Abstract Frequency modulation (FM) is believed to play a major role in encoding information in tonal vocal communication. However, most studies aimed at investigating acoustic variability rely on the manual measurement of acoustic parameters, and the implementation of multivariate techniques, which represent only partially sound FM. Thus, innovative modelling approaches able to capture local FM dynamics are needed for tonal sounds modelling. These kinds of methodological developments in bioacoustics might be key for further understanding animal behaviour and ecology. We propose the application of a functional data analysis (FDA) approach to model extracted FM patterns, which entails the transformation of acoustic signals into continuous functions. We describe a Raven-to-R FDA workflow for modelling tonal sounds and we highlight two of its potential applications for classification aims and FM analysis in relation to behavioural, social, and environmental factors. For illustrative purposes, the approach performance was tested on FM patterns (contours) of signature whistles emitted by bottlenose dolphins (Tursiops truncatus), to investigate their acoustic variability and potential information content. Our results show that tonal sounds are inherently of functional nature and can be treated as such for FM analysis. We found that building the FDA sound curves using 25 B-spline basis functions was the most appropriate setting for the functionalisation of dolphin whistles. Our approach was able to accurately reconstruct the FM patterns of the modelled tonal sounds and even to capture their fine-scale modulations, with a precision of 80\%. Functional clustering showed an accuracy up to 89\% for short whistles. Estimated functional coefficients of the regression model suggested that different contextual factors have a significant effect on the FM patterns of SWs. The adoption of an FDA approach appears to be quite promising for the study of frequency-modulated tonal sounds, both for classification aims and in-depth FM analysis. The limitations and advantages of the approach were discussed. We developed and tested a new methodology able to fill knowledge gaps in animal communication, and we anticipate it will lead to advances in the ethological and ecological understanding of wild animal populations.

A functional data analysis approach for modelling frequency-modulated tonal sounds in animal communication / Silvia Labriola, Maria; Pammer, PETRA OSWINE; Pedrazzi, Giulia; Giacomini, Giancarlo; JONA LASINIO, Giovanna; Pace, DANIELA SILVIA. - In: METHODS IN ECOLOGY AND EVOLUTION. - ISSN 2041-210X. - n/a:n/a(2025), pp. 1-15. [10.1111/2041-210X.14446]

A functional data analysis approach for modelling frequency-modulated tonal sounds in animal communication

Petra Oswine Pammer
Membro del Collaboration Group
;
Giulia Pedrazzi
Membro del Collaboration Group
;
Giancarlo Giacomini
Membro del Collaboration Group
;
Giovanna Jona Lasinio
Penultimo
Methodology
;
Daniela Silvia Pace
Ultimo
Conceptualization
2025

Abstract

Abstract Frequency modulation (FM) is believed to play a major role in encoding information in tonal vocal communication. However, most studies aimed at investigating acoustic variability rely on the manual measurement of acoustic parameters, and the implementation of multivariate techniques, which represent only partially sound FM. Thus, innovative modelling approaches able to capture local FM dynamics are needed for tonal sounds modelling. These kinds of methodological developments in bioacoustics might be key for further understanding animal behaviour and ecology. We propose the application of a functional data analysis (FDA) approach to model extracted FM patterns, which entails the transformation of acoustic signals into continuous functions. We describe a Raven-to-R FDA workflow for modelling tonal sounds and we highlight two of its potential applications for classification aims and FM analysis in relation to behavioural, social, and environmental factors. For illustrative purposes, the approach performance was tested on FM patterns (contours) of signature whistles emitted by bottlenose dolphins (Tursiops truncatus), to investigate their acoustic variability and potential information content. Our results show that tonal sounds are inherently of functional nature and can be treated as such for FM analysis. We found that building the FDA sound curves using 25 B-spline basis functions was the most appropriate setting for the functionalisation of dolphin whistles. Our approach was able to accurately reconstruct the FM patterns of the modelled tonal sounds and even to capture their fine-scale modulations, with a precision of 80\%. Functional clustering showed an accuracy up to 89\% for short whistles. Estimated functional coefficients of the regression model suggested that different contextual factors have a significant effect on the FM patterns of SWs. The adoption of an FDA approach appears to be quite promising for the study of frequency-modulated tonal sounds, both for classification aims and in-depth FM analysis. The limitations and advantages of the approach were discussed. We developed and tested a new methodology able to fill knowledge gaps in animal communication, and we anticipate it will lead to advances in the ethological and ecological understanding of wild animal populations.
2025
animal communication; bioacoustics; bottlenose dolphin; FDA; FM; signals; signature whistles; tonal sounds
01 Pubblicazione su rivista::01a Articolo in rivista
A functional data analysis approach for modelling frequency-modulated tonal sounds in animal communication / Silvia Labriola, Maria; Pammer, PETRA OSWINE; Pedrazzi, Giulia; Giacomini, Giancarlo; JONA LASINIO, Giovanna; Pace, DANIELA SILVIA. - In: METHODS IN ECOLOGY AND EVOLUTION. - ISSN 2041-210X. - n/a:n/a(2025), pp. 1-15. [10.1111/2041-210X.14446]
File allegati a questo prodotto
File Dimensione Formato  
Labriola -A-functional-data_2025.pdf

accesso aperto

Note: early view version
Tipologia: Versione editoriale (versione pubblicata con il layout dell'editore)
Licenza: Creative commons
Dimensione 2.22 MB
Formato Adobe PDF
2.22 MB Adobe PDF

I documenti in IRIS sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.

Utilizza questo identificativo per citare o creare un link a questo documento: https://hdl.handle.net/11573/1731616
Citazioni
  • ???jsp.display-item.citation.pmc??? ND
  • Scopus ND
  • ???jsp.display-item.citation.isi??? ND
social impact