Broad spectrum artificial light at night increases the conspicuousness of camouflaged prey

11 1. The growing global prevalence of energy efficient broad spectrum lighting 12 threatens to disrupt an array of visually guided ecological processes. Broad 13 spectrum lighting likely better enables the discrimination of colour, yet it’s 14 potential to increase the conspicuousness of camouflaged prey at night remains 15 little explored. 16 2. Using a well-established visual model, we quantified the impacts of four 17 spectrally distinct narrow and broad spectrum lighting technologies on the 18 conspicuousness of three different polymorphic colour variations of intertidal 19 littorinid snail, as viewed by three model predators.


Introduction
The prevalence of Artificial Light at Night (ALAN) has increased dramatically due to the expansion of urbanised areas worldwide (Falchi et al., 2016;Kyba et al., 2017).
Estimates indicate that 23% of the world's surface between 75°N and 60°S is affected by ALAN (Falchi et al., 2016) with a rate of increase of 2.2% between 2012 and 2016 (Kyba et al., 2017).While these developments herald a new age of simplicity in night time travel and security, an array of deleterious repercussions have been documented for humans and animals alike (Kempenaers et  As technologies develop, there has been a shift from narrow spectrum low-pressure sodium (LPS) towards luminaires that emit across a broader range of wavelengths (Elvidge et al., 2010;Davies et al., 2013), including High Pressure Sodium (HPS), Metal Halide (MH), and more recently Light Emitting Diodes (LED's) (Kyba et al., 2017).It is projected that LED bulbs will account for 85% of the global street lighting market by 2028 (Northeast Group LLC, 2019).Numerous concerns have been raised regarding the unforeseen ecological impacts of broad spectrum lighting (see Davies & Smyth, 2018 for an overview).Perhaps the most intuitive, yet little quantified of these impacts is the encroachment of light at night that enables colour guided behaviours previously only possible during the daytime (Davies et al. 2013;Briolat et al. 2021) or possibly under a full moon.
Camouflage is employed by a vast number of organisms to reduce conspicuousness.
While methods of camouflage vary considerably, the most common strategy is known as background matching (Michalis et al., 2017), where an organisms colouration and patterning resembles its typical habitat.Cryptic colouration can dramatically alter conspicuousness and is an essential predator avoidance strategy in many species (Stuart-Fox et al., 2003;Cheney et al., 2009;Cournoyer & Cohen., 2011), particularly for sessile organisms that cannot rely on evasion.Many cryptic species exhibit polymorphic variations in their colouration, that can be selected for in spatially and temporally complex environments (Duarte et al., 2018).Given their selective disparity, the maintenance of varied colour morphs within a population is thought to be a complex phenomenon (Karpestam et al., 2016).Alongside stochastic processes such as genetic drift, it is thought small scale environmental heterogeneity is predominantly responsible, where particular colourations are more resistant to thermal extremes or better able to background match and reduce conspicuousness to predators (Johannesson & Ekendahl, 2002;Phifer-Rixey et al., 2008).
The potential for broad spectrum lighting to impact the conspicuousness of camouflaged prey is clear.Such impacts may alter the balance of predator-prey interactions, population dynamics and the genetic structure of polymorphic populations.Its effect on the conspicuousness of camouflaged prey by predators at night has been little quantified (although see Briolat et al. 2021).Here, we provide evidence that a transition towards broad spectrum lighting can improve a predator's ability to discriminate prey species against a natural background.Our analysis spans three contrasting predator visual systems in the intertidal environment, with predation occurring both in air and in water accounting for the interaction of inherent optical water properties with the spectral composition of the artificial light field.brown are most common) that help them reduce conspicuousness to predators against the fucoid macroalgae on which they live (Crothers 2012).These snails are intertidal grazers of this macroalgae, and are more active during the night when the risks of dessication and predation are at their lowest.We selected three common predator models in temperate intertidal ecosystems that represented an array of differing predation modes and visual systems (Table 1).The herring gull (Larus argentatus) is a diurnal predator foraging for intertidal gastropods primarily in air and can discriminate complex colours using tetrachromatic vision (Crescitelli, 1958;Liebman, cited in Hart, 2001;Hart, 2001;Ödeen & Håstad, 2003).ALAN has demonstrated impacts on avian activity rhythms (Dominoni 2015) and foraging strategies (Santos et al. 2010;Dwyer et al. 2013) that make nocturnal predation of L. argentatus under man-made light sources possible.The common blenny (Lipophrys pholis) and green shore crab (Carcinus maenas) were selected as in water predators.L. pholis is a mostly diurnal predator and a trichromat capable of complex colour discrimination (Loew & Lythgoe, 1978).nocturnal predation by fish in response to ALAN exposure is well documented (Becker et al. 2013;Bolton et al. 2017).C. maenas is a mostly nocturnal predator (Silva et al. 2010) and a dichromat less able to discriminate a broad range of colours from a background (Martin & Mote, 1982).

