From the lab to the field: integrating bioacoustics and eco acoustics

Both small- and large-scale work currently drives bioacoustic research: taxonomists may use sounds to identify or describe a single species, while acoustic observatories collect large volumes of data that require automated analysis. In this symposium, we will review the state-of-the-art and provide suggestions for collaboration between researchers working at opposite ends of the scale.

Topics covered will include data collection from observatories, how that data can be analysed (including annotation and identification) and sustainably archived for future researchers. We are fortunate to be experiencing a surge in the rate that acoustic data are collected, which presents both opportunities and challenges. The symposium, therefore, aims to bridge all stages of the data life cycle from collection, through analysis to publication. In particular, we address the problems of dataset annotation and how annotations are shared.

A core topic that symposium speakers will cover is achieving future-proof research. Automated species identification, particularly for less well-studied taxa, is often performed against an ever-changing taxonomy as researchers discover new species and redefine species concepts. A genuinely automatic observatory must, therefore, be adaptable and training sets must be regularly updated to reflect new scientific understandings and bridge the fields of bioacoustics and ecoacoustics.

Symposium details

Held at the XXVI International Bioacoustics Congress - 31th aug. to 5th sep. 2019

Chairmen: Dr. Ed Baker, Dr. Klaus Riede (Zoological Research Museum Alexander Koenig), Mr. Tomás H. Rostirolla (Arquivo Bioacústico Catarinense UFSC)

Visit the IBAC site

Oral speakers

Recognising the Wolf Howls

Mr. Sougata Sadhukhan

Wildlife Institute of India

Bilal Habib

Study of population ecology is imperative in wildlife conservation studies. The conservation status of any species is dependent on its population size, the smaller the size of the population, the more vulnerable it is to extinction. The population size and distribution of endangered Indian wolves (Canis lupus pallipes) are entirely unknown due to lack of standardised population estimation methods for this cryptic long-ranged species. Mark-Capture-Recapture based population estimation method falls short in estimating the population of wolf as they lack identifiable coat patterns. Our Acoustics Based Survey method offers a potential method for detecting vocal animals like wolves from as far as 3km away.
To standardise individual identification from wolf howls, we collected data from three captive wolves as known samples and howls of 9 pack of free-ranging wolves. Twelve parameters were measured from the fundamental frequency of each howls. DFA showed 100% accuracy in identifying 44 known howls from 8 individuals. The same function (DF1 and DF2) was applied over 52 howls from 8 known individual and classified using Centroid Clustering with Minkowski 2nd order function. 64.7% howls were identified correctly. The primary aim of this study was to check the feasibility of wolf pack census through howling. We are still working to improve the accuracy and reliability of this protocol. Estimating the wolf population can be achieved scientifically by further improvement of this technique and systematic sampling, which in turn will help us to get a proper status of wolves in India and continuous population monitoring.

sougata.mail@gmail.com

Assessing biodiversity with sound: Do acoustic diversity indices reflect diversities of bird communities in grassland habitats?

Ms. Pooja Panwar

University od Arkansas- Fayetteville

J.D. Willson (University of Arkansas- Fayetteville)

Measuring biodiversity is a challenging task for research in ecology and conservation. Acoustic indices calculated from soundscape has been suggested as a solution for biodiversity assessment. However, the relationship between on-the-ground truth and acoustic indices still needs to be investigated. This study applied the emerging tools of ecoacoustics to examine the avian diversity of the Prairies in the US. In this study, we investigated the suitability of five acoustic indices: Entropy, Acoustic Complexity Index (ACI), Acoustic Diversity Index (ADI), Biodiversity Index (BIO) and Normalized Difference Soundscape Index (NDSI) to measure the avian richness of Prairies at different level of degradation. Twenty-one 24-hour long recordings were made at nine sites of remnant, restored, and degraded prairies each in northwest Arkansas (US) between May-July 2018.
Using “soundecology” package in R, the five acoustic indices were calculated for randomly selected one-minute recordings. A linear mixed-effect model highlighted a significant correlation between the ACI values and the number of bird vocalizations composing the soundscape. A comparison of ACI value for prairies with different degradation scale, showed that remnant and restored prairies had significantly higher ACI values than degraded grasslands. This was in agreement with avian richness estimated using traditional point count surveys. Acoustic indices offer new opportunities to monitor grassland bird communities faced with the challenge of human-induced disturbances and other proxies like climate and land use changes.

