A fish can change its stripes: investigating the role of body colour and pattern in the bluelined goatfish

Introduction

Animal colouration has several important ecological functions and many animals have the capacity to alter their individual colouration and pattern depending on social or environmental context (Aspengren, Sköld & Wallin, 2009). Morphological colour change typically occurs over long timeframes such as days or months (Sugimoto, 2002). By contrast, physiological, or rapid, colour change tends to happen more quickly on the order of milliseconds to hours (Duarte, Flores & Stevens, 2017). Rapid colour change provides individual plasticity with the potential for quick alterations and flexibility in colour and pattern depending on their behavioural requirements (Nilsson Sköld, Aspengren & Wallin, 2013). There are numerous reports of marine animals rapidly changing colour, with cephalopods considered the masters of this phenomenon. Cephalopods use dynamic colouration for camouflage (Hanlon, 2007) and as an intraspecific communication tool (Brown, Garwood & Williamson, 2012; Mäthger et al., 2009). Many studies on colour change in fishes have demonstrated a role in camouflage (Akkaynak et al., 2017; Allen et al., 2015; Ramachandran et al., 1996; Stevens, Lown & Denton, 2014; Watson, Siemann & Hanlon, 2014) but have also indicated that rapid alterations in fish colouration are used as conspicuous signals in courtship (Erisman & Allen, 2005; Svensson et al., 2006), competition (Berglund & Rosenqvist, 2001) and mimicry (Cheney et al., 2009). Some marine fishes use colouration in signalling, such as male minnows (Phoxinus phoxinus) where colourful ornaments may signal status, courting activity and superior quality (Kekäläinen et al., 2010). Moreover, some fishes adopt or enhance vertical banding patterns to communicate information (Berglund & Rosenqvist, 2001; Erisman & Allen, 2005; Galván-Villa & Hastings, 2018; Miller & Allen, 2006; Morris, Mussel & Ryan, 1995).

The bluelined goatfish (Upeneichthys lineatus) occurs along the southeastern coast of New South Wales and has the capacity for rapid colour change. U. lineatus can rapidly modify its colouration and body pattern from plain buff/white to a prominent dark red and white vertically banded pattern in less than ten seconds (Tosetto, Hart & Williamson, 2023). There are anecdotal reports of U. lineatus adopting a red colour when resting at night (personal observation) but during the day they are significantly more likely to shift colour and adopt a dark red pattern when actively feeding (Tosetto, Hart & Williamson, 2023), seemingly making them more conspicuous (Fig. 1). U. lineatus feed in a similar fashion to other goatfish using a shovelling action where their snout is buried in the substrate with their body angled more than 30 degrees relative to the substrate (Krajewski et al., 2006). This body position potentially restricts both their field of view and mobility, which may reduce their capacity to detect and escape from predators. When U. lineatus feed and adopt the red banded colouration, they regularly have conspecifics and associated fish following them, and a previous study found that heterospecifics will spend significantly longer interacting with U. lineatus that are displaying the dark red banding while foraging (Tosetto, Hart & Williamson, 2023).

Goatfish are zoobenthivores (benthic carnivores) and found throughout coastal areas in both temperate and tropical systems (Uiblein, 2007). Their capacity to form feeding aggregations through nuclear-follower foraging associations is well reported (Hanaway-Moore & Keeler, 2012; Lukoschek & Mccormick, 2002; Sazima et al., 2007). Group foraging increases the feeding success of follower fish by providing access to otherwise unobtainable prey (Ternes et al., 2018) while the nuclear fish (goatfish) may benefit through anti-predator mechanisms such as increased vigilance and dilution effects by the followers (Cresswell et al., 2003). Many studies, particularly in the case of goatfish, suggest that nuclear fish are identified by followers via bottom disturbance and the resulting sediment plume (Krajewski, 2009; Sazima et al., 2007). However, foraging associations may also be influenced by local enhancement, a process whereby individuals use social cues such as fish presence or body position to indicate the location of suitable food patches (Brown & Laland, 2003; Ryer & Olla, 1992; Waite, 1981). There is also evidence of fish using referential gestures (Vail, Manica & Bshary, 2013) to signal food and instigate collaboration but whether distinct changes in colour and pattern play a role in signalling foraging success has not been widely explored. It is possible that the dark red banding pattern exhibited by U. lineatus when foraging is an intra- or interspecific signal to indicate increased food acquisition via a ‘safety in numbers’ approach.

