Call recognition and individual identification of fish vocalizations based on automatic speech recognition: An example with the Lusitanian toadfish

Boat noise affects meagre (Argyrosomus regius) hearing and vocal behaviour

Aquatic noise has increased in last decades imposing new constraints on aquatic animals' acoustic communication. Meagre (Argyrosomus regius) produce loud choruses during the breeding season, likely facilitating aggregations and mating, and are thus amenable to being impacted by anthropogenic noise. We assessed the impact of boat noise on this species acoustic communication by: evaluating possible masking effects of boat noise on hearing using Auditory Evoked Potentials (AEP) and inspecting changes in chorus sound levels from free ranging fish upon boat passages. Our results point to a significant masking effect of anthropogenic noise since we observed a reduction of ca. 20 dB on the ability to discriminate conspecific calls when exposed to boat noise. Furthermore, we verified a reduction in chorus energy during ferryboat passages, a behavioural effect that might ultimately impact spawning. This study is one of few addressing the effects of boat noise by combining different methodologies both in the lab and with free ranging animals.

Fish sounds and boat noise are prominent soundscape contributors in an urban European estuary

Passive acoustic monitoring is a valuable tool for non-intrusive monitoring of marine environments, also allowing the assessment of underwater noise that can negatively affect marine organisms. Here we provide for the first time, an assessment of noise levels and temporal soundscape patterns for a European estuary. We used several eco-acoustics methodologies to characterize the data collected over six weeks within May 2016 - July 2017 from Tagus estuary. Biophony was the major contributor dominated by fish vocalizations and the main driver for seasonal patterns. Maritime traffic was the major source of anthropogenic noise, with daily patterns monitored using 1584 Hz third-octave band level. This indicator avoided biophony and geophony, unlike other indicators proposed for the EU Marine Strategy Framework Directive. Furthermore, the frequency overlap between anthropophony and biophony demands precautionary actions and calls for further research. This study provides an assessment that will be useful for future monitoring and management strategies.

Boat noise interferes with Lusitanian toadfish acoustic communication

Anthropogenic noise is considered a major underwater pollutant as increasing ocean background noise due to human activities is impacting aquatic organisms. One of the most prevalent anthropogenic sounds is boat noise. Although motorboat traffic has increased in the past few decades, its impact on the communication of fish is still poorly known. The highly vocal Lusitanian toadfish (Halobatrachus didactylus) is an excellent model to test the impact of this anthropogenic stressor as it relies on acoustic communication to attract mates. Here, we performed two experiments to test the impact of boat noise on the acoustic communication of the Lusitanian toadfish. Using the auditory evoked potential (AEP) technique, we first compared the maximum distance a fish can perceive a boatwhistle (BW), the mate attraction acoustic signal, before and after embedding it in boat noise. Noises from a small motorboat and from a ferryboat reduced the active space from a control value of 6.4-10.4 m to 2.0-2.5 m and 6.3-6.7 m, respectively. In the second experiment we monitored the acoustic behaviour of breeding males exposed to boat noise playbacks and we observed an increase in the inter-onset interval of BWs and a disruption of the usual vocal interactions between singing males. These results demonstrate that boat noise can severely reduce the acoustic active space and affect the chorusing behaviour in this species, which may have consequences in breeding success for individuals and could thus affect fitness.

Rapid coral reef assessment using 3D modelling and acoustics: acoustic indices correlate to fish abundance, diversity and environmental indicators in West Papua, Indonesia

BACKGROUND

Providing coral reef systems with the greatest chance of survival requires effective assessment and monitoring to guide management at a range of scales from community to government. The development of rapid monitoring approaches amenable to collection at community level, yet recognised by policymakers, remains a challenge. Technologies can increase the scope of data collection. Two promising visual and audio approaches are (i) 3D habitat models, generated through photogrammetry from video footage, providing assessment of coral cover structural metrics and (ii) audio, from which acoustic indices shown to correlate to vertebrate and invertebrate diversity, can be extracted.

METHODS

We collected audio and video imagery using low cost underwater cameras (GoPro Hero7™) from 34 reef samples from West Papua (Indonesia). Using photogrammetry one camera was used to generate 3D models of 4 m2 reef, the other was used to estimate fish abundance and collect audio to generate acoustic indices. We investigated relationships between acoustic metrics, fish abundance/diversity/functional groups, live coral cover and reef structural metrics.

RESULTS

Generalized linear modelling identified significant but weak correlations between live coral cover and structural metrics extracted from 3D models and stronger relationships between live coral and fish abundance. Acoustic indices correlated to fish abundance, species richness and reef functional metrics associated with overfishing and algal control. Acoustic Evenness (1,200-11,000 Hz) and Root Mean Square RMS (100-1,200 Hz) were the best individual predictors overall suggesting traditional bioacoustic indices, providing information on sound energy and the variability in sound levels in specific frequency bands, can contribute to reef assessment.