ppanwar@uark.edu

Tracking Woodpeckers in the wild in the age of deep neural networks

Dr. Juliette

University of Mons

Olivier Verlinden (University of Mons)

Of the eleven European species of woodpeckers, ten drum and seven use long-distance advertising calls. These signals are all species-specific and easy to spot in forest soundscapes for a trained ear. For a few years now, we have studied the automation of this process. Recording sound continuously is not an issue beyond battery management considerations. The actual challenge is to handle the data deluge. Detecting signals is a multi-stage process; first the most obvious negatives are eliminated using the ACI, then the remaining audio is segmented, and finally the segments are assessed. For this last stage and for the identification of the species, very deep neural networks provide the analysis skills and the ease of handling large batches of data. This is made possible by 1) the accessibility of legacy deep image nets that can be retrained at will for specific image problems and 2) the transformation of sound problems into image problems via spectrograms. Either for detection or species identification, the accuracy of deep nets on a Xeno-Canto-based test set easily reaches 95%. The performance on real-life datasets remains impressive but is hindered notably by the wealth of biodiversity that could not be taken into account during training. The way image invariants do not apply to spectrograms is another concern. Nevertheless, deep neural networks make the analysis of large datasets a reality.

juliette.flontin@umons.ac.be

Minding miners: Using bioacoustic indices to remotely monitor both the presence and behaviour of a species that causes avifaunal

Dr. Paul McDonald

University of New England

Ahmad Barati (University of New England)

Bioacoustic indices have proven very successful in monitoring community level changes in soundscapes across a range of habitats and scales. However, their potential use as a tool for estimating the occurrence of the vocalisations of specific species is still poorly understood. We examined the ability of indices to separate areas that were either inhabited by, or freefrom, the Noisy Miner (Manorina melanocephala). This species is ideally suited for indexbased monitoring, as it is highly vociferous with a diverse, functionally referential acoustic repertoire. Monitoring miner presence is also of key importance as the species has negative conservation impacts on avifaunal diversity due to its extreme territoriality. Bioacoustic indices did differ between sites occupied by Noisy Miners and those where the species was absent, with indices also changing significantly at sites following Noisy Miner colonisation. Beyond simple presence/absence data, some indices also correlate with Noisy Miner density at different sites. Finally, through targeted parameterisation of acoustic index settings, we were able to demonstrate their potential for quantifying the relative abundance of specific Noisy Miner vocalisation types, and thus their associated behaviours, within sites. Together, these data demonstrate the enormous potential of acoustic indices to rapidly provide information of use at both the soundscape and focal species level that is highly adaptable to other taxa.

paul.mcdonald@une.edu.au

Long-term acoustic recordings: a solution for the rock ptarmigan?

Mr. Thibaut Marin-Cudraz

Equipe de Neuro-Ethologie Sensorielle, University of Lyon/Saint-Etienne

The rock ptarmigan (Lagopus muta) is a bird living in high altitudes in France and considered as a sentinel species of the alpine environment. The ever changing weather of mountains and the low call rate of the bird makes it a challenge to monitor. The current monitoring of the species is based on localizations of males during the mating season by acoustic cues of males during point count protocol. The inconsistency between counting days and localizations problems due to sound propagation in alpine environments question the protocol validation. According to these problems, long-term acoustic monitoring might be an appealing solution.
Indeed, numerous studies showed that acoustic signals convey individual information about the emitters. Bioacousticians can exploit these informations obtained through long period of time to estimate the number of individuals in an area and furthermore monitor population dynamics but present two drawbacks: (1) The weather might corrupt the recordings and mask the calls and reduce individuals identification. (2) The acoustic availability of the animals: individuals might not be identified if they don’t vocalize enough. Despite these strong constraints, our study covering three consecutive years of long-term recordings shows that bioacoustics is relevant as a monitoring tool for this species. Moreover, the temporal scale covered by our study allow to obtain finer informations than by point count such as the number and time of presence of individuals in the area of interest. Hence, the precision of the data obtained by acoustic monitoring is dependent of its duration, especially for secretive species.