If the colour and pattern change exhibited by U. lineatus is indeed a signal, then potential receivers must have the sensory systems to detect this signal. If fish are using chromatic cues and it is the red colour providing the signal, then fish must possess visual pigments with the spectral sensitivity to discriminate those longer wavelengths (Endler, 1992; Marshall & Vorobyev, 2003). If it is an achromatic cue and receivers are using the contrast of the vertical banding as the signal, regardless of the colour, then fish must have the visual acuity to resolve those bands from an ecologically relevant distance (Cronin et al., 2014). Recent findings suggest U. lineatus are trichromats, with three cone pigments having wavelengths of maximum absorbance at 413, 494 and 524 nm, and can discriminate the ‘red’ colours when longer wavelengths of light are available (Tosetto et al., 2021). The visual acuity of U. lineatus is estimated at 6.2 cpd and the maximum distance from which the banding could be resolved in clear waters is approximately seven meters (Tosetto et al., 2021). Whether associated heterospecifics possess the spectral sensitivity to perceive longer (red) wavelengths or have the acuity to resolve the banding from a suitable distance is unknown. We do know that many coastal fish have visual pigments that are tuned to match the wavelengths of light that are most abundant in these habitats, typically in the short (blue) and medium (green) wavelength portion of the visible spectrum, and generally lack pigments that absorb strongly at longer (red) wavelengths of light (Bowmaker, 1984; Lythgoe & Partridge, 1991). Nevertheless, several coastal teleosts possess visual pigments that are similar to U. lineatus, so discrimination of red colouration is possible. The visual acuity of fish found in temperate coastal environments have not been widely published but fish inhabiting similar environments have visual acuity in the region of 10 cpd (Caves, Sutton & Johnsen, 2017), suggesting that the banding of U. lineatus may be resolved from 10 m away in suitable conditions. Given that the red colour cannot always be perceived, such as in low light when cone photoreceptors are not active or deeper water environments where there is little red light, and that some of the associated fish may not have the spectral sensitivity to discriminate the longer wavelengths of light, it is also possible that it is the change in patterning rather than the colour of U. lineatus that provides a meaningful signal.

In this study, we used three-dimensional (3D) printed model fish to investigate if changes in colour and / or patterning of U. lineatus play a role in signalling potential food patch locations to both conspecifics and associated heterospecifics. Such 3D printed models provide a way to produce identical objects while only manipulating the traits of interest (in this case colour) and preserving the ecological setting (Behm et al., 2018; Bulté et al., 2018). A similar approach has been employed across a range of behavioural studies, including using model bird eggs in evaluating egg rejection behaviour (Igic et al., 2015) and decoy female turtles to assess the effect of body size on mate choice (Bulté et al., 2018). Further, zebrafish have demonstrated similar attraction to both live conspecifics and 3D printed replicas (Ruberto, Polverino & Porfiri, 2017), suggesting that 3D models are an effective tool for assessing fish response to different colour morphs.

Here, we deployed 3D model goatfish in three different colour morphs; a red and white vertically banded model, a black and white vertically banded model and a plain white model with no banding to examine differences in fish behaviour and interactions in proximity to the different model variants. Using both a red striped and a black striped model fish allowed us to determine whether fish were responding to a chromatic or achromatic signal to discern between the colouration or the vertical banding. We also included a plain white model fish to control for interest in the 3D model fish. This study allowed us to remotely obtain information on the communities and behaviours of both conspecifics and heterospecifics in response to these different colour patterns via the use of remote underwater video, without the influence of human observers. Specifically, we examined whether U. lineatus or associated heterospecifics displayed differences in their behaviour (passing, attracted or foraging) in proximity to different coloured models. We also explored if the total amount of time that U. lineatus and associated heterospecifics spent around the different model fish varied. We predicted that both U. lineatus and associated heterospecifics would spend a greater amount of time around the red and black models when compared with the white models. We also predicted that if the red patterning is a foraging signal to conspecifics, then there would be a greater number of U. lineatus feeding near the red models than the black or the white models. Furthermore, because some coastal fishes may not be able to discriminate long wavelength reflecting colours, we expected that some associated heterospecifics would not be influenced by the colour and so anticipated no difference between the black and white banded model fish and the red and white banded model fish.