Data Acquisition & Initial Processing
The receptor noise model established by Vorobyev and Osario (1998) was used to determine the discernibility of prey against their background by a number of predator species.This model relies upon three key parameters: 1) the reflectance spectra of prey species and the background on which they typically reside; 2) the spectral sensitivities of each photoreceptor possessed by a predator; and 3) the irradiance spectrum of light striking the prey individual and the background against which it is camouflaged.Sixty seven L. fabalis and L. obtusata individuals were collected from the fucoid macroalgae, Fucus vesiculosus, using fifteen 30cm quadrats in May 2020, along the mid-tide gullies of the Portwrinkle section of Whitsand Bay 50°21'N, 4°18'W, South West U.K. Both species are most commonly found on F. vesiculosus, however can occur on other species including Fucus serratus and Ascophyllum nodosum.Each group of Littorina were divided into pots based on the quadrat they were sampled from.
Hyperspectral reflectance spectra were quantified ex situ in sunlight using an Ocean Insight OCEAN-HDX-XR spectrometer with a wavelength response from 200-1100nm, fitted with a 3m long 1000μm fibre optic probe.The spectrometer was calibrated before each pot was measured using a WS-1-SL Spectralon® Diffuse Reflectance Standard.
Measurements were taken at the top of their shell along the last whirl, holding the fibreoptic probe at a 5mm distance above each individual and pointing down.Shells were air dried prior to measurement reducing specular reflection.Two measurements were also taken from the frond and vesicle of the seaweed F. vesiculosus, which were averaged to create a single, representative background reflectance spectrum.F. vesiculosus was selected as a model background as Littorina species are known to favourably reside upon fucoid macroalgae where they can employ cryptic background matching (Johannesson & Ekendahl, 2002).All reflectance spectra were standardised to a 1nm resolution through averaging, and readings outside of the 350nm-750nm range were omitted.The averaged reflectance spectra for the three colour morphs of littorinid snail and background algae are given in Figure 1C.Ethical approval was not required as no animals were removed from their native environment and no invasive, stressful or harmful procedures were performed.
To determine different colour morphs, Littorina were classified visually into Brown, Olive and Citrine (Yellow) classifications using a colour scheme presented by Rolán-Alvarez et al., (2012), as no orange specimens were found (n = 35 Brown, n = 15 Olive, n = 17 Yellow).The number of individuals per morph allowed replication for the receptor noise model and statistical analysis.These qualitative classifications were validated statistically using Multivariate Analysis of Variance performed on a Bray-Curtis dissimilarity matrix calculated from the raw reflectance data using CRAN: Vegan (Oksanen et al., 2007) in R v3.6.1 (R Core Team, 2020).Prior to use in the receptor noise model, the raw Littorina reflectance spectra were smoothed by a parameter of 0.2 using the 'procspec' function of the R package 'pavo 2' (Maia et al., 2019), to remove unwanted electrical noise.

