Proceedings of Meetings on Acoustics

Size: px

Start display at page:

Download "Proceedings of Meetings on Acoustics"

Gwen Collins
6 years ago
Views:

Proceedings of Meetings on Acoustics Volume 19, 2013 http://acousticalsociety.

1 Proceedings of Meetings on Acoustics Volume 19, ICA 2013 Montreal Montreal, Canada 2-7 June 2013 Animal Bioacoustics Session 2aAB: Conditioning, Segmentation, and Feature Extraction in Bioacoustic Signals 2aAB8. Complexity index and proportional variability to study dolphin whistles Carmen Bazúa Durán*, E. Julieta Sarmiento Ponce, Brenda P. González Leal and Camila Rodríguez Bohorquez *Corresponding author's address: Departamento de Física, Facultad de Ciencias, UNAM, Circuito exterior s/n, México, 04510, D.F., Mexico, Dolphin whistles are emitted especially during social interactions and feeding activities involving group cohesion, individual recognition and recruitment. This paper presents a new methodology to describe and compare the whistle repertoire of dolphins. It consists on first extracting the whistle contour using Matlab BELUGA, then classifying whistles into whistle types using Matlab ARTwarp, next classifying whistle types into four general categories (high complexity, low complexity, linear long, and linear short), and finally computing a complexity index and a proportional variability of the whistle repertoire. The method was tested with whistles from captive and wild bottlenose dolphins, Tursiops truncatus, and from wild Guiana dolphins, Sotalia guianensis. Results obtained showed that this very simple method is useful to describe the whistle repertoire and to compare it according to the activity of dolphins, and between species. It is necessary to implement new methodologies like this one to better understand how dolphins are using whistles, since acoustic communication is the most important sense in dolphin species. This is specially important in areas where dolphins are exposed to humans, and where underwater visibility is limited, like Laguna de Términos, a Marine Protected Area in Mexico [work supported by PAPIIT-UNAM]. Published by the Acoustical Society of America through the American Institute of Physics 2013 Acoustical Society of America [DOI: / ] Received 23 Jan 2013; published 2 Jun 2013 Proceedings of Meetings on Acoustics, Vol. 19, (2013) Page 1