Materials & Methods

Remote underwater video apparatus

The Remote Underwater Video (RUV) apparatus consisted of a T-shaped structure constructed with PVC (10 mm diameter). The structure was weighted by placing 12 mm reinforcing bar (Whites 12 × 600 mm rib reinforcing bar) inside the PVC piping and sealing each end with PVC caps. This expedited deployment into sediment and negated the need for additional weights on the RUV. A single GoPro™ digital video camera (Hero 2018 model) was placed 600 mm away from the model fish (Fig. 2). We did not use bait to attract fish as the aim of the study was to ascertain if there was a spontaneous effect of the different colour morphs on fish community structure and behaviours.

Field sites

Experiments were conducted between July and September in 2018 and 2019 at 15 sites in New South Wales, Australia, from Sydney (35°52′39.77″S, 151°14′30.03″E) down to Eden (37°03′56.73″S, 149°53′59.09″E). Sites were chosen based on the presence of rocky reef with adjacent sand flats and the presence of U. lineatus. The RUVs were randomly deployed at five sites in Sydney, two sites in Botany Bay, six sites in Jervis Bay, one site in Kioloa, and one site in Eden (Fig. 3) (see Table S1 for details of all sites and GPS coordinates). Field work permits were issued by the Marine Fieldwork Manager at Macquarie University (TR-18-2893, TR-19-4335 and TR-19-4181).

Abundance and behaviour

Fish abundance was the total number of individual conspecifics and heterospecifics that interacted with the 3D model fish. For each interacting fish we recorded the species and its behaviour: passing, attracted or foraging. Passing was defined as no obvious interaction or interest in the model fish. Attracted was logged when an individual demonstrated some curiosity in the 3D model fish, such as stopping to look at the model fish, an obvious slowing in swim speed to investigate the model, or an obvious change in swimming direction towards the model fish. Foraging was recorded as any fish that probed or ate from the sediment within a 2 m radius of the model fish (Fig. 4).

We only recorded individuals where we could clearly see their eyes and markings and only one behaviour was recorded per individual. The scoring of behaviours was hierarchical whereby if an individual was both attracted to and fed near the model we recorded the behaviour as feeding given that the overall hypothesis is whether the change in colour and pattern is a foraging signal. Similarly, if it passed by and then demonstrated attracted behaviour, the attracted behaviour was recorded. If a fish left the field of view of the camera and a fish identical in size with distinct markings returned within 10 s, it was deemed to be the same fish and was not recorded again. If a similar looking fish returned into the screen after more than 10 s then it was recorded as a new fish unless clear markings identified it as the previous fish.

Results

In total there were 58 species of fish that interacted with the different model fish. The species recorded, including their frequency around each of the coloured models, is outlined in Table 1.

Abundance and behaviour—associated heterospecifics

Overall, there was no significant interaction between behaviour of the heterospecifics and the colour of the model fish (X² = 2.512, P = 0.642). Analysis of the main effects showed the colour of the model did not influence heterospecific abundance (F = 0.030, P = 0.970) but we did see overall differences in heterospecific behaviour (F = 32.943, P < 0.001) regardless of goatfish model colour. Associated heterospecifics fish passed the models significantly more than they were attracted (t = 2.776, P = 0.0160) or foraged (t = 8.065 P < .0001). There were also significantly more heterospecifics attracted to the models than foraged around the models (t = 5.085, P < 0.001) (Fig. 6).

Individual analysis of the nine key heterospecifics found no significant interactions between the model fish colour and their behaviour. Analysis of the main effects revealed some differences in behaviour regardless of model colour. O. lineolatus passed and showed attraction to the models significantly more times than they foraged around them (F = 5.370, P = 0.005), Atypicthys strigatus were attracted significantly more times than they foraged (F = 5.710, P = 0.004). Eupetrichthys angustipes passed the models significantly more times than fed near the models (F = 10.170, P < 0.001) and P. georgianus passed significantly more times than displayed attraction or fed near the models (F = 11.340, P < 0.001). All pairwise comparisons for the main effects are provided in Table S3.

