Maltese Coastline Never Sleeps: The Effects of Artificial Light at Night (ALAN) on the Local Infralittoral Assemblages—A Case Study

: Aside from the most notorious threats, the Mediterranean Sea faces novel and poorly explored impacts from artificial light at night (ALAN), which influences natural light–dark cycles and affects marine ecosystems. This study investigates the impact of ALAN on coastal infralittoral assemblages in Malta, where such effects remain unexplored. Using Baited Remote Underwater Videos (BRUVs)


Introduction
The Mediterranean Sea represents a biodiversity hotspot of global importance and it faces a spectrum of anthropogenic threats [1].One of these is the increase in global light emissions, which necessitates evaluating the impacts on the physiology, behaviour, and ecology of organisms [2].Many species have daily cycles or scototaxic behaviours that can be disrupted by changes in natural lighting [3].Artificial Light at Night (ALAN) alters natural light-dark cycles and it is recognised as a threat to biodiversity due to its effects on animal behaviour, migration, reproduction, and biological interactions [4].ALAN disrupts activities such as foraging, hiding, and resting, affecting predator-prey dynamics [5,6] and reducing the availability of darkness for nocturnal species [7].As human development in coastal zones increases, organisms in shallow coastal or intertidal systems face growing risks from ALAN [3].The diminishing availability of naturally dark spaces also threatens species that rely on darkness for rest and recovery, potentially leading to a loss of refuge for these species [8].Moreover, exposure to ALAN might affect key species resulting in potential cascade effects to the community with case studies like Paracentrotus lividus [9], Caretta caretta [10], and Lymnaea stagnalis [11].
Predicting ecosystem reactions to environmental change is crucial for scientific and political agendas [12].However, single-site studies struggle to generate broad predictions due to context-specific responses [13,14].While ALAN research has focused mainly on terrestrial ecosystems, studies on its effects on marine ecosystems are still limited [15].Experimental networks now apply uniform treatments across multiple sites to address this gap, providing generalisable data on environmental changes [7].Current EU policies lack explicit considerations for ALAN's impact on fish, focusing instead on the broader marine conservation issues.
Peregrym et al. [16] highlighted how Mediterranean islands' pollution from ALAN is directly correlated with the population density.This is also the case of the Maltese archipelago that, according to Caruana et al. [17], results in one of the most polluted areas of the Mediterranean Sea.Nevertheless, significant gaps in knowledge persist regarding the impact of ALAN on biodiversity and ecosystems within the Maltese region.To our knowledge, the only study is the EU LIFE Yelkouan Shearwater project which highlighted ALAN's impact on Puffinus yelkouan, a seabird breeding on the Maltese north coast [18].The rapid increase in ALAN, due to inadequate planning, underscores the need for a comprehensive examination of its effects on marine biota in Maltese waters, which has not yet been studied in any of its ecological components.
Given the scarcity of empirical evidence in Maltese waters, this study aims to provide valuable insights from the study of the effects of ALAN on the coastal fish (but not limited) community.With the use of Baited Remote Underwater Videos (BRUVs), in this study we hypothesised that ALAN influences nocturnal fish activity, particularly predatory behaviour, thereby altering assemblage structure.This study offers initial empirical evidence of ALAN's impact on fish behaviour and assemblage structure in Maltese marine ecosystems.By analysing the relationship between artificial light exposure and marine communities, we support the idea to develop sustainable management strategies for the mitigation of the impact of ALAN in the Maltese archipelago.