CONCLUSION

Acoustics and 3D modelling contribute to low-cost, rapid reef assessment tools, amenable to community-level data collection, and generate information for coral reef management. Future work should explore whether 3D models of standardised transects and acoustic indices generated from low cost underwater cameras can replicate or support 'gold standard' reef assessment methodologies recognised by policy makers in marine management.

Individual vocal recognition across taxa: a review of the literature and a look into the future

Individual vocal recognition (IVR) has been well studied in mammals and birds. These studies have primarily delved into understanding IVR in specific limited contexts (e.g. parent-offspring and mate recognition) where individuals discriminate one individual from all others. However, little research has examined IVR in more socially demanding circumstances, such as when an individual discriminates all individuals in their social or familial group apart. In this review, we describe what IVR is and suggest splitting studies of IVR into two general types based on what questions they answer (IVR-singular, and IVR-multiple). We explain how we currently test for IVR, and many of the benefits and drawbacks of different methods. We address why IVR is so prevalent in the animal kingdom, and the circumstances in which it is often found. Finally, we explain current weaknesses in IVR research including temporality, specificity, and taxonomic bias, and testing paradigms, and provide some solutions to address these weaknesses. This article is part of the theme issue 'Signal detection theory in recognition systems: from evolving models to experimental tests'.

The Use of Soundscapes to Monitor Fish Communities: Meaningful Graphical Representations Differ with Acoustic Environment

Many marine animals produce sounds in several phases of their life cycles, either actively or as a byproduct of their activities, such as during mate attraction or when moving. Recent studies of underwater soundscapes have proved passive acoustic monitoring to be a cost-effective, non-invasive tool to understand ecological processes, especially when sampling in adverse conditions or at great depth. Four days of sound recordings at three seamounts from the Azorean archipelago were examined to assess the suitability of different sound graphical representations to characterize different acoustic environments that contrast in the contribution of vocal fish communities. Long-term spectrograms, sound pressure level, spectral probability densities and the Acoustic Complexity Index (ACI) were computed for two shallow seamounts (Formigas and Princesa Alice, c. 35 m) and one deep seamount (Condor, 190 m) using graphics with different time spans. Only in Formigas, which presented the highest occurrence of fish sounds, was it possible to observe temporal patterns of fish vocal activity in the graphical representations. We highlight that habitats with a higher diversity and abundance of sounds are the most suitable targets for these methods, while in locations with a low prevalence of fish sounds a combination of several methods would be recommended.

Characterization of the acoustic community of vocal fishes in the Azores

Sounds produced by teleost fishes are an important component of marine soundscapes, making passive acoustic monitoring (PAM) an effective way to map the presence of vocal fishes with a minimal impact on ecosystems. Based on a literature review, we list the known soniferous fish species occurring in Azorean waters and compile their sounds. We also describe new fish sounds recorded in Azores seamounts. From the literature, we identified 20 vocal fish species present in Azores. We analysed long-term acoustic recordings carried out since 2008 in Condor and Princesa Alice seamounts and describe 20 new putative fish sound sequences. Although we propose candidates as the source of some vocalizations, this study puts into evidence the myriad of fish sounds lacking species identification. In addition to identifying new sound sequences, we provide the first marine fish sound library for Azores. Our acoustic library will allow to monitor soniferous fish species for conservation and management purposes.

Seasonal Variation of Captive Meagre Acoustic Signalling: A Manual and Automatic Recognition Approach

Many species rely on acoustic communication to fulfil several functions such as advertisement and mediation of social interactions (e.g., agonistic, mating). Therefore, fish calls can be an important source of information, e.g., to recognize reproductive periods or to assess fish welfare, and should be considered a potential non-intrusive tool in aquaculture management. Assessing fish acoustic activity, however, often requires long sound recordings. To analyse these long recordings automatic methods are invaluable tools to detect and extract the relevant biological information. Here we present a study to characterize meagre (Argyrosomus regius) acoustic activity during social contexts in captivity using an automatic pattern-recognition methodology based on the Hidden Markov Model. Calls produced by meagre during the breading season showed a richer repertoire than previously reported. Besides the dense choruses composed by grunts already known for this species, meagre emitted successive series of isolated pulses, audible as ‘knocks’. Grunts with a variable number of pulses were also registered. The overall acoustic activity was concurrent with the number of spawning events. A diel call rhythms exhibit peak of calling activity from 15:00 to midnight. In addition, grunt acoustic parameters varied significantly along the reproduction season. These results open the possibility to use the meagre vocal activity to predict breeding and approaching spawning periods in aquaculture management.