Modelling Predator Visual Systems
An extensive literature search was carried out to locate each predator's lambda max (λmax) values, the wavelength at which each photoreceptor maximally absorbs light (Table 1).We were unable to source spectral sensitivity data measured specifically from the Herring Gull.Where spectral sensitivities for UV sensitive (UVS) avian species have been unavailable in the past, many studies utilise the sensitivities of the blue tit (Cyanistes caeruleus) as a model for an average UVS bird (Håstad et al., 2005;Avilés, 2008).To reinforce the validity of the herring gull results in our study, the majority of its photoreceptor absorbance curves are derived from published sensitivities from the Laridae family (Crescitelli, 1958 Measurements were collected from urban lighting installations around Cornwall, U.K. at ground level to accurately record the irradiance that animals are exposed to.It was assumed fish and crab predators viewed Littorina while submerged.To account for the different attenuations of artificial light wavelengths in seawater, irradiance spectra for their models were obtained using the HYDROLIGHT radiative transfer numerical model to simulate the passage of light from each source through 3m of water (i.e.3m depth) with a chlorophyll concentration of 0.3 mg m -3 .HYDROLIGHT output ranged between 400nm-700nm, with values between 350nm-400nm and 700nm-750nm set to zero.In air and in water irradiance spectra for each light source are given in Figure 1 A and B respectively.

Visual Modelling
The visual modelling section of the experiment was carried out using CRAN: pavo 2 (Maia et al., 2019).
The spectral absorbance curves of the photoreceptors in the eyes of each predator were modelled from their λmax values using the standard visual pigment template of Govardovskii et al. (2000) and Hart & Vorobyev (2005).For the herring gull, this function required the input of λcut, Bmid and ocular media transmission data, owing to their more complex visual system involving cone oil droplets.λcut values were estimated using the average of all available avian values from Hart & Vorobyev (2005).Pavo 2's standard ocular media transmission for avian visual systems, "bird" (Hart et al., 2005), was also used.In the absence of Bmid data, the oiltype argument was used to calculate Bmid using regression equations from Hart & Vorobyev (2005).
Quantum catch values for each photoreceptor were then calculated by using the vismodel function which integrates the spectral absorbance curves with the reflectance of the prey subject and its background, and the hyperspectral irradiance of the lighting technology being tested.Quantum catch refers to the proportion of photons that are captured by each receptors photopigment when viewing a subject.A total of 36 outputs were created, to obtain data for the three polymorphs as perceived by the three predators under the four lighting conditions.As in previous studies on colour discrimination, a von Kries adaptation coefficient was applied to each visual model to account for colour constancy in different lighting conditions (Siddiqi et al., 2004;Cournoyer & Cohen, 2011).The averaged background reflectance spectra of Fucus vesiculosus and each lighting technologies irradiance data were also included in this calculation.Each visual model's relative argument was set to FALSE to obtain raw photon catches that are suitable for use in pavo 2's coldist function (Maia et al., 2019).
For all 36 vismodel outputs, Euclidean colour distances (ΔS) were calculated in units of Just Noticeable Difference (JND) between prey and background quantum catches using the coldist function.JND values greater than 1 approximate the minimum level at which a single (prey) can be perceived (Cournoyer & Cohen, 2011; Bitton, 2019) with higher values indicating a stronger contrast between the prey and their natural background.To obtain colour distances, photoreceptor densities must be input and quantum catches must be weighted against the Weber fraction (noise-to-signal ratio) of the cones.It was assumed the herring gull and common blenny have a Weber fraction of 0.1 and 0.05 respectively, based on known avian and fish values (Olsson et al., 2017).For the crab, we have followed widely used protocols for unavailable data and used a Weber fraction of 0.05 (Matz et al., 2006;Cournoyer & Cohen, 2011;Bitton et al., 2019) as median estimate of published data that range between 0.02 in humans to 0.1 in some birds (Matz et al., 2006).For the herring gull's photoreceptor proportions, we used values that represent an average UVS bird (1:2:2:4) utilised by Seymoure et al. (2019) in a similar experiment.This is an accurate estimation as gull species are known to have a high proportion of long wavelength sensitive (LWS) cones (Hart, 2001).The common blenny's proportions were based on those typically seen in diurnal percomorphs (1:2:2), with a single cone surrounded by four double cones (Ali & Anctil, 1976;White et al., 2004).Due to unavailable data, the shore crab's proportions were set to 1:1, maximising its ability for colour discrimination (Lettieri et al., 2009).While this approximation may affect the magnitude of absolute values obtained from the model, the relationship between them will be maintained (Cheney et al., 2009;Lettieri et al., 2009), meaning that the relationships and contrasts between light types and colour morphs within each predator modelled in our study will still be valid.We cannot, however, make statistical comparisons on the effect of artificial lighting between the predators.Neural values were calculated using the noise argument as described by previous artificial lighting experiments (Ronald et al., 2017), indicating bright conditions and a high photoreceptor saturation.

