Ultraviolet (UV) vision is widespread among teleost fishes, of which many exhibit UV skin colours for communication. However, aside from its role in mate selection, few studies have examined the information UV-signalling conveys in other socio-behavioural contexts. Anemonefishes (subfamily, Amphiprioninae) live in a fascinating dominance hierarchy, in which a large female and male dominate over non-breeding subordinates, and body size is the primary cue for dominance. The iconic orange and white bars of anemonefishes are highly UV-reflective, and their colour vision is well-tuned to perceive the chromatic contrast of skin, which we show here decreases in the amount of UV-reflectance with increasing social rank. To test the function of their UV-skin signals, we compared the outcomes of staged contests over dominance between size-matched Barrier Reef anemonefish (Amphiprion akindynos) in aquarium chambers viewed under different UV-absorbing filters. Fish under UV-blocking filters were more likely to win contests, whereas fish under no-filter or neutral-density filter were more likely to submit. For contests between fish in no-filter and neutral density filter treatments, light treatment had no effect on contest outcome (win/lose). We also show that sub-adults were more aggressive towards smaller juveniles placed under a UV filter than a neutral density filter. Taken together, our results show that UV-reflectance or -contrast in anemonefish can modulate aggression and encode dominant and submissive cues when changes in overall intensity are controlled for.

Experiment 1: effect of UV on staged contest outcome

To test the importance of UV-signalling in anemonefish interactions, fish were paired against size-matched opponents (n = 70 trials) within 1.0 cm SL (Δ_SL range = 0.0 cm to 0.85 cm; mean Δ_SL ± s.d. = 0.32 ± 0.24 cm). Fish were used in one to six contests, depending on the number of size-matched pairings (average number of contests per fish ± sd = 3.0 ± 1.5). No repeat pairings, nor pairs comprised of fish from the same anemone were made. Each pairing was pseudo-randomly assigned to one of three treatments: 1) no-filter versus UV-filter (n trials = 21), 2) no-filter versus ND (n trials = 15), and 3) ND versus UV-filter (n trials = 34). At the beginning of each contest, fish from each pair were transferred by hand-net into one section of the tank. They were given an adjustment period of 30-min with an opaque barrier placed in front of the acrylic window to prevent any pre-contest interactions. Immediately prior to contest commencement, the barrier was removed, and interactions were video recorded for 20-min using a Go-Pro (Hero 3) mounted on a tripod with a full view of the experimental tank. At 20-min, the contest was terminated, and fish were returned to their aquaria.

Analysis of the footage involved recording key agonistic and submissive behaviours when anemonefish interacted with each other at the acrylic window. The outcome (winner and loser) of contests was decided by the first observation of submissive posturing displayed by either fish, subsequently deemed the loser. Because of the dual role that head-shaking serves in agonism and submissiveness, and the occasional subtlety between dorsal leaning and ventral leaning, we avoided ambiguity by adhering to a strict criterion that postures were only submissive when immediately preceded by an attack (i.e., charge or bite) from the opponent.  Upon observing submission, the video time stamp (in seconds), and outcome (loser and winner fish IDs and corresponding treatments) was recorded and footage analysis was terminated. 

Experiment 2: effect of UV on aggression directed at subordinates

To further investigate the potential role of UV-signalling in anemonefish social interactions, we compared the effect of UV light and reduced UV light on the level of aggression directed by larger fish at smaller (less-dominant) fish. In this experiment, anemonefish pairings had a SL difference over 1.0 cm (Δ_SL range = 1.1 cm – 3.5 cm; mean Δ_SL ± s.d. = 2.1 ± 0.72 cm), which exceeded the minimum threshold size difference between adjacent hierarchical ranks. This perceived difference in rank was further emphasised by almost exclusively pairing sub-adults with juveniles. Prior to trial commencement, individual fish were randomly assigned (via coin toss) to either the UV-filter or ND-filter treatment and similar to the previous experiment were given a 30-min adjustment period with an opaque barrier placed in front of the acrylic window to prevent any prior interaction. Trials commenced after removing the opaque barrier and were video recorded for 20-min, after which video recording stopped and fish were returned to their aquaria.

Footage analysis involved counting the number of different agonistic behaviours directed by the larger fish towards the smaller fish at the acrylic window. Behaviours exhibited by smaller fish were limited to typical submissive postures in response to attacks. Trial footage was analysed for the full recorded 20-min (n = 20).

Skin colour measurement and analysis

After the behavioural experiments were completed, the spectral reflectance of multiple skin patches (Figure 2) was measured for anemonefish (n = 47) used in the experiment. Measurements were done across a 300-700 nm range using an Ocean Optics Jaz spectrometer (Ocean Optics, Largo, FL, USA) with pulsed Xenon light source (Jaz-PX module), and a bifurcated 200 µm fibre optic cable (Thorlabs, NJ, USA). Reflectance measurements were taken at a 45° angle against skin patches, relative to a Spectralon 99% diffuse reflectance standard (Lab-sphere, North Sutton, NH, USA). All measurements were taken out-of-water and in a dark room by two personnel. One person maintained the fibre in position above skin patches, while the second person took measurements and gradually exposed skin patches caudal-rostrally by rolling back a seawater-soaked towel that wrapped fish to protect their skin from desiccation and eyes from the bright light. We ensured consistency among measurements by targeting the same locations across anemonefish and compensated for changes in body contour/relief by adjusting the held distance of the fibre accordingly. Three measurements were taken per skin patch (within 0.5 cm² – 1 cm²) and the average was used in subsequent analyses. To minimise stress from air exposure, all measurements were taken within two min per fish and no stress-related colour changes were observed. Post-measurement, all fish promptly recovered in a bucket of seawater and were then returned to aquaria. A low number of measurements (39/225, ~ 17%) were discarded due to either being saturated or containing artefacts.

Sea anemone (Stichodactyla gigantea) reflectance was based on averaged (n = 10) measurements recorded in-situ using a submersible spectrometer (USB2000 Ocean Optics) with a 100µm fibre and relative to a 99% Spectralon reflectance standard positioned next to the anemone. Reflectance measurements of tentacles used natural daylight as a light source at midday during non-overcast conditions at ~3m depth.

Downwelling spectral irradiance measurements were taken at ~5m depth on the reef (average of n = 3) using a submersible spectrometer (USB2000 Ocean Optics) with a 100µm fibre and cosine corrector.

Microsoft Excel

Windows Media Player (or equivalent video player)

File (de-)compression software.