Research was conducted in Praia do Tofo in the Inhambane province on the southern Mozambican coast ( Figure 1 ) between July 2021 and March 2024. The study site, known as Giants Castle (-23.86°S, 35.55°E), lies within Tofo Bay and consists of a rocky reef wall at a depth of 26 – 35 metres below sea level, stretching approximately 120 metres. The reef wall runs from north to south with a steep slope descending along the eastern side and features several “cleaning station” regions, which are recognisable by their abundance of cleaner fish. The site experiences a predominant north-south current, and water visibility ranges from 1 to 25 metres depending on weather conditions, tidal cycles, and particulate matter in the water column. Giants Castle was chosen as the study site for this research due to the high frequency of elasmobranch sightings compared to other sites in the region (Marshall, 2008; Rohner et al., 2013; Venables et al., 2020; Keeping et al., 2021), which maximises the likelihood of capturing these species with the remote cameras. Giant’s Castle is also one of the more accessible diving areas from the boat launch site, allowing for frequent camera deployments.

Trial camera deployments were conducted to identify areas with the highest frequency of elasmobranch sightings. Two areas characterised by large schools of sea goldies (Pseudanthias squamipinnis) provided the most consistent sightings whilst facing east across the Giant Castle reef wall (hereafter cleaning station A and cleaning station B) (All Out Africa, unpubl. data) and were therefore selected for camera deployments in the present study ( Figure 2 ).

An Open Ocean Camera^© (OOCAM) (Clear Robotics Limited, 2021) was used to collect LL-RUV data for the study. The OOCAM was encased in an acrylic tube housing, equipped with temperature, pressure and light sensors, collecting environmental data every second during video recordings. It features a Sony IMX219 sensor with a resolution of 12.3 megapixels, a 400 Wh battery pack of approximately 148 hours recording capacity and a Raspberry Pi computer chip. The Raspberry Pi computer chip was set to have the camera record at three- to four-hour intervals between 06:00 and 18:00 at 1080p resolution, a frame rate of 15 frames per second (FPS) and a 90° field of view. The camera system was modified with a Nauticam vacuum check and leak detection system (Nauticam, 2023). To ensure camera stability and prevent equipment damage or loss from strong currents, the camera housing was attached to a metal frame and secured to a rock on the reef using ski rope ( Figure 3A ). While deployments in this study were conducted on rocky reef substrates, LL-RUV systems like the OOCAM can also be deployed on sandy or soft-bottom substrates if adequately weighed down to ensure stability. Care must be taken during reef deployments to avoid damaging coral structures. Researchers deployed the camera system at cleaning station B facing east across the narrow part of the reef using a lift bag. After each deployment, the camera was retrieved by the researchers and video data was downloaded for analysis.

In addition to the OOCAM, GoPro™ Hero 4 cameras were positioned at cleaning station A. GoPro™ cameras were mounted in a cement block, tied to a rock on the reef using ski rope to prevent camera loss ( Figure 3B ). GoPro™ cameras were set to record at 1080p resolution, 60 FPS and offered a 170° field of view. External battery packs were employed for the first three deployments in August 2022. Subsequent deployments used standard GoPro™ batteries. In the same manner as the OOCAM, GoPro™ cameras were deployed and retrieved by researchers. Camera deployments were conducted according to field conditions and the dive schedule of the commercial dive operator supporting this research.

Video footage was analysed using VLC Media Player Version 3.0.18 (VideoLan, 2006). Sightings of elasmobranchs were systematically recorded by trained citizen scientists and monitored by staff from the organisation All Out Africa (AOA). All contributing citizen scientists received standardised training on shark and ray identification via presentations by AOA staff. AOA staff administering training and monitoring data input had a minimum of M.Sc. degree in the field of marine science. Each sighting began when an elasmobranch individual entered the video frame for the first time and ended after it left the frame and did not return for more than five minutes (Oliver et al., 2011). For the purpose of this study, only sightings where animals could be identified to species level were included. Uniquely identifying markings such as spot patterns, predatory bite marks, injuries, or tail length were used to differentiate between individuals. Citizen scientists recorded the time the elasmobranch entered and exited the frame and the direction of travel. In addition, camera type (OOCAM or GoPro™), camera position on the reef (cleaning station A or cleaning station B), video file name and date were noted. A new data entry row was used for every new animal entering the video frame. All records of elasmobranch sightings on remote video cameras were confirmed by the researchers, who then also recorded additional information on displayed behaviours ( Table 2 ), co-occurrence with other elasmobranchs and the presence of “hitchhiker” species (i.e. cobia (Rachycentron canadum) and remoras (Echeneis spp.)). Further assessment of elasmobranch behaviours was restricted to species with 25 or more sightings to ensure data reliability and reduce the risk of potential anomalies skewing the results of this study. A random subset of 10% of the video footage analysed by citizen scientists was reviewed by AOA staff for quality control purposes.

