Bird predation poses a strong selection pressure on fish. Since birds must enter the water to catch fish, a combination of visual and mechano-acoustic cues (multimodal) characterize an immediate attack, while single cues (unimodal) may represent less dangerous disturbances. We investigated whether fish could use this information to distinguish between non-threatening and dangerous events and adjust their anti-predator response to the perceived level of risk. To do so, we investigated the anti-predator behavior of the sulphur molly (Poecilia sulphuraria), a small freshwater fish which is almost exclusively preyed on by piscivorous birds in its endemic sulfide spring habitat. In a field survey, we confirmed that these fish frequently have to distinguish between disturbances stemming from attacking birds (multimodal) and those which pose no (immediate) threat such as bird overflights (unimodal). In a laboratory experiment, we then exposed fish to artificial visual and/or acoustic stimuli presented separately or combined. Sensitivity was high regardless of stimulus type and number (< 96% of fish initiated diving), but fish dove deeper, faster and for longer when both stimuli were available simultaneously. Based on the system’s high rates of bird activity, we argue that such an unselective dive initiation with subsequent fine-tuning of diving parameters in accordance to cue modality represents an optimal strategy for these fish to save energy necessary to respond to future attacks. Ultimately, our study shows that fish anticipate the imminent risk posed by disturbances linked to bird predation through integrating information from both visual and acoustic cues.

Topic: We investigated the importance of visual and acoustic stimuli in aerial predator avoidance of a small freshwater fish, the sulphur molly Poecilia sulphuraria. Our first aim was to explore the diversity of disturbances to which sulphur mollies are exposed in their natural habitat. We then tested whether fish increased their anti-predator response with the number of predator stimuli by presenting  a moving object above the surface (unimodal visual), a playback of an impact sound (unimodal acoustic) or both stimuli simultaneously (bimodal). 

DATA_SI_BirdActivityField: Bird activity
We recorded all sightings of birds in the predefined study area within a 30-min period. For each bird, we determined species (to lowest feasible taxonomical level), entry and exit times, as well as the number of attacks. Bird sightings were categorized as either 'stay' or 'overflight' (i.e. flying through the transect without landing or attacking). We calculated presence time as the mean time birds spent in the study area (excluding overflights) and attacks as the total number of bird attacks launched in a sampling period. 

DATA_SI_DivingGroupmeans: Lab experiment – predator stimuli 
We exposed fish to simulated bird attacks presenting them with one of three stimulus types: unimodal visual, unimodal acoustic or bimodal (visual + acoustic). In a laboratory setting, we tested a total of 5 groups (12 individuals each; n = 60). Data provided here is group-pooled, but individual tracks are available upon request.

DATA_SI_DivingField: Supplement – predator stimulus validation
We assessed fish's dive reaction in response to simulated bird attacks using a focal shoal sampling approach. Six focal shoals were chosen along the sulphidic stream habitat (but in correspondance with sites reported in Culumber et al. 2016). The application of the acoustic stimulus was not feasible under field conditions. We stimulated fish 10–12 times with a minimum of 60 s between exposures and recorded diving responses within a frame of interest that allowed for the detection of individual fish. As the information of a detected threat often propagates through a group, we determined the interval from the point that 50% of fish had initiated diving until 50% had resurfaced.  

DATA_SI_DivingBodysize: Supplement – lab experiment
For the experiment, test fish were visually matched for size. To verify that groups did not significantly differ in size, body sizes were estimated from video images taken during the experiment. To minimize a systematic error introduced through fish’s distribution in the z-axis (i.e. distance to camera), measurements were only taken for individuals that were very close to the tank’s front glass and not occluded by another fish. Monitoring of water paramters (DO, Temp) was carried out prior to a group's first trial and repeated after the last trial.

2021_Lukas_Multisensory.R (see 2021_Lukas_Multisensory_Report for a PDF version)
All analyses were performed in R (R Core Team, 2020, version 4.0.2).

SIAudio_Stimulus_Impact.wav
The acoustic stimulus consisted of an underwater recording of a projectile hitting the water (duration: 1 s)