Time-in-view–all associated fishes (conspecifics and heterospecifics)

Overall, there was no difference in the time that conspecifics spent around the different models (F = 0.286, P = 0.751). There was also no difference in the time spent by all the heterospecifics around the models (F = 1.150, P = 0.317). However, some differences in the time spent around the different model fish were observed for juvenile C. auratus (F = 5.296, P = 0.006), S. granulatus (F = 6.141, P = 0.007), G. subfasciatus (F = 3.331, P = 0.038), O. lineolatus (F = 3.102, P = 0.046) and A. australis (F = 9.828, P = 0.006). Juvenile C. auratus and S. granulatus spent significantly longer around the red model fish than the black or white fish. Gerres subfasciatus spent longer around the black model fish than the white model fish and O. lineolatus and A. australis spent more time around the white model fish than the black model fish. There were no differences in time spent around the different model fish for A. strigatus (F = 3.411, P = 0.041), P. spilurus (F = 0.803, P = 0.461), E. angustipes (F = 0.917, P = 0.406) or P. georgianus (F = 1.699, P = 0.197) (Fig. 7. See Table S4 for all pairwise comparisons).

Discussion

In this study we assessed whether the striped body pattern of U. lineatus represents a foraging signal that encourages group feeding. We further sought to understand whether, if this is the case, the relevant signal perceived by the associated fish is achromatic or influence by colour. Overall, we observed 58 different fish species and over 4,000 individual fish interacting with the 3D models. There was no interaction between the behaviour of conspecifics and the model fish colour. We did observe significantly more foraging interactions by conspecifics around the model fish than passing or attracted interactions irrespective of colour. We also recorded more interactions with the red and black model fish when compared with the white models regardless of behaviour. However, no difference was observed when assessing the total time that conspecifics spent around the different models. Associated heterospecifics did not exhibit differences in behaviour in response to different models. We did record more passing interactions for some key heterospecifics when compared with foraging or attracted interactions around the model fish. There were also significant differences in the time that some fish species spent in the proximity of the different models, suggesting that the striped body pattern of U. lineatus may be providing an interspecific signal to some species.

The results suggest that conspecifics may be responding to the difference in pattern but whether they are using the dark red banding as a foraging signal is not clear from these results, suggesting any intraspecific signalling in U. lineatus may be complex. Upeneichthys lineatus was the third most abundant fish that interacted with the 3D models, only surpassed by two species that we observed in large schools, T. novaezelandiae and G. subfasciatus. Higher numbers of conspecifics were observed in proximity to the red and black model fish than the white control fish, but we did not observe any differences between the black and red coloured models. It is possible that U. lineatus can recognise the banded pattern but cannot or do not distinguish between the red and black-striped variants, suggesting that the body pattern of U. lineatus encodes mainly achromatic information. The colour of the banding may or may not be important for U. lineatus depending on the context. It may be that a more universally reliable signal is achromatic which is perceivable under all circumstances, whereas when the observer is viewing the goatfish banding from a distance, or where red light is unavailable, the red pattern is less noticeable. We also observed significantly more goatfish foraging near the models than passing by or showing attraction, irrespective of colour, but it was rare that we observed conspecifics eating alongside the models, with most probing the sediment for food with their barbels. Whether foraging increased around the deployed models was not explored in this study, and any future work should consider the inclusion of a negative control. The static nature of the models, and the fact that they were likely not placed in preferred food patches, may explain why we did not see U. lineatus feed, and possibly why there were no differences in the amount of time conspecifics spent around the models. Without stereo-cameras, we were unable to accurately assess fish size although many conspecifics appeared smaller than our model fish. A previous study found that conspecifics will regularly adopt a white colouration when following focal goatfish (Tosetto, Hart & Williamson, 2023), suggesting that red patterning may also be a dominance signal in U. lineatus. These findings are indicative of social learning within U. lineatus populations and further studies that explore the role of colour in this system are warranted. We rarely observed single-species foraging groups amongst adult goatfish but this may be because adult U. lineatus share resource requirements (Powell, 1989). Rather, adult conspecifics could be using the body position and colour as an indicator of nearby food but recognise the dark red banding as foraging success and avoid competition by finding an alternative patch to feed. Intraspecific signals may be more complex than the colour change alone and more dynamic models that combine appropriate food resources would allow us to extrapolate whether colour change is being used as an intraspecific signal in U. lineatus.