Study Area
The Ċirkewwa harbour, located at coordinates 35.98333 • N latitude and 14.33333 • E longitude, lies along the northwest coast of Malta, marking the northernmost point of the island.Within the study area, two sites were chosen for the purposes of this study, which were denoted as Reef and Harbour (Figure 1).
Predicting ecosystem reactions to environmental change is crucial for scientific and political agendas [12].However, single-site studies struggle to generate broad predictions due to context-specific responses [13,14].While ALAN research has focused mainly on terrestrial ecosystems, studies on its effects on marine ecosystems are still limited [15].Experimental networks now apply uniform treatments across multiple sites to address this gap, providing generalisable data on environmental changes [7].Current EU policies lack explicit considerations for ALAN's impact on fish, focusing instead on the broader marine conservation issues.
Peregrym et al. [16] highlighted how Mediterranean islands' pollution from ALAN is directly correlated with the population density.This is also the case of the Maltese archipelago that, according to Caruana et al. [17], results in one of the most polluted areas of the Mediterranean Sea.Nevertheless, significant gaps in knowledge persist regarding the impact of ALAN on biodiversity and ecosystems within the Maltese region.To our knowledge, the only study is the EU LIFE Yelkouan Shearwater project which highlighted ALAN's impact on Puffinus yelkouan, a seabird breeding on the Maltese north coast [18].The rapid increase in ALAN, due to inadequate planning, underscores the need for a comprehensive examination of its effects on marine biota in Maltese waters, which has not yet been studied in any of its ecological components.
Given the scarcity of empirical evidence in Maltese waters, this study aims to provide valuable insights from the study of the effects of ALAN on the coastal fish (but not limited) community.With the use of Baited Remote Underwater Videos (BRUVs), in this study we hypothesised that ALAN influences nocturnal fish activity, particularly predatory behaviour, thereby altering assemblage structure.This study offers initial empirical evidence of ALAN's impact on fish behaviour and assemblage structure in Maltese marine ecosystems.By analysing the relationship between artificial light exposure and marine communities, we support the idea to develop sustainable management strategies for the mitigation of the impact of ALAN in the Maltese archipelago.

Study Area
The Ċirkewwa harbour, located at coordinates 35.98333°N latitude and 14.33333° E longitude, lies along the northwest coast of Malta, marking the northernmost point of the island.Within the study area, two sites were chosen for the purposes of this study, which were denoted as Reef and Harbour (Figure 1).The notable distinction between the sites predominantly resides in the influx of artificial light each site receives, given that the majority of other environmental variables (e.g., sea surface temperature, weather conditions,) are not expected to differ much across both study sites due to their proximity to one another.
In relation to light, the terms press and pulse disturbances are used.A press disturbance signifies a continuous exposure to light, as seen with the persistent illumination from sources like buildings or streetlights in the Harbour habitat.In contrast, a pulse light disturbance involves the external light exposure during the sampling activities.

Reef (Dark)
The selection of the site was based on the putative absence of artificial light whilst being characterised by the same habitat as the light-polluted harbour site, thereby enabling a comparative analysis (refer to Section 2.3 Harbour).The photometer reading at the deployment location (35.989253 • N, 14.327892 • E) within the site was 0 Lm.
The site is exposed to the prevailing north-westerly winds which occur on an average of 20.7% of all the days in a year [19].According to the Marine Strategy Framework Directive (MSFD) habitat classification, the operational site is classified as shallow infralittoral with its topography being denoted as coastal water [20].At the site, the prevailing seabed composition is a rocky reef substrate that features substantial rocky outcroppings of a horizontal orientation.From the collected data it was clear that the extensive rocky reefs support various coralligeneous and macroalgal species.Posidonia oceanica meadows were also present in the surrounding areas.

Harbour (Already-Present ALAN)
The site was chosen in an artificial light-polluted area along the north-easterly quay that serves as a departure point for the Blue Lagoon Ferry.The photometer reading at the deployment location (35.988169 • N, 14.329861 • E) within the harbour site was 80 Lm, going up to 145 Lm when charter boats pass by.
The site is exposed to north-easterly winds which show no dominance on the Maltese Islands [19].According to the Marine Strategy Framework Directive (MSFD) habitat classification, the operational site is classified as shallow infralittoral with its topography being denoted as coastal water [20].The prevailing seabed composition at the recording site was a rocky reef substrate, with some sparsely distributed sand pockets in adjacent regions.Within proximity to where the BRUV apparatus was placed, the terrain was also interspersed with sizeable rocks and debris.P. oceanica meadows were also present in close proximity to the operational monitoring site.
The harbour site presents an advantageous environment for examining the impact of artificial light on fish and other benthic organisms due to its exposure to multiple sources of persistent artificial illumination.The principal sources of artificial light at the treatment site include streetlights in proximity, illumination emitted by the Gozo Channel Ferry (crosses over to the sister island of Gozo) and charter boats (either stationed at the quay or traversing the area), and emissions from nearby dining establishments.The incidence of charter boats varied across sampling instances, with some sessions witnessing a heightened presence and others displaying a comparatively negligible presence.Moreover, the durations of charter boat presence exhibited notable fluctuations.

