identifying the signal of interest is time-intensive, error-prone, and becomes unfeasible with large data volumes. Therefore, machine-driven algorithms are increasingly applied to various bioacoustic signal identification challenges. Nevertheless, biologists still have major difficulties trying to transfer existing animal- and/or scenario-related machine learning approaches to their specific animal datasets and scientific questions. This study presents an animal-independent, open-source deep learning framework, along with a detailed user guide. Three signal identification tasks, commonly encountered in bioacoustics research, were investigated: (1) target signal vs. background noise detection, (2) species classification, and (3) call type categorization. ANIMAL-SPOT successfully segmented human-annotated target signals in data volumes representing 10 distinct animal species and 1 additional genus, resulting in a mean test accuracy of 97.9%, together with an average area under the ROC curve (AUC) of 95.9%, when predicting on unseen recordings. Moreover, an average segmentation accuracy and F1-score of 95.4% was achieved on the publicly available BirdVox-Full-Night data corpus. In addition, multi-class species and call type classification resulted in 96.6% and 92.7% accuracy on unseen test data, as well as 95.2% and 88.4% regarding previous animal-specific machine-based detection excerpts. Furthermore, an Unweighted Average Recall (UAR) of 89.3% outperformed the multi-species classification baseline system of the ComParE 2021 Primate Sub-Challenge. Besides animal independence, ANIMAL-SPOT does not rely on expert knowledge or special computing resources, thereby making deep-learning-based bioacoustic signal identification accessible to a broad audience.

Wild harbour porpoises (Phocoena phocoena) mainly forage during the night and, because they rely on echolocation to detect their prey, this is also when they are most acoustically active. It has been hypothesised that this activity pattern is a response to the diel behaviour of their major prey species. To test this hypothesis, we monitored the acoustic activity of two captive harbour porpoises held in a net pen continuously during a full year and fed by their human keepers during daylight hours, thus removing the influence of prey activity. The porpoises were exposed to similar temperature and ambient light conditions as free-ranging animals living in the same region. Throughout the year, there was a pronounced diel pattern in acoustic activity of the porpoises, with significantly greater activity at night, and a clear peak around sunrise and sunset throughout the year. Clicking activity was not dependent on lunar illumination or water level. Because the porpoises in the pen are fed and trained during daylight hours, the results indicate that factors other than fish behaviour are strongly influencing the diel clicking behaviour pattern of the species.

Graphical AbstractBrief summary of the early-career job market in marine biology and conservation.