Responses of associated heterospecifics to the model fish varied, indicating that interspecific signalling may be multifaceted and driven at a species level. No difference in the abundance or behaviour of associated heterospecifics in response to the different coloured models was observed but there were some differences in behaviour unrelated to model colour. Pseudocaranx georgianus passed the models significantly more than fed or were attracted, but as a mid-water schooling species that don’t naturally occur on rocky reefs (Folpp et al., 2013) this is not surprising. Atypichthys strigatus were clearly attracted to the models but no feeding was observed. Atypichthys strigatus are zooplanktivores that commonly respond to disturbance events to feed in the water column (Glasby & Kingsford, 1994; Kuiter & Kuiter, 1997). It is likely they were attracted to the head down feeding position of the models, but the absence of disturbance may have meant no food was made available. We also observed two wrasse species, O. lineoata and E. angustipes, displaying attraction to and passing by models more than we observed them feeding. Wrasse are the most opportunistic carnivores in exploiting fishes that foraging through sediment (Aronson & Sanderson, 1987). Like A. strigatus the feeding position, regardless of colour, may provide an initial attraction for the wrasse but the absence of foraging means infauna are not disturbed and subsequently, there are no food resources to exploit. Including sediment disturbance in these models would likely provide insight into the motivation of these follower fishes. However, if the dark red banding of U. lineatus is indeed a signal then we would expect to see a difference in response to the different models regardless of food availability. It is possible that the foraging signal is multifaceted, but it is likely that these species respond to the nose down feeding position of any benthic fish in the anticipation of food.

Assessing the time heterospecifics spent around model fish provides greater insight into associations and potential signals. Both O. lineolata and A. australis are benthic foragers and both spent more time around the white model fish, contrary to our expectations. Due to the opportunistic nature of wrasse (Sazima et al., 2005), it is possible they are attracted to any fish exhibiting a nose down feeding position. A. australis are generalist feeders and will often feed in shallow waters during daytime hightide to avoid predation (Gannon et al., 2015; Taylor, Becker & Lowry, 2018) and possibly do not rely on increased food or vigilance from interspecies group feeding. Furthermore, given coastal fish are often dichromats lacking the visual pigments to see in the longer (red) wavelengths (Marshall, Carleton & Cronin, 2015) and some species of marine fish have excellent contrast sensitivity (Santon, Münch & Michiels, 2019), the white model may have been more obvious to opportunistic fish. We observed no variation in time spent around the models for P. georgianus or A strigatus but as they are both predominately planktivores, the absence of food meant there was little benefit spending time in proximity to the model. Surprisingly, the blacksaddle goatfish P. spilurus did not differ in time spent around different models either. Personal observations suggest there is dietary partitioning as P. spilurus were often seen foraging on rocky reefs rather than bare sand, and regularly seen in schools, thus there may be no extra foraging or vigilance benefits gained from foraging with U. lineatus.

Three species of fish that spent significantly more time around the red and black model fish than the white control fish were G. subfasciatus, S. granulatus and juvenile C. auratus, all of which are benthic carnivores (Edgar, 1997). We observed that G. subfasciatus spent more time interacting with the black-banded model fish, suggesting a response to the banded pattern. While G. subfasciatus and U. lineatus both forage over bare sand it is possible they are not targeting the same prey. Dietary partitioning has been demonstrated between G. subfasciatus and U. tragula (bartailed goatfish) via different feeding strategies. Gerres subfasciatus use their highly protrusible mouth to target surface polychaetes, while goatfish use their barbels to extract prey living deeper below the substrate surface (Linke, Platell & Potter, 2001). Perhaps U. lineatus expose infauna that G. subfasciatus can then exploit from the surface, thereby providing benefits to both fish without competition for resources.