Field Sampling
A horizontal-facing single camera BRUV constructed in-house from easily accessible materials was employed for the study.The BRUV was stationary and placed on the seafloor during sampling.
The frame of the BRUV, which spans an area of approximately 355 mm by 400 mm, was constructed using Polyvinyl Chloride (PVC) piping and provided with 4 kg of weight to enhance the bottom stability.The bait arm measured a length of 1 m from the central point of the BRUV with an elevation of 260 mm above the seabed and accommodated a bait bag attached to the end of the arm.Two goose-neck arms for torch attachments was then affixed to the structure armed with two AL2600XWP-II (Black Molly V) torches (Bigblue Dive Lights, Clearwater, FL, USA) used during each replicate recording.Each torch has a maximum light output of 2600 Lm at 6500 K colour temperature.Moreover, the torches were also equipped with integrated red LEDs which were used as pseudo-CTRL conditions given the lowest visibility red light has on fish sight [21,22].The beam angle of the torches was set at 120 • .
The tray mount was equipped with a GoPro Hero 10 Action Video Camera TM (GoPro, San Mateo, CA, USA), which served as the primary means for capturing video sequences.Footages were recorded at a resolution of 1080 p, and a frame rate of 60 frames per second.
Prior to each deployment, 250 g of bait mix was enclosed in the bait bag.This included a mixture of chopped Bullet tuna (Auxis rochei) and European sprat (Sprattus sprattus) in a 4:1 ratio, respectively.Both fish species have a high oil content, which makes them ideal baits [23].

Sampling Design and Video Analysis
During August 2023, a total of eighteen video samples were collected, using three light treatments (High-White Light Intensity (H), Low-White Light Intensity (L), and Red Light (R)) across two Habitats (Harbour and Reef), with three replicates per treatment (Table 1).The specified light settings for each treatment were achieved using torches attached to the BRUV system (refer to Section 2.4).The torches emitted the light intensities defined in Table 1, ensuring consistent application of all treatments.Sampling involved collecting two samples per night with each being one-hour long: one from the harbour habitat and one from the reef habitat.Sampling began at 10:00 p.m., ensuring synchronisation with the circadian rhythm of fish.Photometer readings of each study area were measured to monitor ambient light levels.

Observation Protocols
Each video sequence was analysed at 0.4x playback speed for thorough assessment.Count-N was computed for each species, representing the cumulative count of individuals observed during specific monitoring periods, following methodologies from previous studies [24][25][26].Count-N emulates in situ slate-transect enumeration, accurately tallying individuals within digital transects.The MaxN approach was avoided due to its potential to misrepresent average fish abundance during hour-long recordings, particularly in cases of sporadic large schools.Count-N ensures comprehensive identification and counting of all fish within the video frame, and was thus applied consistently across all samples.
Viewing videos in greyscale during the red-light treatment enhanced pattern discernment and improved analysis quality.When precise species identification was unfeasible, a broader taxonomic classification, such as genus or family, was assigned.
Key environmental factors that may influence the results were also recorded for each sample.These included wind direction (degrees), wind speed (m/s), lunar phase (% illumination), and cloud coverage (%).Wind factors were examined for their potential impact on fish assemblage [27,28], while the lunar illumination and cloud coverage were scrutinised for their direct influence on light conditions and, consequently, fish assemblages [29,30].