Statistical Analysis
Exceptionally low JND values obtained for LPS in comparative to other lighting technologies provided a highly skewed response variable distribution that did not conform to normality even following log transformation.JND response values were instead investigated using generalised linear models fitted with a gamma error distribution.A two-way analysis of variance was performed on each predator's JND response values to quantify whether the four artificial light sources significantly impacted the conspicuousness of each of the three Littorina colour morphs.Pairwise contrasts were performed using the emmeans package's (Lenth et al., 2019) 'contrast' function to determine significant differences in colour distance between each light source and colour morph's ΔS values.The Tukey method was applied as a P value adjustment to control for inflated type II errors when performing a modest number of multiple tests.

Results
The classification of Littorinid snail colour morphs into Brown, Olive and Yellow was validated using a multivariate analysis of variance performed on a Bray-Curtis dissimilarity matrix calculated from the raw reflectance data of each individual (MANOVA: F 2,64 = 35, P < 0.001) (Supplementary Figure 1).While this validated our classification, a clear distinction can be made between the reflectance spectra of yellow from other colour morphs (Figure 1C).Olive and brown morphs exhibited similar reflectance spectra (Figure 1C) and displayed no clear clustering in the MDS ordination (Figure S1) suggesting that these may actually be one variable 'dark morph'.An extensive review of currently proposed classification systems is beyond the scope of this paper, hence our analysis is based on the classification of Rolán-Alvarez et al., (2012).Further reflectance data across multiple shores is needed before an informed appraisal of current classification systems can be made.
The ability of all three predators, to discriminate the three colour morphs against a fucoid algae background was significantly different depending on which light source was used (Herring Gull: Gamma GLM, χ 2 6,256 = 0.063, P = <0.001;Common Blenny: Gamma GLM, χ 2 6,256 = 0.1472, P = <0.001;Green Shore Crab, Gamma GLM, χ 2 6,256 = 0.5669, P = <0.001).Pairwise comparisons of the conspicuousness of the colour morphs are presented by predator for each artificial and natural light source in Supplementary Tables 1-3, summarised in Table 2 and presented visually in Figure 2.
The JND values of all three prey morphs remained below the minimum threshold of detectability (1) under LPS lighting (Figure 2).As such all three predators are unlikely to be able to differentiate any colour morph from the fucoid algae background when illuminated with LPS lighting, rendering any statistical differences in JND ecologically meaningless.The threshold of detectability was exceeded to varying degrees under the broader spectrum (HPS, LED and MH) light sources, sunlight and the full moon.The shift to broader spectrum (MH, HPS and LED) lighting however, increases the conspicuousness of some colour morphs more than others, depending the predator (Figure 2).When illuminated by LED, MH, the sun or the moon, yellow colour morphs were significantly more conspicuous to herring gulls (Figures 2A and 3A, Table 2, Supplementary Table 1) and shore crabs (Figure 2C and 3C, Table 2, Supplementary Table 3) compared to brown and olive morphs.This was also the case when illuminated by HPS lighting, except brown morphs were also more conspicuous than olive.
The switch to broad spectrum lighting had a lesser impact on the conspicuousness of the three colour morphs to the common blenny (Figures 2B and 3B, Table 2, Supplementary Table 2).In sunlight, yellow colour morphs were most conspicuous, while in moonlight and LED light, yellow and brown colour morphs were equally more conspicuous than olive.When illuminated by HPS lighting, brown morphs were more conspicuous than olive, but not yellow, and yellow morphs were equally as conspicuous as olive.When illuminated with MH lighting, brown colour morphs were significantly more conspicuous than yellow but not olive morphs, while olive and yellow morphs were equally as conspicuous.