Two fish that spent a significantly higher proportion of time around the red-banded goatfish models were S. granulatus and C. auratus. There is little information available on the visual ecology of S. granulatus but they are habitat specialists, predominately found on seagrass (Harvey et al., 2013). In this study, we regularly observed S. granulatus with the model fish when they had been inadvertently placed on seagrass beds. Given seagrass depth limit is related to light attenuation underwater (Duarte, 1991), it is possible S. granulatus possesses visual pigments sensitive to a broad range of wavelengths present in shallow water, including longer wavelengths, and can perceive the red colouration. The benefits to S. granulatus associating with goatfish are not clear but despite seagrass edges providing greater food availability (Macreadie et al., 2010) prey species will regularly use the interior of seagrass patches for protection from predators (Smith et al., 2011). Interestingly, in Port Phillip Bay on Australia’s south coast the bluestriped goatfish (Upeneichthys vlamingii) and bridled leatherjackets (Acanthaluteres spilomelanurus) are regularly found in deeper seagrass beds (3–6 m) (Smith, Jenkins & Hutchinson, 2012). Perhaps the foraging activity of U. lineatus provides prey to S. granulatus allowing them to remain in the safe interior of the seagrass patch. Juvenile C. auratus are also regularly found foraging with U. lineatus over exposed sand flats adjacent to rocky reef habitats, a distribution thought to balance requirements of food acquisition and predator protection (Ross et al., 2007). It has been reported that juvenile C. auratus have visual pigments with sensitivity in the longer wavelength area of the spectrum (around 570 nm) (Robinson et al., 2017) suggesting the capacity to discriminate the colour red. Perhaps juvenile C. auratus are using the red patterning of the goatfish to indicate suitable food resources whilst also benefiting from increased predator protection. Furthermore, adult snapper also adopt a red body colouration (Booth et al., 2004). Skin pigmentation requires a dietary source of carotenoids (Doolan et al., 2008) which likely come from crustaceans found in the benthos. Given that U. lineatus likely obtain carotenoids from their diet, it may be that C. auratus also follow goatfish as a source of the quality food that they require. It is worth noting that there are some outliers in these data that may have influenced the results. More data are required to strengthen the associations that individual species may have with U. lineatus. There is an exciting opportunity to investigate the drivers of this relationship in more detail and gain insight into some of the important associations and the complexities in signalling taking place in coastal temperate environments.

The increased time that some fish spend around the red-and black-banded models suggests they were responding to the changes in colour and pattern, but without associated bottom disturbance it is difficult to tease out the benefits for associated heterospecifics. Early studies on multi-species foraging associations suggest follower fish are initially anticipatory in their behaviour, following nuclear fish whenever they see them regardless of their behaviour (Strand, 1988). A more recent study (Krajewski, 2009) suggests the initial driver of attraction is bottom disturbance, regardless of the source. In both cases it is reported that fish will only forage with a nuclear fish when there is evidence of them feeding successfully. To date, any examination of group feeding featuring goatfish as the nuclear fish has focused on bottom disturbance as the social cue. Two goatfish species regularly assessed in foraging associations are Pseudupeneus maculatus (spotted goatfish) and Parupeneus barbernius (dash-and-dot goatfish) (Krajewski, 2009; Lukoschek & Mccormick, 2002; Sazima et al., 2006; Sazima et al., 2007). Both are anecdotally reported to change colour, but any changes in colour have not been considered in foraging associations to date. It is possible that bottom disturbance and foraging signals, such as visual cues, are not mutually exclusive and foraging signals reinforce the initial attraction to disturbance. To explore this relationship, (Krajewski, 2009) used models of P. maculatus and disturbed the underlying sediment but found no interaction between the visual appearance of the fish and disturbance. However, changes in goatfish colour were not considered, just the neutral appearance of the fish. Our current study focused on the possible signal of the dark red banding but did not include any bottom disturbance. If the two are not mutually exclusive, then it is essential that any studies combine potential foraging signals with more dynamic models.