Statistical Analysis
A raw data matrix was constructed using the Count-N values for each identified species per sample, which was then processed and analysed using the software Primer v7 and PERMANOVA+ [31].To balance contributions from species with varying frequencies, a dispersion-weighting routine was applied followed by a square root transformation.The transformed matrix of abundances was used to compute a Bray-Curtis Index into a triangular distance matrix [32].As diversity descriptors, the number of species (S) and the number of individuals (N) were extracted along with calculation of the Shannon Index (H).
A SIMPER analysis [33] identified species contributing to at least 70% of the assemblage within the Light Intensity and Habitat factors.
Significant differences were assessed using a PERMANOVA analysis within a twofactors design: Light Treatment (Red Light (R), Low-Intensity White Light (L), High-Intensity White Light (H)) and Habitat (Reef, Harbour).Significant differences between levels of the factors were explored through pair-wise testing.Non-metric Multidimensional Scaling (nMDS) was then employed to visualise the assemblage against the tested factors in an ordination plot.The BEST analysis [33] was then used to identify which environmental variables best explain the patterns observed in the biological data.
Principal Coordinates Ordination (PCO) [34] was used to visualise and understand community composition and dissimilarity patterns in response to different light and site treatments.Vector overlays were applied to visualise and compare the abundances of the dominant species in the dataset.

Descriptive Metrics
A total of 23,985 individuals belonging to eighteen species of fish, two species of mollusc, one species of cnidarian, and one species of polychaete were recorded and identified during this study from the recorded videos.Of the 23,985 reported individuals, 19,200 individuals belong to the species Boops boops (the full list of species is reported in Appendix A).

Impact of Light Intensity and Habitat on Community Composition
Differences within the full assemblage for the two factors, Light Intensity and Habitat, were explored with resulting significant differences for both factors (Table 2 and Figure 2).PERMANOVA results showed that Light Intensity (p = 0.0001) had the most significant impact on community composition, followed by Habitat (p = 0.0005) and their interaction (p = 0.0003), all strongly significant (p > 0.001).Pair-wise t-tests indicated significant differences under varying light conditions: Red vs. Low Light (t = 2.2347), Red vs. High Light (t = 3.5938), and Low vs. High Light (t = 2.267), with the lowest p-value (0.0022) signifying the strongest difference between Red-and High-Light conditions (Table 2).

Effects of Light Intensity on Biodiversity Metrics
Boxplots depicting the diversity metrics (S, N, and H) for Reef and Harbour habitats under the three different light treatments were computed (Figure 3) highlighting the effects of the levels of light conditions on community composition and biodiversity across the two habitats.In the reef habitat, light intensity appears to have a pronounced effect on individual abundances with low-light conditions attracting significantly more individuals than red-light and high-light conditions.When looking at the Harbour Habitat, we observed significantly higher levels of S and N when the samples were treated with red light.PERMANOVA results showed that Light Intensity (p = 0.0001) had the most significant impact on community composition, followed by Habitat (p = 0.0005) and their interaction (p = 0.0003), all strongly significant (p > 0.001).Pair-wise t-tests indicated significant differences under varying light conditions: Red vs. Low Light (t = 2.2347), Red vs. High Light (t = 3.5938), and Low vs. High Light (t = 2.267), with the lowest p-value (0.0022) signifying the strongest difference between Red-and High-Light conditions (Table 2).

Effects of Light Intensity on Biodiversity Metrics
Boxplots depicting the diversity metrics (S, N, and H) for Reef and Harbour habitats under the three different light treatments were computed (Figure 3) highlighting the effects of the levels of light conditions on community composition and biodiversity across the two habitats.In the reef habitat, light intensity appears to have a pronounced effect on individual abundances with low-light conditions attracting significantly more individuals than red-light and high-light conditions.When looking at the Harbour Habitat, we observed significantly higher levels of S and N when the samples were treated with red light.

PERMANOVA of Shannon Diversity Index
Nevertheless, from the PERMANOVA results of the H index (Table 3), non-significant differences were found both for the factors Habitat and Light (p = 0.2471).However, the significant p-value for their interaction (p = 0.0021) suggests that the combined effect of habitat and light conditions might have a noteworthy impact.