This research used remote underwater video, which is a non-invasive sampling method. No animal was captured, handled or removed from its natural environment for the purpose of this study. Research activities described in this paper were approved by the University of Gibraltar’s ethics committee under the ethics policy for research involving animals and their environment (ID 080/2023/UniGib). Fieldwork was conducted with the permission of the Museu de História Natural de Moçambique and adhered to the safety and scientific regimen of All Out Africa.

Between July 2021 and October 2023, the OOCAM was deployed six times, resulting in 442 hours of remote underwater video data. Recording lengths per deployment varied from 21 hours to 134 hours, with an average of 63 hours. 58 GoPro™ deployments between August 2022 and March 2024 attained 82 hours of video data. Each deployment varied from 0.3 to 5.5 hours with an average duration of 1.4 hours per deployment. External battery packs were employed for the first three GoPro™ deployments in August 2022 to prolong battery life. However, the external extension cables caused corrosion to the seals of the cameras’ underwater housings, resulting in camera flooding. Consequently, subsequent deployments used standard GoPro™ batteries. LL-RUV data is summarised in Table 3 .

The LL-RUV recordings provide evidence of the occurrence of 14 elasmobranch species at the studied cleaning stations, all of which are classified as critically endangered, endangered, vulnerable or data deficient on the IUCN Red List ( Table 4 ). The OOCAM recorded a total of 254 elasmobranch sightings during the study period. These were reef manta rays (n = 63), smalleye stingrays (n = 58), oceanic manta rays (n = 40), spotted eagle rays (n = 30), blotched fantail rays (n = 20), fevers of shortfin devil rays (n = 4, total of 15 individuals), bowmouth guitarfish (n = 10), blacktip sharks (n = 19), Jenkins whiprays (n = 5), pink whiprays (n = 1), scalloped hammerhead sharks (n = 1), grey reef sharks (n = 1), whale sharks (n = 1) and whitespotted wedgefish (n = 1). The GoPro™ camera station yielded 86 elasmobranch sightings: Oceanic manta rays (n = 21), reef manta rays (n = 18), smalleye stingrays (n = 14), spotted eagle rays (n = 10), bowmouth guitarfish (n = 8), blacktip sharks (n = 8), fevers of shortfin devil rays (n = 5, total of approximately 196 individuals), blotched fantail rays (n = 1) and scalloped hammerhead sharks (n = 1). In combination, reef manta rays (n = 81), smalleye stingrays (n = 72), oceanic manta rays (n = 61), spotted eagle rays (n = 40) and blacktip sharks (n = 27) were the most frequently recorded animals at the studied cleaning stations ( Table 4 ).

Of the elasmobranch species recorded more than 25 times (reef manta ray, smalleye stingray, oceanic manta ray, spotted eagle ray and blacktip shark), each exhibited either cleaning or cruising behaviours. No feeding or courtship behaviours were observed ( Figure 4 ). Out of the 81 sightings of reef manta rays, 78% displayed cleaning behaviour. The remaining 22% of sightings showed reef manta rays cruising over the reef with their cephalic lobes rolled and mouth closed. Oceanic manta rays also predominantly exhibited cleaning behaviour (72% of sightings), with the remaining 28% of sightings showing oceanic manta rays cruising over the reef. Smalleye stingrays were primarily observed cleaning (93% of sightings). The remaining observations of smalleye stingrays (7%) showed them cruising over the reef without slowing down or circling back into frame. Unlike manta rays and smalleye stingrays, spotted eagle rays were solely observed cruising in mid-water (n = 13) or close to the reef (n = 27), either individually (n = 13) or fevers of up to five individuals (n = 27). No cleaning behaviour or other behaviours were recorded by RUV cameras. Blacktip sharks, the only shark species recorded by LL-RUV cameras more than 25 times, were observed cruising either mid-water (n = 13) or close to the reef (n = 14). As was the case with spotted eagle rays, no other behaviours were observed for blacktip sharks.

RUV cameras captured 14 instances of between two and five reef or oceanic manta ray individuals, passing in and out of the camera’s field of view, alternately circling the cleaning station. On three occasions, individuals of both species were observed alternately travelling over the cleaning station within approximately one minute of each other. On eight occasions, RUV cameras recorded two smalleye stingray individuals travelling over the cleaning station within less than eight minutes of each other, four of which were observed in frame together. Both reef and oceanic manta rays were recorded at the cleaning station within less than one minute of a smalleye stingray (n = 5), some even within three to six seconds of the smalleye stingray (n = 3). Reef manta rays were also observed circling the cleaning station alternately with bowmouth guitarfish (n = 3). Of these, there were two instances where both the reef manta ray and bowmouth guitarfish were captured in frame together. In one instance, both simultaneously turned and swam off in different directions ( Supplementary Materials 1 ), in the other the bowmouth guitarfish passed behind the reef manta ray with no noticeable reaction by either animal. On one occasion, six reef manta ray individuals, two smalleye stingray individuals and one bowmouth guitarfish were observed alternately travelling over the cleaning station, moving in and out of frame repeatedly, for more than 90 minutes. Additionally, a smalleye stingray and a pink whipray were recorded once in frame together at the cleaning station, both cleaning and travelling in and out of frame together.