thibaut.marincudraz@gmail.com

Can My Dinner Hear Me? The Perception of Bats' Echolocation Signals by Orthoptera Insects

Dr. Tomasz Postawa

Institute of Systematics and Evolution of Animals PAS

Anna Marchewka (Zoological Garden in Poznań, Poland), Paulina Turowicz (University of Wrocław, Poland), Szilárd Bücs (Centre for Bat Research & Conservation, Romania)

Predation is one of the strongest factors in natural selection. Under this biological interaction, prey have developed a number of defense mechanisms (passive and active), of which the most common are behavioral. Some predators specialize in eavesdropping the courtship songs of their prey, leading to a tradeoff being developed during the mating period that balances sexual selection and the pressures of predation. Tettigoniids comprise the main part of the diet of the M. blythii, one of the two European species of bat who apply this "eavesdropping" strategy to foraging. In turn, Orthoptera species of insect are sensitive to ultrasound, which induces reactions such as pausing or stopping their mating songs, curtailing flight, and flying away from the source of the sound, actions that are indicative of defense strategies developed against insectivorous bats. We tested the responses of several tree species – Gryllus campestris, Tettigonia cantans and T. viridissima – to echolocation calls of the Myotis blythii (who feed on Orthoptera) and Eptesicus serotinus (who do not feed on Orthoptera), using playback experiments. Research was conducted under natural conditions in regions where both species of bats naturally occur (Poland and Romania). The results found no reaction to the M. blythii feeding sequences (predator), yet reactions from all three Orthoptera species (stopping or pausing vocalization) to the feeding sequences of E. serotinus (not a predator). These results demonstrate that defense strategies to avoid echolocation predators is, to a certain extent, dependent on the range of prey audibility, in this case Orthoptera insects.

tpostawa@gmail.com

Sensitivity of bush-crickets to climate change revealed by large-scale acoustic monitoring

Dr. Yves Bas

CESCO-MNHN and CEFE-CNRS

Jérôme Sueur [ISYEB-MNHN]; Christian Kerbiriou [CESCO-MNHN]; Jean-François Julien [CESCO-MNHN]; Isabelle Le Viol [CESCO-MNHN]

Monitoring biodiversity over large spatial scales is crucial for assessing the impact of
growing climate changes. New technologies such as high frequency autonomous recording
and machine learning for vocalisation classification unleashed the potential of passive
acoustic monitoring schemes intending to track species trends in the long term. Such

monitoring schemes have been developed for bats in several countries recently, and bush-
crickets (or katydids) have the advantage to be acoustically active in the same frequency

range and during the same periods as bats, so species acoustic activity of both groups can be monitored through the very same recordings. French bat monitoring scheme Vigie-Chiro allowed to record bush-cricket activity on 5922 localities during full nights between 2014 and 2018. We used this dataset to track short-term responses of 16 species of bush-crickets to climate change in France. Acoustic data were analysed through a generalised linear mixed model to investigate the interactions between seasonal phenology, seasonal temperature anomalies and species climatic niche (position and width). Seasonal phenologies showed a significant negative interaction with seasonal temperature anomalies, i.e. an average shift of 3 days.°C-1. Results also showed non-linear effects of seasonal temperature anomalies, species with relatively cold climatic niche being negatively affected by anomalies over 1°C whereas species with warm climatic niche took advantage of anomalies over 2°C. These short-term and complex effects demonstrate the value of large-scale acoustic monitoring to study climate change effects on biodiversity, and especially to measure species sensitivity to temperature anomalies, thus highlighting conservation issues.

yves.bas@mnhn.fr

The challenges of automatic counting and identification of weak-flying insects using optoacoustic methods

Dr. James Bell

Rothamsted Research, Harpenden, Herts, UK

Alex Dye (Rothamsted Research); Kirsty L. Hassall (Rothamsted Research); Ilyas Potamitis (TEI Crete)