Discussion
While ALAN is now well documented to increase predation pressure on prey populations (Frank 1988 (Briolat et al. 2021).The results of this study indicate that broader spectrum lighting technologies (HPS, LED and MH) increase the conspicuousness of prey species at night by reducing the efficacy of cryptic background matching when compared to narrow spectrum lighting.This may have profound implications for the fitness of cryptic species that rely on camouflage for their survival (Coker et al., 2009;Imperio et al., 2013).It should be noted however, that while prey species may be more conspicuous under broad spectrum lighting, conspicuousness does not necessarily scale linearly with colour distance (Santiago et al. 2020).Further behavioural research is needed to verify the suprathresholds of JND at which prey items become conspicuous to predators, however these were beyond the scope of this initial research.Nonetheless, littorinid prey remained under the threshold of detectability when illuminated by LPS lighting at night, and above this threshold when illuminated by modern broad spectrum lighting indicating that they have become detectable to predators at night where LPS lighting has been replaced.The magnitude of broad spectrum lighting's effect on the conspicuousness of prey was largely dependent on the colour morph being perceived.While some variability was observed, Yellow Littorina were most commonly more affected by broader spectrum lighting sources (HPS, LED, MH), likely owing to the greater distinction between their spectral reflectance and that of the Fucus vesiculosus background.This suggests that polymorphic colour variations that do not employ background matching techniques may be selectively preyed upon when illuminated by broad spectrum light, leading to altered population structure.Broad spectrum ALAN could therefore have impacts on the structure of polymorphic populations similar those seen on the peppered moth (Biston betularia) in the UK during the early 20 th century (Cook 2003).This would lead to greater homogeneity in polymorphic populations affected by broad spectrum ALAN, where more conspicuous colourations have been extirpated through enhanced predation or forced to migrate to habitats better suited for crypsis.In all predators studied, a shift from LPS to broader light types (HPS, LED, MH) increased the ability to perceive prey.This is likely because the broader spectral composition stimulates the multiple photoreceptors of predators (Davies et al., 2013), enhancing colour discrimination through visual opponent mechanisms that rely on the differences between receptor signals (Vorobyev & Brandt, 1997;Cournoyer & Cohen, 2011).While each of the broad light sources provoked a largely similar response in most cases, some notable differences were found between predator responses under different lighting technologies.It is likely that these differences would be more prominent if a broader selection of predator species were studied, given the diverse range of photoreceptor sensitivities that can be exhibited.For instance, visually guided behaviours in predators with spectral sensitivities that extend further into shorter UV wavelengths such as lizards, arachnids and reptiles will likely be most affected by MH technologies that can emit light in the UV range (Davies et al., 2013).The short wavelength, blue peak in LED lighting will also be more likely than other light types to affect marine organisms as it can penetrate further into the ocean (Davies et al., 2014).
The impact of broad spectrum lighting on conspicuousness is also variable between receivers.When viewed by the common blenny for example, the relative conspicuousness of yellow colour morphs was not as impacted by broad spectrum lighting compared to the herring gull and shore crab.This is likely because the photoreceptors of the common blenny are more tightly clustered and centred on the green portion of spectrum (Figure 1E).When attempting to interpolate the real-world outcomes of visual models, multiple predators, and their relative impacts on prey populations need to be accounted for.
The potential ecological repercussions that arise from the proliferation of modern broad spectrum lighting have been discussed at length (Gaston et al., 2012;Davies et al., 2013;Davies & Smyth, 2018), many of which arise from the facilitation of visually guided behaviours previously limited to the day (Davies et al., 2013).A variety of mitigation methods are available for planners and environmental managers when considering the ecological impacts of ALAN.These include reducing the amount of light used, shielding lights to prevent spill into the surrounding environment, part night lighting during times of peak demand, and manipulating the spectra of lighting to minimise ecological impacts (Gaston et al. 2012).Given that broad spectrum facilitates colour discrimination by predators and consequently increases the conspicuousness of prey, it is intuitive to suggest using narrow spectrum lighting to avoid these impacts.In the absence of colour however, nocturnal predators will use luminance contrast perception.We suggest a review of the colour vision systems of nocturnal predators in a given ecosystem should be undertaken to identify those wavelengths of light that minimise luminance contrast perception of prey items against backgrounds.Managers should remain aware however, that the impacts of ALAN extend beyond those on camouflage to impact all aspects of organism biology, and that all parts of the visual spectrum will likely have some ecological impact (Davies & Smyth 2018).
This study has demonstrated that broad spectrum artificial lighting has the potential to increase the conspicuousness of camouflaged prey species at night and leave colour variations with less effective background matching at greater risk of predation.If selective predation of colour morphs is sufficiently affected by the proliferation of LED lighting (Kyba et al., 2017), this could reduce prey populations and alter the genetic structure of naturally polymorphic populations.