SIMPER Analysis
By looking at the SIMPER analysis results for the factors Light Intensity and Habitat at each of their treatments, we can observe which species had the greatest contribution to the composition of the assemblages (Tables 4 and 5).

PERMANOVA of Shannon Diversity Index
Nevertheless, from the PERMANOVA results of the H index (Table 3), non-significant differences were found both for the factors Habitat and Light (p = 0.2471).However, the significant p-value for their interaction (p = 0.0021) suggests that the combined effect of habitat and light conditions might have a noteworthy impact.

SIMPER Analysis
By looking at the SIMPER analysis results for the factors Light Intensity and Habitat at each of their treatments, we can observe which species had the greatest contribution to the composition of the assemblages (Tables 4 and 5).

BEST Analysis
The BEST analysis (Table 6) was also conducted to determine which environmental variables best explain the variation in biological data.Wind direction emerged as the most influential single factor, with a positive correlation of 0.103.This indicates a moderate explanatory power of the wind direction on assemblage composition.However, combinations of variables, including wind speed, lunar phase, and cloud coverage, resulted in negative correlations, with the highest negative correlation being −0.159 for all four variables combined (Table 6).This suggests that while wind direction alone has some significance, the combined influence of these environmental factors is more complex, potentially indicating non-linear or interactive effects that diminish their individual explanatory power.

Principal Coordinates Ordination (PCO) Analysis
From the results of the Principal Coordinates Ordination (PCO), highlighting the significance of each principal coordinate in explaining ecological variation, PCO1 and PCO2 captured the most significant portion of the variance (60.97%) (Table 7).From the interpretation of the two-dimensional plot (Figure 4), the x-axis representing PCO1, strongly correlates with light conditions, while the y-axis representing PCO2, appeared to be associated with habitat types.

PCO5
2280.9 5.62 85.22 From the interpretation of the two-dimensional plot (Figure 4), the x-axis represe ing PCO1, strongly correlates with light conditions, while the y-axis representing PCO appeared to be associated with habitat types.From the vectors overlay for the dominant species (Figure 4), as identified in the SI PER analysis, we observed a divergent behaviour of species.T. trachurus showed increas frequency with higher light intensity, particularly in the reef habitat.In contrast, the pr ence of A. imberbis and S. scriba seemed to be related, favoured by the presence of R Light regardless of the Habitat composition.The polychaete H. carunculata presen showed to be strongly correlated with the Habitat composition, favouring the more i pacted Harbour rather than the Reef.The species B. boops, while not showing appar signs of correlation with Light Intensity, seemed to prefer the Reef habitat, probably d to the greater exposure to the open sea.From the vectors overlay for the dominant species (Figure 4), as identified in the SIMPER analysis, we observed a divergent behaviour of species.T. trachurus showed increased frequency with higher light intensity, particularly in the reef habitat.In contrast, the presence of A. imberbis and S. scriba seemed to be related, favoured by the presence of Red Light regardless of the Habitat composition.The polychaete H. carunculata presence showed to be strongly correlated with the Habitat composition, favouring the more impacted Harbour rather than the Reef.The species B. boops, while not showing apparent signs of correlation with Light Intensity, seemed to prefer the Reef habitat, probably due to the greater exposure to the open sea.

ALAN as a Threat to Biodiversity: Evolution and Habituation
Artificial Light at Night (ALAN) threatens biodiversity by disrupting nocturnal behaviours and causing phenological changes in migratory species.Additionally, ALAN can create selection pressure that alters genetic compositions, favouring non-light-sensitive species and potentially leading to the loss of light-sensitive species and genotypes [35,36].While some species may adapt through habituation, where repeated exposure leads to diminished responsiveness, this can affect predator-prey dynamics [36,37].
In this study, ALAN's impact on fish behaviour was evident.B. boops was more abundant in red light at the site with existing ALAN compared to the dark site.Conversely, in low-intensity white light, B. boops was more abundant at the dark site.This suggests habituation to ALAN, reducing typical predator avoidance behaviours in low-light conditions.
The PCO analysis (Figure 4) supports this hypothesis, showing red-light samples clustering closely regardless of habitat type, indicating habituation to constant light disturbance.Low-and high-light samples displayed more variance, likely due to specific habitat conditions.This reduced impact of constant light disturbance highlights the role of habituation in moderating ALAN's effects on fish behaviour.

