Sound production and intraspecific sound signal structure
Males of O. croaticus produced a single type of acoustic signal, named pulsatile sound, during intersexual (male-female) interactions conducted within the reproductive season (April – October). Four resident males ( ± s.d. = 49.1 ± 0.8; range: 48.0 – 50.0 mm LT; 40.9 ± 1.8; range: 38.9 – 42.2 mm LS; 1.2 ± 0.1; range: 1.1 – 1.4 g W ; 1.7 ± 0.1; range: 1.5 – 1.9 Fulton’s K ) produced sounds when interacting with females, while the other four males remained mute and did not court. We recorded 372 sounds produced by the four males (mean 93.0 sounds per male). These signals were reminiscent of brief felid purrs. They were produced in a relatively irregular repetition pattern, at a rate of 4.7 sounds per min, while the maximum recorded rate was 10 sounds per min (7.7 ± 1.4; range: 6 – 10) (Figure 1, Table 1 ). Structurally, the sounds are short duration signals, lasting around 450 ms (442.0 ±132.6; range: 156.8 – 952.8 ms), and composed of a variable number of short pulses (14.2 ± 4.0; range: 5 – 32) of around 15 ms (14.5 ± 1.9; range: 9.7 – 22.9 ms) (Figure 2a-e, Table 1 ). The individual pulse unit structure differed markedly between sounds, exhibiting one to three peaks with variable amplitude. Generally, the amplitude of a sound changed gradually, first increasing and then steadily decreasing throughout the sound, with the first two or three pulses being the loudest (Figure 2a-e ). The pulse repetition rate varied from 26.0 to 38.0 Hz (32.5 ± 1.6 Hz), while the sound pulse period (PP) averaged 32 ms (31.9 ± 1.4; range: 27.6 – 37.5 ms). PP changed with water temperature, with higher values occurring at lower temperatures (Figure 3 ). Fatigue averaged 1.1 ± 0.08 (range: 0.9 – 1.4). The sound peak frequency varied from 89 to 340 Hz (137.4 ± 38.3 Hz), although several higher frequency components were also present, especially in the range 0.5 - 1.5 kHz. Energy extended from 0.05 to 2 kHz (Figure 2a-e, Table 1 ), with most of the sound energy within 0.05 – 0.6 kHz. Frequency modulation of the sounds ranged from 0.7 – 1.1 Hz. Additionally, the calling effort varied between males from 0.37 to 0.60 (0.49 ± 0.09), indicating that some individuals emitted sounds more frequently than others (Table 1 ). Interestingly, sounds were never organised in bursts, which are usually composed of several consecutive sounds produced with regular inter-sound interval, as observed in some sand gobies.
Intraspecifically, soniferous O. croaticus males differed significantly in all acoustic features (Kruskal-Wallis H test, χ2 = 8.53 – 41.97; d.f. = 3; N = 80; P< 0.05) except for calling effort (Kruskal-Wallis Htest, χ2 = 5.67; d.f. = 3; N = 7; P> 0.05) (Table 1 ). Some acoustic features were significantly correlated (Spearman rank correlations; rS = - 0.90 – 0.97; P < 0.05), while the strongest correlation was observed between NP and DUR (rS = 0.97), PRR with FMf (rS = 0.7), and PRR with PP (rS = -0.9). Note that for pairwise correlation between sound variables, we only consideredrS > 0.7 due to the limited sample size. These relations are expected as DUR results in part from NP, PRR and PP are redundant metrics, and FMf is the PRR of the initial part of the sound. In addition, some acoustic features were also correlated with physical features LS (PF and FMi, rS = -0.53; P< 0.05 for both variables), W (DUR and NP, rS = 0.53 and 0.55; P < 0.05), and water temperature (PRR, PP, PF and FMi, rS = -0.67 – 0.69; P < 0.05) (Figure 4 ).