Table 1:
The λmax values used to model the spectral sensitivities of the herring gull, common blenny and green shore crab.
Using a well established photoreceptor noise-limited chromatic discrimination model(Vorobyev & Osario, 1998), we determine the conspicuousness of three statistically distinct colour morphs of Littorinid snail (Littorina obtusata and Littorina fabalis) illuminated by 20 th century narrow spectrum lighting (Low Pressure Sodium (LPS)], and modern broad spectrum lighting [High Pressure Sodium (HPS); Light Emitting Diodes (LEDs); and Metal Halide (MH)] as viewed by three different predators.Solar and lunar irradiances were also included in the model as natural reference points.This modelling approach has been used extensively to quantify the perceptibility of camouflaged prey species (Stuart-Fox et al., 2005; Cournoyer & Cohen, 2011; Marshall et al., 2015) and removes the risk of extraneous variables affecting predation that could arise experimentally.L. obtusata and L.fabalis are found commonly on fucoid macroalgae (Fucus serratus, Fucus vesiculosus and Ascophyllum nodosum) throughout the UK intertidal environment and exhibit a range of colour polymorphisms (yellow, olive and ; Liebman, cited in Hart, 2001; Hart, 2001; Ödeen & Håstad, 2003).Therefore, our herring gull visual model represents the best possible approximation.The modelled absorbance spectra of the photoreceptors in the eyes of each model predator are given in Figure 1 D-F.Hyperspectral irradiance measurements previously collected by Davies et al. (2013) at a 1nm resolution between 350-750nm (MAYA2000 Pro) were used to represent the environmental light spectrum under each lighting technology (LPS, HPS, LED, MH).Conspicuousness was also modelled under sunlight and moonlight to provide natural light sources for comparison.Sea surface solar irradiances were collected from the L4 buoy of the Western Channel Observatory (50.250°N; 4.217°'W) at midday on June 24 th 2014 under clear sky conditions using an Satlantic Hyperspectral Radiometer.Lunar irradiances were downloaded from (http://www.olino.org/blog/us/articles/2015/10/05/spectrum-of-moon-light).Measurements were made using a SpecBos 1211 spectroradiometer (51.424°N, 5.409°E) and collected during a clear full moon night on the 14 th April 2014.The street lighting technologies represent an assortment of artificial light sources that were used in the 20th and 21st century and each possess a unique spectral composition, with LPS lighting typically emitting narrow spectrum irradiance at 590nm (Davies et al., 2014) and HPS, LED and MH emitting across a broader spectral range.HPS emits yellow/orange light similar to LPS although across a broader spectrum.LED lighting typically has wavelength peaks in the blue and green range (Elvidge et al., 2010), while MH is able to emit light within the UV range (Davies et al., 2013).
Similar trends have been documented in a variety of species in response to habitat changes brought on by climate change (Roulin 2014; Delhey & Peters 2017; Jones et al. 2020).This may also have a deleterious effect on species that exhibit garish colouration for sexual display at the expense of crypsis (Keren-Rotem et al., 2016), further exacerbating population decline by increasing the predation risk of viable mates.