Automatic detection, classification, and quantification of sciaenid fish calls in an estuarine soundscape in the Southeast United States

In the Southeast USA, major contributors to estuarine soundscapes are the courtship calls produced by fish species belonging to the family Sciaenidae. Long-term monitoring of sciaenid courtship sounds may be valuable in understanding reproductive phenology, but this approach produces massive acoustic datasets. With this in mind, we designed a feature-based, signal detector for sciaenid fish calls and tested the efficacy of this detector against manually reviewed data. Acoustic recorders were deployed to collect sound samples for 2 min every 20 min at four stations in the May River estuary, South Carolina, USA from February to November, 2014. Manual analysis of acoustic files revealed that four fish species, belonging to the family Sciaenidae, were the major sound producers in this estuarine soundscape, and included black drum (Pogonias cromis), silver perch (Bairdiella chrysoura), spotted seatrout (Cynoscion nebulosus), and red drum (Sciaenops ocellatus). Recorded calls served as an acoustic library of signature features that were used to create a signal detector to automatically detect, classify, and quantify the number of calls in each acoustic file. Correlation between manual and automatic detection was significant and precision varied from 61% to 100%. Automatic detection provided quantitative data on calling rates for this long-term data set. Positive temperature anomalies increased calling rates of black drum, silver perch, and spotted seatrout, while negative anomalies increased calling rates of red drum. Acoustic monitoring combined with automatic detection could be an additional or alternative method for monitoring sciaenid spawning and changes in phenology associated with climate change.

Automatic detection of fish sounds based on multi-stage classification including logistic regression via adaptive feature weighting

This paper presents a method for automatic detection of fish sounds in an underwater environment. There exist two difficulties: (i) features and classifiers that provide good detection results differ depending on the underwater environment and (ii) there are cases where a large amount of training data that is necessary for supervised machine learning cannot be prepared. A method presented in this paper (the proposed hybrid method) overcomes these difficulties as follows. First, novel logistic regression (NLR) is derived via adaptive feature weighting by focusing on the accuracy of classification results by multiple classifiers, support vector machine (SVM), and k-nearest neighbors (k-NN). Although there are cases where SVM or k-NN cannot work well due to divergence of useful features, NLR can produce complementary results. Second, the proposed hybrid method performs multi-stage classification with consideration of the accuracy of SVM, k-NN, and NLR. The multi-stage acquisition of reliable results works adaptively according to the underwater environment to reduce performance degradation due to diversity of useful classifiers even if abundant training data cannot be prepared. Experiments on underwater recordings including sounds of Sciaenidae such as silver croakers (Pennahia argentata) and blue drums (Nibea mitsukurii) show the effectiveness of the proposed hybrid method.

Acoustic Complexity of vocal fish communities: a field and controlled validation

The Acoustic Complexity Index (ACI) is increasingly applied to the study of biodiversity in aquatic habitats. However, it remains unknown which types of acoustic information are highlighted by this index in underwater environments. This study explored the robustness of the ACI to fine variations in fish sound abundance (i.e. number of sounds) and sound diversity (i.e. number of sound types) in field recordings and controlled experiments. The ACI was found to be sensitive to variations in both sound abundance and sound diversity, making it difficult to discern between these variables. Furthermore, the ACI was strongly dependent on the settings used for its calculation (i.e. frequency and temporal resolution of the ACI algorithm, amplitude filter). Care should thus be taken when comparing ACI absolute values between studies, or between sites with site-specific characteristics (e.g. species diversity, fish vocal community composition). As the use of ecoacoustic indices presents a promising tool for the monitoring of vulnerable environments, methodological validations like those presented in this paper are of paramount importance in understanding which biologically important information can be gathered by applying acoustic indices to Passive Acoustic Monitoring data.

Automatic fish sounds classification

The work presented in this paper focuses on the use of acoustic systems for passive acoustic monitoring of ocean vitality for fish populations. Specifically, it focuses on the use of acoustic systems for passive acoustic monitoring of ocean vitality for fish populations. To this end, various indicators can be used to monitor marine areas such as both the geographical and temporal evolution of fish populations. A discriminative model is built using supervised machine learning (random-forest and support-vector machines). Each acquisition is represented in a feature space, in which the patterns belonging to different semantic classes are as separable as possible. The set of features proposed for describing the acquisitions come from an extensive state of the art in various domains in which classification of acoustic signals is performed, including speech, music, and environmental acoustics. Furthermore, this study proposes to extract features from three representations of the data (time, frequency, and cepstral domains). The proposed classification scheme is tested on real fish sounds recorded on several areas, and achieves 96.9% correct classification compared to 72.5% when using reference state of the art features as descriptors. The classification scheme is also validated on continuous underwater recordings, thereby illustrating that it can be used to both detect and classify fish sounds in operational scenarios.