Behavioural observations

We first conducted a pilot study over 2 days to examine potential habituation by fish engaged in cleaner–client interactions to the presence and noise of a SCUBA diving observer. Four observers watched a total of 12 juvenile cleaner wrasses at one site, which was not part of the main study. We studied juveniles for two reasons: (1) at this location, they are more active cleaners than adults (IMC, SCM, personal observations) and (2) their cleaning behaviour is not yet influenced by sexual motivation, making it more consistent across individuals. Each cleaner was observed, in the absence of any motorboat noise, by a motionless diver hovering or kneeling on the sand 2–3 m (horizontal distance) away for three successive 20-min periods, which mirrors the experimental treatment (see below). During each 20-min period, we recorded the species identity of each visiting client and duration of each inspection (i.e., when the cleaner surveys the body surface, gills or buccal cavity of the fish client).

Following the pilot study, we searched for juvenile cleaner wrasses at eight additional sites, and each diver observed the first cleaner he/she encountered. All observations lasted 60 min, divided into three 20-min periods: (1) a pre-exposure, control period (no motorboat noise); (2) an exposure period (motorboat noise present); and (3) a post-exposure period (no motorboat noise). At the end of the pre-exposure period, all diving observers released a tethered surface marker buoy to signal the start of the exposure period. A research assistant drove (estimated speed: 4–10 knots) one of two small outboard motorboats (boat with 25 hp engine chosen randomly on each day) multiple times past the marker buoys, no closer than 10 m from the buoy (for diver safety) and no further than 100 m. This treatment was expected to generate noise at a broad range of frequencies including 1–3000 kHz, which is audible to most fish species⁵², at levels of 80–150 dB re 1 μPa (RMS, full spectra), which have been shown to affect behaviour in reef fish (e.g. ref. ⁶). We recorded the experimental motorboat noise underwater to measure the actual range of frequencies and levels generated around a cleaner (see below).

During each of the three 20-min periods, we recorded the number and species of fish visiting the focal cleaner and the duration of each inspection. We also noted whether the interaction was initiated by the client by adopting a stereotyped incitation posture, which increases the likelihood of being inspected³⁵, whether clients jolted during the interaction, and whether clients chased the cleaner. We moved to a new site after each observation to ensure that cleanerfish were only observed once. Because fish were not marked, a few individual clients might have contributed to the clientele of more than one cleanerfish; however, this is unlikely to have occurred often given the high density of fish at the study sites. Other boats were not observed or heard during observations.