Pulse and Press Dynamics
In this study, the intensity of light had a more significant impact on fish and marine biota than the distinction between press and pulse disturbances.The PERMANOVA test indicated that Light Exposure was a more significant factor than Habitat in influencing community composition, nevertheless, the Habitat played a significant role in shaping the local community.This suggests that external disturbances, such as light intensity, might be more influential than habitat changes, although different communities might respond in contrasting ways as in the case of benthic river assemblages [38].
Notably, in red-light samples, there was a higher abundance of fish in the Harbour habitat with existing ALAN compared to the Reef habitat without ALAN.This indicates an attraction to constant light disturbance in the Harbour.This trend was supported by experiments with low-and high-light intensity disturbances, where certain species showed a preference for altered light conditions.These findings highlight fish preferences and behavioural shifts in response to light disturbances, contributing to the understanding of marine community interactions with artificial light.

External Factors
Additional habitat factors, such as seabed type and fishing activities paired with environmental variables, like wind speed and direction and lunar illumination, might have influenced the study's outcomes.Topographic factors such as relief, complexity, and morphology significantly influence fish assemblages, with variations in terrain attributes affecting community composition across different seascapes [39].Coastal urbanization and resource extraction can reduce fish diversity and abundance [40].Additionally, habitat type influences the circadian cycles of fish, with seagrass beds supporting nocturnal species and sandy seabeds harbouring diurnal species [41].This highlights the need to fully characterise the habitat and consider its role within the broader landscape to accurately assess its influence on the fish community [39].Environmental factors like wind speed, wind direction, and lunar illumination influenced sample variations [42].Wind speed showed low variability due to controlled sampling conditions, while wind direction seemed to have an effect on the underwater conditions and fish distribution, as per results of the BEST analysis.High lunar illumination without cloud cover introduced natural light, complicating the control of variables.The interplay of these factors highlights the complexity of isolating individual impacts on fish communities.