In the last five years, there has been an exponential interest in real-time detection of insect threats. Researchers have used electronics to explore the possibilities of detection and have focussed efforts on camera traps with image analysis, radar and lidar all with varying degrees of success. Bioacoustic methods, particularly those using acoustic and optical methods are playing an increasingly important role, often complementing image analysis when identifications are uncertain, as is the case for many insect groups. Here, we discuss the technical challenges of accurately identifying aphid pests to species level using optoacoustic methods. Aphids migrate just above crop height, producing a wingbeat signal that is an order of magnitude weaker than mosquitoes, making this group hard to detect. Given that there are more than 600 species in the UK and that these are difficult to differentiate between species even with a microscope, the challenges for automatic detection seem fraught with difficulty. We discuss the technical challenges of accurately identifying aphid pests to species level using opto-acoustic methods. Using fast Fourier transform, a highpass filter and a detrending step to ‘clean’ the signal, we plot the frequency spectra generated by a small number of aphid species to indicate the high variability of wingbeat recordings within and between species. Yet, from these we can estimate to reasonable confidence the fundamental frequency, the dominant frequency, the bioacoustic index and the temporal entropy, all of which show some role in aphid species classification.

james.bell@rothamsted.ac.uk

Acoustic metrics to characterize underwater acoustic diversity in the Southern Ocean

Dr. Irene T. Roca

Helmholtz Institute for Functional Marine Biodiversity (HIFMB), Carl von Ossietzky University, Oldenburg, Germany

Ilse Van Opzeeland [Helmholtz Institute for Functional Marine Biodiversity; Ocean Acoustics Lab, Alfred-Wegener Institute]

Acoustic metrics (AM) assist our interpretation of acoustic environments by aggregating a complex signal into a unique number. Numerous AM have been developed for terrestrial ecosystems, with applications ranging from rapid biodiversity assessments to characterizing spatial acoustic heterogeneity. However, there has been comparatively little research aimed at understanding how these metrics perform to characterize the acoustic features of marine habitats and their relation with ecosystem biodiversity. We first assess how existing AM vary with different natural marine soundscape compositions. Second, we evaluate how AM perform compared to the signal frequency spectrum to discriminate between marine mammal acoustic assemblages. Basin-wide passive acoustic monitoring data sets from the Antarctic Weddell Sea spanning 8 years and 5 sites were used to develop a supervised machine learning approach to discriminate between acoustic environments. AM performed better than spectral signatures, achieving correct classifications of species richness levels (0-5) of 65% for off-shore sites and 90% for near-shore sites. Discrimination of species identity based on AM achieved >85% correct classifications for 7 marine mammal species. While there was not a single AM that accurately classified acoustic community composition for all sites and years, we show that a combination of AM is a promising approach to characterize underwater marine habitats. AM allow an intuitive ecological interpretation of passive acoustic data, which in the light of ongoing environmental changes, supports the more holistic approach needed to detect and understand trends in species diversity and underwater habitat quality.

irene.torrecilla@hifmb.de

Multi-scale ecoacoustic analysis of natural soundscapes within the INR of Sasso Fratino (Italy)

Dr. Roberta Righini

Department of Earth and Environmental Sciences, University of Pavia (Italy)

GIANNI PAVAN (Interdisciplinary Center for Bioacoustics and Environmental Research, University of Pavia, Italy)

Ecoacoustics is an increasingly emerging interdisciplinary science that investigates natural and anthropogenic sounds and their relationship with the environment. But if on the one hand, it allows a great opportunity to monitor a habitat because, by the placement of many sensing devices it is possible to record continuously and simultaneously different locations, on the other hand, it is generated a huge quantity of data that needs to be archived and analyzed.
The new challenge is to draw up a scaled analysis methodology to optimize resources both in the field (saving memory cards and batteries) both in the lab, reducing time and effort of analyses, while maintaining high the accuracy of the information. This work focuses on the testing of scaled multi-view analyses that are both qualitative, based on visual screening of compact daily spectrograms, and quantitative, by the estimation of the acoustics indices and the analysis of spectrum energy for each 1000 Hz range (from 0 to 24 kHz). A further level of analysis is the identification of species by listening to the sounds and observing high-resolution real-time spectrograms, possibly driven by automatic detection tools.
On a five-year archive of acoustic data collected in the Integral Natural Reserve of Sasso Fratino (Italy), we operated cross-section (horizontal) and time series (vertical) analyses to explore its soundscapes’ spatial-temporal dynamics across different scales.

roberta.righini01@universitadipavia.it

An exploration of high dimensional acoustic features applied to a real-time monitoring network in Borneo