2 Dolphin whistles are emitted especially during social interactions and feeding activities involving group cohesion, individual recognition and recruitment (Janik and Slater, 1998; Sayigh et al., 2007). They are frequency modulated sounds almost omnidirectional (Lammers and Au, 2003), with a duration between 0.1 and 3.6 s and a fundamental frequency between 1 and 35 khz (Bazúa Durán, 2010), and have been the primary focus of dolphin research to understand how dolphins communicate. This paper presents a new methodology to describe and compare the whistle repertoire of dolphins. Current whistle analysis methods disagree in the way whistles are classified (e.g., Buck and Tyack, 1993; Janik, 1999; Datta and Sturtivant, 2002; Deecke and Janik, 2006), thus results are not always comparable (Janik, 1999; Bazúa-Durán and Au, 2002), and only in some occasions results have biological significance (Janik, 1999; López Rivas and Bazúa-Durán, 2010). Therefore, researchers need to develop analysis methods that help in avoiding such bias. METHODOLOGY In this paper we studied the dolphin whistle repertoire by first extracting the whistle contour using Matlab software version 6.5 subroutine named BELUGA (Deecke and Janik, 2006), then classifying whistles into whistle types using Matlab software version 7 subroutine named ARTwarp, a neural network that includes dynamic timewarping and considering 90% of similarity (Deecke and Janik, 2006). A proportional variability of the whistle repertoire (PV) was first calculated by dividing the number of whistle types by the total number of whistles and was expressed as a percentage (Sarmiento Ponce, 2011). Next, whistle types were reclassifyed into four general categories (high modulation, low modulation, linear long, and linear short, Table I) and a weight was assigned to each general category to compute a complexity index of the whistle repertoire (CI), as follows (Sarmiento Ponce, 2011): Both parameters represent a fraction and have values between cero and one (or between 0 and 100 for the proportional variability). The closer the proportional variability is from 100, the more variable the whistle repertoire is and the less each whistle type was used. The closer the complexity index is from one, the more complex the whistle repertoire is. We chose to measure these two new parameters because previous studies that used a classification of whistles into six categories (i.e., upsweep, downsweep, concave, convex, sine, and constant; Bazúa Durán and Au 2002, Bazúa Durán 2004, Cook et al. 2004, Azevedo and Van Sluys 2005) are biased because results depend on the researcher, categories are discrete, and whistles are a continuum (Bazúa Durán and Au 2002). Since ARTwarp could not accomplish whistle categorization after 100 iterations (Deecke and Janik, 2006), the data set was partitioned into two or more parts according to the whistle contour to obtain the whistle repertoire. Cross validation of whistle categorization was made to avoid duplicating whistle type categories. Then, the whistle repertoire was formed by the ARTwarp output of all categorizations made after eliminating duplicated categories. This method was tested with whistles from captive bottlenose dolphins, Tursiops truncatus, housed in two aquaria in Mexico City (Sarmiento Ponce, 2011), from wild bottlenose dolphins in Laguna de Términos, a Marine Protected Area at the southern Gulf of Mexico (Bazúa-Durán and Delgado-Estrella, 2009), and from wild Guiana dolphins, Sotalia guianensis, in Bahía Cispatá at the southern part of Golfo de Morrosquillo, Colombia (Caballero et al., 2007; Bazúa Durán et al., 2009). Captive bottlenose dolphin s age was also considered for analysis as adults older than 13 years in aquarium 1, and adults older than 7 years in aquarium 2. Wild bottlenose dolphin s activity was considered for analysis as feeding, traveling, socializing and resting. Acoustic recordings were made using omnidirectional transducers model 8200 (-198±1 db re 1 V/μPa from 3 Hz to 100 khz) connected to a digital M-Audio Microtrack recorder model 24/96 (sampling at 96 khz with 16 bits), a Sony Digital Audio Tape (DAT) recorder model TCD-D100 (sampling at 48 khz with 16 bits), or a TASCAM recorder model HD-P2 (sampling at 96kHz with 24 bits), for captive bottlenose dolphins, wild bottlenose dolphins, and Guiana dolphins, respectively. Captive bottlenose dolphins were recorded in 2007, wild bottlenose dolphins in 2004, and wild Guiana dolphins in Recordings were analyzed to create files of at least 1.5 s in duration that contained a single whistle. We used 1.5 s, 24 or 48 khz windows to analyze recordings. Whistles were considered as a single entity following Bazúa-Durán and Au (2002). Proceedings of Meetings on Acoustics, Vol. 19, (2013) Page 2

3 TABLE I. Examples of whistle contours that comprised each of the four general whistle categories (high modulation, low modulation, linear long, linear short) used to calculate the complexity index of the whistle repertoire of dolphins. High modulation Less modulation Long Short 4* 2* 3* 1* * These numbers were used to calculate the complexity index of the whistle repertoire RESULTS The data set consisted of 724 whistles for captive bottlenose dolphins, 1113 whistles for wild bottlenose dolphins, and 260 whistles for wild Guiana dolphins. Results show that no differences were found in the structure and use of the whistle repertoire by captive and wild bottlenose dolphins (Table II). Bottlenose dolphin whistles were moderately complex, with a relatively large whistle repertoire, which they used quite extensively. When considering captive bottlenose dolphin s age, adult bottlenose dolphins in aquarium 1 had a smaller repertoire of less variable whistles (38 types, CI=0.57, PV=13.7%) than younger dolphins in aquarium 2 (50 types, CI=0.58, PV=11.2%), and used it more frequently (Sarmiento Ponce, 2011). When considering wild bottlenose dolphin s activity (e.g., feeding, travel, socializing and resting), feeding (26 types, CI=0.47, PV=5.9%) and socializing (21 types, CI=0.62, PV=7.0%) dolphins emitted the largest number of whistle types, but feeding whistles were less complex than social whistles; traveling dolphins emitted the least complex whistles (18 types, CI=0.40, PV=7.3%), while resting dolphins emitted the fewer number of whistle types, but the more complex ones (6 types, CI=0.63, PV=12.0%) (González Leal in progress). Guiana dolphins were very short, thus all whistles of less than 0.21 s were not classified using ARTwarp and were considered as a single category of whistle type (61% of all 260 whistles), that was included as part of the general category of linear short whistles to compute the complexity index (Rodríguez Bohorquez in progress). We chose 0.21 s because it is the duration mean of the 260 Guiana dolphin whistles studied. Guiana dolphin whistles were, then, simple, with a less diverse repertoire that was used frequently (Table II). TABLE II. Number of whistle types, proportional variability and complexity index of the whistle repertoire of dolphins. captive bottlenose dolphins wild bottlenose dolphins Guiana dolphins number of whistle types proportional variability, PV (%) complexity index, CI number of whistles, N Proceedings of Meetings on Acoustics, Vol. 19, (2013) Page 3