Species-Specific Light Effects
In contrast to natural daylight, ALAN pollution introduces significant disparities in light conditions, affecting nocturnal interactions among fish and the environment.Fish exploit these conditions for refuge and foraging, potentially altering predator-prey dynamics in coastal ecosystems [43].Responses to ALAN vary by species, influenced by their visual behaviours and environmental factors [44].
Predatory fish rely on visual cues to hunt, with success decreasing in lower light or higher turbidity [44].In this study, it was observed that artificial light from nearby infrastructure influences predatory behaviour, particularly in T. trachurus, which exhibited increased activity in well-lit areas [45].SIMPER analysis indicated that T. trachurus significantly contributed to the dissimilarity in high-intensity white-light conditions.They were present in every high-intensity white-light sample, absent in red-light samples, and present in only two low-intensity white-light samples.Such a result contrasts with another study where it was observed how T. trachurus would hide from intense white light [46].
Becker et al. [43] noted predatory fish maintaining positions in illuminated areas, a behaviour also observed in T. trachurus, indicating a net energy gain despite higher metabolic costs [45,46].In this study, T. trachurus engaged in 'station holding', actively swimming against the current to stay in well-lit areas to prey, demonstrating a distinct preference for high-intensity light conditions.Bolton et al. [5] reported a markedly higher occurrence of fish predation on the sessile assemblages under artificial lighting conditions compared to the naturally dark nocturnal conditions.Owing to the currents, there was always a bait trail and potentially some spillage from the BRUV's bait bag, in which the fish exhibited interest.However, the study revealed that it was specifically during illuminated nights that the fish significantly exploited the bait bag, as was already described, particularly by T. Trachurus [47].
In the case of Muraena helena, its presence was not uniform across all samples.It exhibited a relatively even distribution between red-light and low-light samples while it was less frequent in high-light intensity environments, especially within the habitat with no pre-occurring ALAN.In instances where this species was observed, it consistently attempted to tear the bait bag, displaying this behaviour irrespective of the prevailing light conditions [48].Notably, in the site with existing ALAN, it was concluded that the BRUV apparatus was positioned in proximity to the den of a solitary M. helena.The frequent sightings in this habitat could potentially be attributed to this single individual.Considering the occasional fishing activities in the area, it is plausible that this individual might have been captured towards the end of the sampling period, explaining its absence thereafter.
The behaviour of A. imberbis contrasts sharply with T. trachurus, a nocturnal fish that forages at night and shelters in dimly lit caves or crevices during the day [49].A. imberbis shows social flexibility as it can be spotted in solitary individuals as well as forming numerous groups or swarms, possibly to minimise predator encounters [50].This behaviour likely contributes to their infrequent presence in high-light intensity environments, as they favour red light and avoid well-lit areas to reduce predation risk [50].
The European common cuttlefish (Sepia officinalis) preferred darker environments, being observed exclusively under red light.Only two individuals were recorded, both in the Harbour habitat.These finding seems consistent with Sykes et al. [51], which found that low-light intensity (100 lux) fostered better growth and survival rates for S. officinalis.The behaviour of station holding suggests that the Harbour habitat provided optimal conditions and a net energy gain for this species.
Rosa et al. [52] found that H. carunculata prefers lower light levels, with increased activity and coral predation at night and during twilight.This study supports these findings, recording 55 specimens of H. carunculata under red light, 49 under low-light intensity, and only 24 under high-light intensity.The consistency of these results with Rosa et al. [52] underscores the bearded fireworm's inclination towards lower light conditions.
Concerning small shoaling fish (<100 mm) such as B. boops and C. chromis, Becker et al. [43] observed higher abundances during nights with floodlights due to improved foraging conditions from plankton attracted to light.From the recorded footage, it was noticed that B. boops and C. chromis were shoaling in all light conditions, however, they were slightly more abundant in the red and low light conditions.On the other hand, from the footage captured in the high-light intensity samples, a notable observation was that B. boops, in particular, avoided approaching the camera apparatus because of the T. trachurus engaged in station holding for an extended period during the sampling.Consequently, B. boops were counted when they passed in the background, yet they were less conspicuous since they were further away from the camera and positioned in an area with reduced light.It is important to note that the possibility of schooling in greater densities by B. boops may have therefore been overlooked due to these factors.

Recommendations for ALAN Mitigation
To mitigate the effects of ALAN on coastal marine environments, a variety of strategies should be adopted.It is essential for future coastal developments to conduct thorough evaluations of ALAN's ecological consequences, ensuring that lighting designs are tailored to lessen adverse impacts on marine organisms.Effective protective steps include utilising full cut-off fixtures that aim light downward, thereby reducing light pollution and achieving a 0% Upward Light Output Ratio (ULOR) [53].Additionally, light intensity should be limited to what is strictly necessary for safety and operational purposes [54].Lighting should be strategically focused on specific areas, preventing unnecessary overflow into ecologically fragile regions, nearby residences, or the sky.Where possible, employing lighting schedules to dim or switch off lights during periods of inactivity can markedly decrease energy use and light pollution.To mitigate the disruptive effects of blue light on marine organisms, it is crucial to select lighting with a Correlated Color Temperature (CCT) below 3000 K.Moreover, reducing reflected light in coastal projects by the use of non-reflective, dark-colored materials can help mitigate the ecological effects of ALAN [53].

