Fin whale song evolution in the North Atlantic

(a) Table showing information of each song analysed in this study: date, time of the first (Time first note) and last note analysed (Time last note), duration and number of INIs measured (Num. INIs). (b). Position, recording equipment and approximate (Approx.) depth for each location sampled. (c). Description of fin whale song measurements made with Raven Pro 1.4 software (Charif et al., 2010). (d) The effect of recording equipment on fin whale song parameters When comparing data from multiple sensors an obvious question is whether the results might be dependent on the specific sensors considered. To investigate the influence of the acoustic recorder type, OBS or ARs, on the song parameters, we analysed the same song fragment, consisting of 209 notes, recorded by the hydrophone channel of an OBS and an AR, specifically an Ecological Acoustic Recorder (EAR) (Lammers et al., 2008). The two instruments were deployed at ~6 km from each other in the Azores region in spring of 2019. Measurements of INIs and 20 Hz peak frequencies of songs recorded by each instrument were compared using a non-parametric paired samples Wilcoxon Test. Differences in HF note peak frequencies could not be tested because of limitations in the sampling rate of the OBS. Results showed that median INIs measured from OBS (16.52) and EARs (16.42) were not significantly different (P-value = 0.47) but median peak frequencies of the 20 Hz note (median OBS = 23.4; EARs = 21.1) were (P-value <0.001). Thus, the use of different recorders did not affect INI measurements but influenced measurements of 20 Hz peak frequencies. The effect of the distance to the source in the analysed fin whale song parameters is included in this study, given that these two recorders were positioned at different distances to the singer. Peak frequencies of the 20 Hz note showed a great variability between equipment types (Fig. S3), which hindered the identification of soft trends (i.e., low changing rate). For this reason, only data from the EARs, the longest dataset (2008 – 2020), were used to study temporal variations of the peak frequencies of the 20 Hz note. All statistical analyses were performed using the software R (v. 4.0.2) (R Core team, 2020).