4 DISCUSSION AND CONCLUSIONS Results obtained showed that this very simple method based on computing the complexity index and proportional variability of the whistle repertoire is useful to describe the whistle repertoire of dolphins and to compare it between species, and according to the activity of dolphins. The Guiana dolphin whistle repertoire was less complex than the bottlenose dolphin whistle repertoire. Both species are coastal, feed during the day, and live in fission-fussion societies, but have differences in their population structure characteristics, thus different social structure and life histories (Mann et al., 2000; Wells and Scott, 2002; García and Trujillo, 2004). Our results, as Azevedo and Van Sluys (2005) and May-Collado and Watzok (2009) studies that described the whistle repertoire using traditional whistle measures, support the hypothesis of Bazúa Durán (2004), also based on traditional whistle measures, that the whistle repertoire of Guiana dolphins may be similar in complexity than the spinner dolphin, Stenella longirostris, whistle repertoire due to population structure similarities (see also May-Collado et al., 2007), and being very different from the bottlenose dolphin whistle repertoire. It is necessary to implement new methodologies like this one to better understand how dolphins are using whistles, since acoustic communication is the most important sense in dolphin species. This is specially important in areas where dolphins are exposed to humans, and where underwater visibility is limited, like Laguna de Términos, a Marine Protected Area in Mexico. ACKNOWLEDGEMENTS This work was supported by projects PAPIIT-UNAM IN and IN granted to CBD. We thank the staff at SIX FLAGS MÉXICO and CONVIMAR for letting us record their dolphins, Ricardo Dorantes and Volker Deecke for their great technical support, and Fondo Mixto CONACyT-Gobierno del Estado de Campeche, Rebeca López, and Salomé Dussán, among others, for data collection. REFERENCES Azevedo, A. F. and Van Sluys, M. (2005). Whistles of tucuxi dolphins Sotalia fluviatilis in Brazil: Comparisons among populations, J. Acoust. Soc. Am. 117, Bazúa Durán, C. (2004). Differences in the whistles characteristics and repertoire of the bottlenose and spinner dolphins, An. Acad. Brasil. Cienc. 76, Bazúa Durán, C. (2010). Sonidos en el mar: el delfín y el camarón tronador, Ciencia 61, Bazúa Durán, C. and Au, W.W.L. (2002). The whistles of Hawaiian spinner dolphins, J. Acoust. Soc. Am. 112, Bazúa-Durán, C. and A. Delgado-Estrella. (2009). Los tursiones (Tursiops truncatus) de la Laguna de Términos, Campeche, in: Memorias del Primer Simposium para el Conocimiento de los Recursos Costeros del Sureste de México, edited by L.E. Amador del Ángel, E. Guevara Carrió, X. Chiappa Carrara, R.Brito Pérez, and R Gelabert Fernández (Universidad Autónoma del Carmen, México), pp Bazúa Durán, C., Vergara, V. and Dussán Duque, S. (2009). We are not silent! Sotalia guianensis calls are mainly ultrasonic, 18 th Biennial Conference on the Biology of Marine Mammals. Québec, Q., Canadá. Oct , P. 28. Buck, J.R. and Tyack, P.L. (1993). A quantitative measure of similarity for Tursiops truncatus signature whistles, J. Acoust. Soc. Am. 94, Caballero, S., Trujillo, F., Vianna, J.A., Barrios-Garrido, H., Montiel, M.G., Beltrán-Pedreros, S., Marmontel, M., Santos, M.C., Rossi-Santos, M., Santos, F.R., and Baker, C.S. (2007). Taxonomic status of the genus Sotalia: Species level ranking for Tucuxi (Sotalia fluviatilis) and Costero (Sotalia guianensis) dolphins, Marine Mammal Sci. 23, Cook, M.L.H., Sayigh, L.S., Blum, J.E., and Wells, R.S. (2004). Signature-whistle production in undisturbed free-ranging bottlenose dolphins (Tursiops truncatus), Proc. R. Soc. Lond. B. 271, Datta, S. and Sturtivant, C. (2002). Dolphin whistle classification for determining group identities, Signal Processing. 82, Deecke, V.B. and Janik, V.M. (2006). Automated categorization of bioacoustic signals: Avoiding perceptual pitfalls, J. Acoust. Soc. Am. 119, García, C. and Trujillo, F. (2004). Preliminary observations on habitat use patterns of the marine tucuxi, Sotalia fluviatilis, in Cispatá Bay, Colombian Caribbean coast, LAJAM 31, Janik, V.M. (1999). Pitfalls in the categorization of behaviour: a comparison of dolphin whistle classification methods, Anim. Behav. 57, Janik, V.M. and Slater, P.J.B. (1998). Context-specific use suggests that bottlenose dolphin signature whistles are cohesion calls, Anim. Behav. 56, Lammers, M.O. and Au, W.W.L. (2003). Directionality in the whistles of Hawaiian spinner dolphins (Stenella longirostris): A signal feature to cue direction of movement?, Mar. Mammal Sci. 19, Proceedings of Meetings on Acoustics, Vol. 19, (2013) Page 4