Mr. Sarab Sethi

Imperial College London

Lorenzo Picinali (ICL); Nick Jones (ICL); Benjamin Fulcher (University of Sydney); Robert Ewers (ICL)

Automated acoustic monitoring of ecosystems provides a cost-effective way to track changes in natural system’s dynamics across temporal and spatial scales. Current methods of analysing soundscapes in a holistic manner typically reduce chunks of audio data to a single index (or a few indices), broadly based on the entropy of the signal. We find that by adopting a deep convolutional neural net used for general purpose audio classification, we are able to embed the audio in a more descriptive, high dimensional feature space. We apply this technique to a real-time passive acoustic monitoring network based in the tropical forests of Sabah, Borneo. We are able to accurately predict metrics of habitat quality from audio alone using this feature embedding, and show how this is not possible using existing approaches. Additionally, we explore potential applications of this approach to general acoustic anomaly detection.

s.sethi16@imperial.ac.uk

Tackling all acoustic taxa inventories in the neotropics

Mr. Tomás Honaiser Rostirolla

Arquivo Bioacústico Catarinense (UFSC)

Fernando Bittencourt de Farias; Alexandre Paulo Teixeira Moreira (Universidade Federal de Santa Catarina)

South America forests comprises some of the most diverse hotspots of biodiversity in the world. While managing such places require integration from diverse specialists and policy makers, knowledge on species distribution and acoustic behaviour still incipient. We’ll explore the ABC (Arquivo Bioacústico Catarinense/ECZ/UFSC) endeavour of annotating all acoustic taxa in the Serra do Tabuleiro State Park, the most biodiverse region from south Brazil, and the efforts of sharing finds and information to scientific and professional communities. Thought bird and anuran repertoires are fairly known, an overwhelming activity of insects still poorly understood, revealing clear needs of deeper field and acoustic research.

onaiser@gmail.com

Designing an Acoustic Observatory

Phil Eichniski

The field of bioacoustics encompasses many scales, from studies of individual species in laboratory settings to studies of the acoustics of entire ecosystems. While studies at the extremities of this range have the potential to benefit each other, at present this requires a large amount of repetitive work. Using the process of developing the Automated Acoustic Observatories project, combined with the ongoing development of the BioAcoustica database I provide an overview of the challenges and potential benefits of developing an integrated bioacoustics research infrastructure.
Ideally, such a system would allow for the permanent storage, annotation and easy sharing of bioacoustic datasets on a worldwide scale. Challenges in achieving this include the fragmentary nature of repositories and the lack of standard annotation and exchange formats. This talk discusses several potential solutions to these problems. Another concern, particularly in insects, is future-proofing bioacoustic collections against a rapidly developing taxonomy and nomenclature. Linking sound recordings to museum specimens and taxonomic databases has the potential to mitigate this problem.
Solving these problems may allow for a future acoustic observatory that can automatically update itself with new knowledge of recently described species, and potentially be used to discover new ones.

Towards the standardization of ecoacoustic indices for cross-comparison of audio datasets

Mr. Sylvain Haupert

MNHN, CNRS UMR7205 ISYEB, Sorbonne Université

Jérôme SUEUR (MNHN, CNRS UMR7205 ISYEB, Sorbonne Université)

Ecoacoustics offers alternative techniques to direct observations for ecologists and conservation biologists. Ecoacoustics is at the crossroads of ecology, acoustics and computer science, taking advantage of the broadening of affordable and easy-to-use passive acoustic recording devices. The tremendous amount of fast growing data collected all around the world, all year-long makes manual annotation by experts impossible. The common global approach consists in reducing full audio recordings into acoustic indices, that are mathematical functions designed to describe the complexity and the diversity enclosed in each audio fragment by single numbers. However, ecoacoustics is actually facing a lack of standardization that prevents direc t comparison of indices computed from different audio datasets. This severe limitation has a substantial impact on the reliability of acoustic indices to describe acoustic richness and diversity among various habitats. To address this issue, we propose to standardize the computation of acoustic indices by 1) calibrating audio recordings, and 2) taking into account the variation of time/frequency resolution based on a multiscale approach. The standardization workflow has been validated on two audio subsets of long-term surveys in tropical rain forest (dB@Nouragues project, French Guiana, France) and cold mountain forest (dB@Risoux project, Jura, France). Preliminary results show that standardization allows cross-comparison of acoustic indices from different audio data set opening new opportunities in ecoacoustics.

