Repertório de emissões sonoras do golfinho-de-dentes-rugosos (Steno bredanensis) (Cetartiodactyla: Delphinidae) na costa do Estado do Rio de Janeiro

Abstract

Sounds emitted by delphinids are very important in social interactions. However, the relationships between whistles and the social contexts of these emissions were studied in only a few species and pulsed sounds were the focus of few studies. Rough-toothed dolphins (Steno bredanensis) is a species rarely studied and information about their acoustic repertoire is based on opportunistic recordings in only a few places in their distribution. This thesis presents a characterization of whistles and pulsed sounds of S. bredanensis in the Rio de Janeiro State Coast. The first section presents a characterization of whistle repertoire and variations of these in relation to behavior states of feeding and travelling and other group characteristics. Recordings were made with a C54XRS hdrophone (-165 dB re: 1V/µPa; frequency response between 0,009 and 100 kHz) and one um digital recorder with 192 kHz sample rate. Seventeen days of recording were done between 2010 and 2015. Twelve acoustic parameters were extracted from 317 whistles of S. bredanensis (Raven Pro 1.4, FFT 512, overlap 50%). The whistles presented a mean duration of 380±179 ms and generally low values of frequency modulation (mean = 2.1±1.6 kHz). Furthermore, 42.3% of the whistles presented segmented contours. In general, frequency modulations concentrate in a small percentage of the duration of the whistles, with constant parts occupying 58.5% of the whistle duration on average. Whistles emitted during feeding and travelling (N=256) were compared through acoustic parameters. Five parameters were different between whistles emitted in these behaviors (MannWhitney, p<0.05) and four of these were selected in the discriminant analysis to separate behaviors: duration, number of inflexion points, ending frequency and frequency at ¾ of the duration of the whistle. The whistle emission rate did not present correlation with group size (r=0.2; p>0.05) and did not vary between behavior states (Mann-Whitney, p>0.05). Groups with calves presented higher whistle emission rate (0.27±0.22) than groups without (0.07±0.09) (Mann-Whitney, p<0.05). Among the 508 whistles selected for analysis, 61.4% were stereotyped (presented characteristic contours, distinct from the others, and repeated in sequence at least in one recording session). These whistles were classified in 28 categories and four subcategories according to contour form. Seventy-one sequences were identified. These were formed by a minimum of two and a maximum of 12 stereotyped whistles of the same contour form, with a mean value of 3.3±1.6 whistles in each sequence. These whistles were found in 11 of the 17 groups and in all of the behavior states. The second section presents an analysis of pulsed sounds. The detections of these sounds was done in Triton, the inspection of these in Wavesurfer and MATLAB and the analysis of acoustic parameters was done in MATLAB. Forty-four pulsed sounds were considered adequate for analysis. These sounds had a mean number of 317±208 pulses per sound and a mean duration of 317±229 ms. The mean duration per pulse was 249±53 µs and the inter-click interval varied from 0.4 to 3.0 ms. The mean values of peak, center frequency, -3dB and -10dB bandwidth were 27.9±6.5, 31.4±5.5, 10.8±2.7 and 31.2±8.2 kHz respectively.