Conclusions
The study's key findings indicate that some habituation occurred in response to press light disturbances, as seen in B. boops, which did not exhibit typical schooling behaviour under low-light intensity in the harbour.Predatory species, particularly T. trachurus, strategically exploited high-light conditions, demonstrating station holding behaviour, primarily under high-intensity white light, due to net energy gain despite increased metabolic costs.Species at lower trophic levels, such as A. imberbis, declined in abundance as light intensity increased.S. officinalis and H. carunculata showed a preference for darker environments.Additionally, fish were more abundant in the Harbour habitat with existing ALAN under red light, suggesting an attraction to press light disturbances in these environments.
As expanding illuminated zones encroach upon natural darkness, elevated levels of ALAN may significantly alter marine ecosystems.Understanding ALAN's impact on marine life can guide protective measures, define thresholds, and optimise lighting strategies to mitigate effects.This study, although pioneering in the Maltese archipelago, provides insights to future coastal developments.While further investigation should take place, ALAN's impact on fish communities is supported worldwide, therefore we suggest planning new coastal infrastructures with a keen eye to light reduction at night, where possible, so as to minimise potential impacts on the coastal environments.

Figure 1 .
Figure 1.Map of the study area with sites used during this exercise as part of the Habitat factor labelled as: Reef and Harbour.Figure 1. Map of the study area with sites used during this exercise as part of the Habitat factor labelled as: Reef and Harbour.

Figure 1 .
Figure 1.Map of the study area with sites used during this exercise as part of the Habitat factor labelled as: Reef and Harbour.Figure 1. Map of the study area with sites used during this exercise as part of the Habitat factor labelled as: Reef and Harbour.

Figure 2 .
Figure 2. Non-metric Multidimensional Scaling plot for all samples denoted by light treatment (R = Red Light, L = Low-Light Intensity, H = High-Light Intensity) and Habitat labels.

Figure 2 .
Figure 2. Non-metric Multidimensional Scaling plot for all samples denoted by light treatment (R = Red Light, L = Low-Light Intensity, H = High-Light Intensity) and Habitat labels.

Figure 3 .
Figure 3. Boxplots illustrating the diversity metrics of species richness (S), total abundance (N), and Shannon diversity index (H') for Reef and Harbour habitats under three different light treatments: Red Light (R), Low-Light Intensity (L) and High-Light Intensity (H).

Figure 3 .
Figure 3. Boxplots illustrating the diversity metrics of species richness (S), total abundance (N), and Shannon diversity index (H') for Reef and Harbour habitats under three different light treatments: Red Light (R), Low-Light Intensity (L) and High-Light Intensity (H).

Figure 4 .
Figure 4. Principal Coordinates Ordination for all samples denoted by light treatment (R = R Light, L = Low-Light Intensity, H = High-Light Intensity) and Habitat labels.Vectors overlay for dominant species, as per the results of the SIMPER analysis, are shown.

Figure 4 .
Figure 4. Principal Coordinates Ordination for all samples denoted by light treatment (R = Red Light, L = Low-Light Intensity, H = High-Light Intensity) and Habitat labels.Vectors overlay for the dominant species, as per the results of the SIMPER analysis, are shown.

Table 1 .
Two-factors study design with the three levels of Light Intensity and the two levels of Habitat.

Table 2 .
PERMANOVA for factors Habitat and Light Intensity with pair-wise test between levels of Light Intensity.Level of significance is indicated (p < 0.01 **; p < 0.001 ***).

Table 3 .
Shannon index PERMANOVA for factors Habitat and Light Intensity.Level of significance is indicated (p < 0.05 *).

Table 3 .
Shannon index PERMANOVA for factors Habitat and Light Intensity.Level of significance is indicated (p < 0.05 *).

Table 4 .
SIMPER analysis for the factor Light Intensity within the levels Red Light, Low Light and High Light showing the dominant species, their average abundance, similarity, and individual contribution to all assemblages.

Table 5 .
SIMPER analysis for the factor Habitat within the levels Reef and Harbour showing the dominant species, their average abundance, similarity, and individual contribution to the overall assemblage.

Table 6 .
Results for the BEST analysis between biological and environmental datasets correlated via Spearman ranking.

Table 7 .
Principal Coordinates Ordination components indicating the Eigenvalue, their contribution in explaining the assemblage variance, and the cumulative contribution.