The quality of the signal, whether chromatic or achromatic, could be influenced by the intensity of the banding. The rapid colour change of U. lineatus from white to dark red is continuous, shifting from a white to paler red when searching for food and the adoption of the dark red colouration when actively feeding (Tosetto, Hart & Williamson, 2023). It is possible that the intensity or saturation of the dark red bands may indicate the food patch quality. Alternatively, the capacity to turn a deeper red may vary within individuals with the fish that turn darker red indicating better body condition. There are reports showing that some species of fish prefer ‘redder’ individuals. For instance, male cyprinid fish (Puntius titteya) prefer redder males (Fukuda & Karino, 2014), female three-spined sticklebacks (Gasterosteus aculeatus) use the intensity of red colouration in males as an indicator of functional fertility (Pike et al., 2010), and male two-spot gobies (Gobiusculus flavescens), prefer to mate with females with bright yellow-orange bellies (Amundsen & Forsgren, 2001). In situations where ‘red’ is not perceivable, an enhanced saturation of the red banding would increase the contrast between the red and white vertical bands, potentially enhancing the conspicuousness of the achromatic signal. When signalling to potential mates and to facilitate spawning, Paralabrax maculatofasciatus (kelp bass) enhance the contrast of their vertical dark bars (Erisman & Allen, 2005) and in sword tails (Xiphophorus sp.), vertical banding on males intensifies during social interactions (Morris, Mussel & Ryan, 1995). Future studies that seek to test whether the dark red colouration could be indicating the quality of food, or individual body condition are certainly warranted.

A diversity of fish can rapidly change colour and increasingly we are learning that the drivers of this are varied. Camouflage remains one of the most studied drivers of colour change and its main function appears to be for signalling. In some cases, dynamic colour change can be used for multiple reasons such as Nassau groupers (Epinephelus striatus) that use colour change for camouflage (Watson, Siemann & Hanlon, 2014) as well as social signalling (Archer et al., 2015). In African dwarf chameleons (Bradypodion spp.) colour change is regularly used in camouflage but it evolved as a conspicuous signal in male contest (Stuart-Fox & Moussalli, 2008). There is anecdotal evidence to suggest U. lineatus adopt a red colour at night, possibly for camouflage while resting. It is possible this colour change is multifunctional, which may have evolved for camouflage and has been co-opted for use in social signalling. Alternatively, the selection for conspicuous social signals may have driven the evolution of colour change in these goatfish. There are anecdotal reports of several other goatfish species having the capacity to rapidly change colour but to date this has rarely been investigated. There is only one other study that we are aware of that quantifies colour change in goatfish. In a mutualistic signalling event, M. martinicus (yellow goatfish) and P. maculatus (spotted goatfish) will shift body colouration, appearing darker to cleaner shrimp and indicating their willingness to be cleaned (Caves, Green & Johnsen, 2018). These findings demonstrate that different species of goatfish use rapid colour change for signalling but in different contexts. Such signals and sensory systems of receivers may have evolved in response to changes in local environments (Endler, 1992).

Conclusions

A total of 58 different fish species interacted with the 3D models. U. lineatus was observed in higher abundance around the coloured model fish than the white control model fish suggesting U. lineatus could recognise the banded pattern but did not distinguish between the red- and black-coloured variants, suggesting that the body pattern of U. lineatus encodes mainly achromatic information. The lack of goatfish foraging near the models could be due to the static nature of the model fish. Alternatively, it could suggest competition for resources between conspecifics. Heterospecifics spent longer time periods around the red and black model fish than the white control fish. Because the model fish were static and did not increase the food availability in the benthos, it was difficult to obtain conclusive results as to whether follower fish benefited from this signal by obtaining access to food resources or were benefiting purely from increased protection. Including different coloured models that incorporate movement and bottom disturbance would provide greater insights into drivers of colour change and subsequent benefits to associated fishes.

Understanding which species in temperate systems are responding to the foraging signal of U. lineatus, as well as understanding the benefits received, will provide greater understanding into this signalling system. It is likely that heterospecific fish obtain additional food resources by foraging with goatfish but this requires further in-situ research to tease this apart. Further, coupling such research with an assessment about the visual systems of different signal receivers will provide greater insight into the complexities and coevolution of signalling systems. Given that there is evidence of other goatfish species using colour change as a signal means there are exciting opportunities to explore the evolution of rapid colour change using the goatfishes (Mullidae) as a model family.

Supplemental Information