Figure 1 .
Figure 1.Spectra used to parametrize visual modelling of the conspicuous of littorinid snail colour morphs to three visual predators.A. In air relative irradiances used in models for the Herring Gull (L.argentatus).Irradiances have been scaled to between 0 and 1 to facilitate comparison of contrasting spectral compositions.B. Model in water relative irradiances (3m depth) used to parametrize models for the common blenny (L.pholis) and shore crab (C.maenas).C. Averaged reflectance specrum collected from yellow, olive and brown morphs of L. obtusata and L. fabalis.The averaged reflectance spectrum for the fucoid algae background is given as a dashed black line.D-F.The modelled spectral absorbance curves of the photoreceptors in the eyes of each visual predator.

Figure 2 :
Figure 2: The impact of four alternative lighting technologies on the conspicuousness of three different colour morphs of intertidal littorinid snail to three predators with contrasting colour vision systems.Plot is derived from colour distance data indicating the chromatic contrast between Littorina and its natural background, as viewed by a predator.Bars represent model mean values, error bars represent 95% confidence limits.Grey dots represent raw Littorina colour distance values.Numbers in bold indicate significant differences between the effects of each light type at the 95% confidence level, where numbers differ within each colour morph grouping (see Supplementary Tables 1-3 for results of pairwise contrasts).The dashed line indicates 1 JND, the minimum threshold of detectability.Where these numbers are shared within a colour morph group, no significant difference can be inferred.

Figure 3 .
Figure 3.The impact of various light sources on the colour distances between camouflaged prey and their background by intertidal predators.Colour distances between yellow, olive and brown colour morphs of L. obtusata and L. fabalis as perceived by the tetrachromatic herring gull (L.argentatus, A,D,G,J,M,P), the trichromatic common blenny (L.pholis, B,E,H,K,N,Q) and the dichromatic shore crab (C.maenas, C,F,I,L,O,R) under Low Pressure Sodium (A-C), High Pressure Sodium (D-F), LED (G-I), MH (J-L) outdoor lighting technologies.Colour distances between different morphs illuminated by the Sun (M,N,O) and Moon (P,Q,R) are also provided.Red points represent the fucoid algae background.
; Becker et al. 2012; Bolton et al. 2017; Underwood, Davies & Queirós 2017; Bennie et al. 2018), few studies have so far evaluated its potential to inhibit cryptic background matching by camouflaged prey

Table 2 .
The impact of contrasting lighting sources on the comparative conspicuousness of yellow (Y), Brown (B) and Olive (O) colour morphs of intertidal littorinid snail (L.obtusata/L.fabalis)tothree predators with contrasting visual systems that hunt in air or water.Summarised from Supplementary Tables3-6.