sylvain.haupert@mnhn.fr

From song libraries to corpora: a case study on extracting Orthoptera songs from environmental recordings

Dr. Klaus Riede

Zoological Research Museum Alexander Koenig

Orthoptera are a major component of soundscapes, particularly in tropical and subtropical forests, but also temperate grasslands. However, most acoustic monitoring projects still focus on vertebrates, particularly birds, frogs or bats. Fortunately, many of these recordings are publicly available or are readily shared by the authors, which facilitates re-use for Orthoptera assessments based on song identification. Using ecoacoustic recordings from North- and South America, Malaysia and Greece I demonstrate their usability for Orthoptera monitoring. The stereotyped temporal structure and well-defined frequency ranges (which reach far into ultrasound for many katydid species) facilitate the development of automatic species recognition algorithms. However, rapid development of algorithmic Orthoptera species recognition is hampered by a lack of labeled reference recordings and incomplete song libraries (Riede 2018). The Orthoptera Species File (http://orthoptera.speciesfile.org/) could serve as a reference song repository, providing a carefully curated taxonomic backbone, while sound libraries such as Bio.acousti.ca (http://bio.acousti.ca/) could serve as a repository for large labeled corpora, needed for development of automatized song recognition.
References:
Riede K (2018) Acoustic profiling of Orthoptera: present state and future needs. Journal of Orthoptera Research 27(2): 203-215. https://doi.org/10.3897/jor.27.23700

klausriede@hotmail.com

From bioacoustics to ecoacoustics: integration challenges

Dr. Ed Baker

The Australian Acoustic Observatory will have 400 continuously recording sensors deployed around Australia that will produce over a terabyte of compressed audio each day for five years. We share insights gained through addressing the challenges of designing this system, including site selection, ensuring sensor hardware is robust to harsh conditions for long durations, protocols for deployment and retrieval of recordings, maintaining data provenance, and development of a public repository capable of making the audio discoverable and navigable.

Contact: ed.baker@york.ac.uk

Poster presentations

Effects of ecology and mating system on acoustic communication in bark and ambrosia beetles

Mr Carol L. Bedoya

University of Canterbury - School of Biological Sciences

Ximena Nelson (UC); Michael Hayes (UC); Richard Hofstetter (NAU); Lawrence Kirkendall (UIB); Eckehard Brockerhoff (SCION)

Bark and ambrosia beetles (Coleoptera: Curculionidae, Scolytinae and Platypodinae) are two closely-related subfamilies of weevils with sound production capabilities. These beetles spend most of their life cycle inside plant tissues and have evolved to display a complex variety of mating systems that are associated with specific acoustic communicatory interactions and stridulatory mechanisms. However, even though bark and ambrosia beetles are some of the few taxa with properly-developed acoustic communication in a medium other than air or water, their sound production remains one of the most understudied areas in bioacoustics. Here, we present the most extensive description to date of the types of calls and stridulatory mechanisms in bark and ambrosia beetles. Furthermore, we evaluated the role of environmental and life history features in the variability of their acoustic parameters, and illustrate the importance of acoustic communication in the evolution of mating systems. Sound production in bark and ambrosia beetles can arise in one sex, both sexes, or neither, and the presence of a stridulatory apparatus appears to be associated with the type of mating system. For example, acoustic communication is absent in species that reproduce using inbreeding polygyny. In monogynous species with elytro-abdominal organs, sound production is dominated by males. Both sexes possessing stridulatory capabilities is the norm in Platypodinae, but not in Scolytinae, and females dictate the acoustic interactions in harem polygynous species.

c.l.bedoya6@gmail.com

Patterns of sound production associated with the reproductive activity of dusky grouper, E. marginatus, at its spawning sites

Ms. Elena Desiderà

University of Padova (Padova, Italy); Université Côte d'Azur (Nice, France)

