Reproductive ethology and association with sounds
We identified nine male behavioural categories within three distinct reproductive phases in O. croaticus : Chase, Lead, Approach and Circling (courtship phase), Nest display, Frontal display, Nest rubbing, Pre-mating (pre-spawning phase) and Spawning (spawning phase). The category Pre-mating included all behaviours that occurred after the female entered the male nest and prior to belly-up position (oviposition ). Spawning was not observed, but we observed the female’s repeated upside-down or belly-up position (an indication ofoviposition ), during which she probably laid eggs on the nest ceiling. Briefly, the behaviours Chase, Lead, Approach and Circling were performed by the male outside the nest (recognised as the “courtship” phase), while Nest display, Frontal display, Nest rubbing, Pre-mating and Spawning were performed within the nest (“pre-spawning” and “spawning” phases). In some cases, Nest display and Frontal display were performed by the male occupying the nest with or without a female inside. However, Nest rubbing, Pre-mating and Spawning were always performed by the male when the female was inside the nest. The frequency, duration and overall percentage of these categories were scored during ethological analysis. The first dataset included seven video recordings (210 min) where at least one sound occurred per recording. Overall, we observed 410 behavioural acts (102.5 per male). The behavioural categories Nest display (29.3%), Pre-mating (22.7%) and Approach (19.1%) were most frequently observed, while Circling (1.2%), Chase (1.0%) and Lead (0.7%) were rarely recorded (Figure 5a-d ). Concerning the association of sound production with behavioural acts, of the 410 behavioural acts, 99 acts (24.1%) were accompanied by sound production. The behaviours Pre-mating (303 sounds), Nest display (27 sounds), Frontal display (20 sounds), Nest rubbing (16 sounds) and Spawning (5 sounds) were coincided with sounds, while Approach, Circling and Lead never co-occurred with sound production (Figure 5a ). During four Chase acts, only one sound was recorded. Only one spawning act (Spawning) was observed in this study, during which five sounds were produced. Unfortunately, spawning sounds were not used in the comparative purposes due to their limited occurrence. The chi-square (χ2) test of independence indicated that behavioural categories Nest display and Pre-mating were significantly associated with sound production (χ 2 = 138.3; d.f. = 5; P < 0.05; residual score: 1.5 and 41.5, respectively), while other categories failed to support this hypothesis (Figure 5a ).
To compare male behaviour when soniferous or mute, a second dataset of eight video recordings was considered (190.5 min) of the same four males but in which no sound production was documented. In these recordings, we have observed 324 male behavioural acts (averaging 81.0 per male), of which Approach (38.3%), Lead (31.8%) and Nest display (20.1%) were the most frequent categories (Figure 5b ). Contrary, Chase, Circling, Pre-mating or Spawning were not documented within these recording sessions. In general, there is an obvious dissimilarity between the frequency of the behavioural categories that were or were not accompanied by sounds. Specifically, Pre-mating, one of the two behaviours significantly associated with sound production, decreased from an average of 22.7% in the trials with sounds to 0% in the trials without sound. In addition, Nest display, Nest rubbing and Frontal display acts produced during sound emission decreased from 29.3%, 13.9% and 12%, to 20.1%, 3.7% and 6.2% in frequency in experiments without the sounds, respectively. On the other hand, Approach and Lead were more frequent during the mute behaviours (31% and 38%, respectively) than during sound production (19% and 0.7%, respectively) (Figure 5b ). Overall, the behavioural rate decreased from 55.8% to 44.1% when males produced sounds in comparison to when they were mute (soniferous vs. mute males: means 2.79vs. 1.55), though the differences were not significant (Wilcoxon signed-rank test, P > 0.05) (Figure 6a ). Importantly, the number of times the females entered the male nest differed significantly between the two datasets (means 3.71 vs 0.71), as female nest entrance was more frequent when males produced sound than when they were mute (Wilcoxonsigned-rank test, P < 0.05) (Figure 6b ).
When producing sounds, the occurrence frequency between the courtship (Chase, Lead, Approach and Circling) and pre-spawning (Nest display, Frontal display, Nest rubbing, Pre-mating) phases of reproduction did not differ between soniferous males (Wilcoxon signed-rank test,P > 0.05), though their duration did (Wilcoxon signed-rank test, P < 0.05). Generally, males exhibited courtship-related behaviours less frequently and for a shorter period compared to pre-spawning behaviours ( = 22.5 vs. 79.7 for frequency and = 173.4 vs.2393.3 for duration).