5 López-Rivas, R.M. and Bazúa-Durán, C. (2010). Who is whistling? Localizing and identifying phonating dolphins in captivity, Appl. Acoust. 71, Mann, J., Connor, R.C., Tyack, P.L., and Whitehead, H. (2000). Cetacean Societies: Field Studies of Dolphins and Whales (University of Chicago Press, Chicago). May-Collado, L.J., Agnarsson, I., and Wartzok, D. (2007). Phylogenetic review of tonal sound production in whales in relation to sociality, BMC Evolutionary Biology, 7, May-Collado, L.J. and Wartzok, D. (2009). A characterizations of Guyana dolphin (Sotalia guianensis) whistles from Cost Rica: The importance of broadband recording systems, J. Acoust. Soc. Am., 125, Sarmiento Ponce, E.J. (2011). Comportamiento acústico de tursiones (Tursiops truncatus) en cautiverio: Comparación entre dos acuarios, Professional thesis, Facultad de Ciencias, Universidad Nacional Autónoma de México, Mexico. 138 pp. Sayigh, L.S., Esch, H.C., Wells, R.S., and Janik, V.M. (2007). Facts about signature whistles of bottlenose dolphins, Tursiops truncatus, Anim. Behav. 74, Wells, R.S. and Scott, M.D. (2002). Bottlenose Dolphins, in: The Encyclopedia of Marine Mammals, edited by W.F. Perrin, B. Würsig and J.G.M. Thewissen (Academic Press, New York), pp Proceedings of Meetings on Acoustics, Vol. 19, (2013) Page 5

Increased Number of Whistles of Bottlenose Dolphins, Tursiops truncatus, Arising from Interaction with People

FULL PAPER Ethology Increased Number of Whistles of Bottlenose Dolphins, Tursiops truncatus, Arising from Interaction with People Junko AKIYAMA 1) and Mitsuaki OHTA 1) 1) Laboratory of Animal and Human