L. Di Iorio (CHORUS); P. Guidetti (Université Côte d'Azur); C. Mazzoldi (University of Padova); A. Navone (Tavolara MPA)

Groupers are fish of key ecological and economic value. The most iconic Mediterranean grouper is the dusky grouper, Epinephelus marginatus. Fishing caused dramatic population declines across the Mediterranean Sea, where the species is listed as endangered. This species is abundant only in effective Marine Protected Areas (MPAs) but knowledge on its reproduction is still limited. Passive acoustic surveys are valuable tools for studying and monitoring reproductive sites of soniferous species. The dusky grouper is known to emit at least two types of sounds, one of which is associated with courtship displays. Nonetheless, no observation of direct co-occurrence between sound production and reproductive behaviours has ever been made in the wild.
In this study, conducted in the “Tavolara-Punta Coda Cavallo” MPA (Sardinia, Italy), we assessed if and how Dusky grouper acoustic activity was linked to its reproductive behaviours, and compared sound production patterns at two spawning sites over one month, during the peak reproductive season. By coupling underwater direct visual observations to simultaneous passive acoustic recordings, we could attribute Dusky grouper sound production to its courtship behaviours. The courtship call was found five-times more often when courtship was observed than when it was not, being significantly linked to the documented courtship displays. Both sites showed similar diel sound production with one site, falling in the MPA no-take and no-access zone, showing a more pronounced acoustic activity. The study consolidates the value of passive acoustics within monitoring programs aiming at protecting this endangered species during a vital phase of its life cycle.

elena.desidera@phd.unipd.it

Phenology of tropical acoustic communities in New Caledonia under invasion of the ant Wasmannia auropunctata

Mr. Valentin Chevalier

IMBE IRD 237 & ISYEB MNHN

Hervé Jourdan, Laure Desutter-Grandcolas, Amandine Gasc

In the context of the current biodiversity crisis, ecoacoustics has gained much interest from academics, stakeholders and managers, as it allows studying biodiversity richness, composition and phenology as well as levels of perturbations to ecosystems. In New-Caledonia, soundscapes were collected in shrublands, paraforests and forests, invaded and not invaded by the exotic ant Wasmannia auropunctata. Analysis of a first dataset collected during the dry season showed that crickets communities – which dominate the soundscape – are affected by W. auropunctata and absent from invaded locations, leaving the tropical soundscapes drastically quieter than non-invaded sites. This is especially true for the night soundscapes of forests, while shrublands and paraforests show milder differences. Raising the interest of park managers in developing acoustic approaches to detect local invasions of W. auropunctata, we pursued our research in 2018 to investigate the effect of season on soundscapes phenology and validate the impact of the invasive species on sound-producing taxa – with a focus on crickets and birds communities – during the humid season. Allying analysis at both the species level (by description of each species of crickets singing in the soundscape recordings) and the community level (by description of animal communities using acoustic diversity indices), we compare composition and phenology of acoustic communities at different seasons and evaluate the potential of each season to reveal the presence of the invasive ant. This work contributes to propose fast, global and interpretable tools for conservation actors.

valentin.chevalier.vch@gmail.com

Recording slugs to improve slug pellets

Ms. Samantha de Silva

Newcastle University

s.m.de-silva2@ncl.ac.uk

Gastropod damage to crop plants has a significant economic impact on agriculture and horticultural industries worldwide, with the grey field slug (Deroceras reticulatum (Müller)) considered as the main mollusc pest in the United Kingdom. The prevailing form of crop protection are pellets containing the active ingredient, metaldehyde. Metaldehyde is known to cause paralysis and ultimately death after threshold amounts are ingested. The paralysing effects were suggested to result in inadequate pellet consumption; greater understanding of the interaction between consumption and the paralysing effects of metaldehyde could reveal an area of potential manipulation to be targeted by novel molluscicide formulations. An audio sensor was used to record individual slugs feeding on a variety of pellet types, including commercially available pellets and novel metaldehyde formulations. A graphical.mlapp application was used to quantify the time each bite was taken; the length of each bite and the total number of bites. There was significant individual variation in number of bites on a non-toxic pellet but this was not observed on metaldehyde formulations. Slugs took significantly fewer bites from metaldehyde pellets than from non-toxic pellets. There was no significant difference in the length of the bites between non-toxic and metaldehyde pellets.

s.m.de-silva2@ncl.ac.uk

Scratchpads developed and conceived by (alphabetical): Ed Baker, Katherine Bouton Alice Heaton Dimitris Koureas, Laurence Livermore, Dave Roberts, Simon Rycroft, Ben Scott, Vince Smith