Both manta ray species and smalleye stingrays were observed travelling with “hitchhiker” species ( Figure 5 ). Smalleye stingrays were accompanied by between one to twelve cobia in 63% of sightings, with an average of 4.2 cobia per sighting. In contrast, reef manta rays were observed with cobia in only 6% of sightings, with an average of 1.4 individuals. Oceanic manta rays were not observed in association with cobia. Both reef and oceanic manta rays were observed travelling with remoras, although the incidence and number of remoras were higher for oceanic manta rays (75% of sightings, with an average of 3.2 individuals and a maximum of 8 individuals) compared to reef manta rays (31% of sightings, with an average of 2.2 individuals and a maximum of 3 individuals). Smalleye stingrays were observed with remoras on three occasions (4% of sightings), with an average of one individual.

This study has reported the first insights into using long-life, remote and unbaited cameras to observe the occurrence and behaviour of threatened and data-deficient elasmobranchs. The recorded species, in order of abundance of sightings, were reef manta rays, smalleye stingrays, oceanic manta rays, spotted eagle rays, blacktip sharks, bowmouth guitarfish, blotched fantail rays, Jenkins whiprays and scalloped hammerhead sharks. Pink whiprays, grey reef sharks, whale sharks and whitespotted wedgefish were only observed once. Shortfin devil rays were observed in fevers of more than 50 individuals on numerous occasions. In total, 14 species were observed ( Table 2 ), compared to the 37 species known to occur in the region ( Table 1 ). Several species listed in Table 1 were recorded only rarely, based on reports from local dive centres, including the great white shark (Carcharodon carcharias), Java shark (Carcharhinus amboinensis), shorttail nurse shark (Pseudoginglymostoma brevicaudatum), silvertip shark (Carcharhinus albimarginatus), tiger shark (Galeocerdo cuvier), bottlenose wedgefish, giant devil rays (Mobula mobular), ornate eagle rays, and cownose rays (Rhinoptera bonasus), and were not captured by the remote cameras. More than 20 observations of unidentified sharks, visible only as silhouettes and too far from the camera for clear identification, could correspond to undetected species such as bull sharks, silvertip sharks and dusky sharks. For research on pelagic sharks, BRUVs may be more effective, as bait attracts sharks to the camera, allowing for species identification. Additionally, bottom-dwelling ray species, such as bluespotted maskrays (Neotrygon caeruleopunctata), bluespotted stingrays (Taeniura lymma), and torpedo rays (Torpedo marmorata, Torpedo sinuspersici), were not detected by the camera system. These limitations are further discussed in section 4.4.

Through direct human observation techniques, such as dive and snorkel surveys and photo-identification, as well as acoustic and satellite telemetry, much is known about the presence of some of the more frequently observed species, such as manta rays, devil rays and whale sharks, along the Inhambane coastline. Such studies have addressed their population size and structure (Brooks et al., 2010; Marshall et al., 2011), trends in sightings (Rohner et al., 2013; Venables et al., 2024) and movement patterns (Rohner et al., 2018; Venables et al., 2020; Daly et al., 2023; Marshall et al., 2023). Much less is known about the lesser-seen elasmobranch species, such as smalleye stingrays, bowmouth guitarfish and other ray and wedgefish species. LL-RUV observations from the present study provide valuable insights that can inform conservation efforts, such as fisheries management, to protect habitats important to the livelihood of rare and threatened species.

Observed behaviours at the study site revealed distinct patterns among various elasmobranch species, shedding light on their utilisation of these critical habitats. Remote underwater video data demonstrated that reef manta rays, smalleye stingrays and oceanic manta rays are the most frequent elasmobranch users of the studied cleaning stations, engaging in mutualistic cleaning interactions with cleaner fish. Such cleaning interactions are vital for their health, as cleaner fish remove and ingest ectoparasites, dead tissue and other unwanted materials from clients’ bodies (Youngbluth, 1968; Losey and Margules, 1974; Grutter, 1996; Cheney and Côté, 2001), thereby improving the overall health of client species and potentially reducing the risk of infections (Grutter et al., 2003; Vaughan, 2018). Manta rays’ use of certain cleaning stations over others has previously been linked to the proximity to coastal feeding grounds (Rohner et al., 2013; Murie and Marshall, 2016). In the Praia do Tofo region, reef and oceanic manta rays can be observed feeding within a well-established year-round whale shark feeding aggregation area, located two kilometres south of the study site (Rohner et al., 2018). Such sightings support the theory of manta rays visiting cleaning stations in proximity to good feeding sites, which would explain the high number of manta ray sightings at the studied reef.

Cleaning behaviour by manta rays has previously been described as the ray reducing its swimming speed when approaching a cleaning station, hovering or circling above the reef and making repeated passes over a cleaning station while being inspected by cleaner fish (Marshall, 2008; Jaine et al., 2012; Kitchen-Wheeler, 2013). LL-RUV footage provides evidence of similar cleaning behaviours in smalleye stingrays. These rays were observed either hovering over the cleaning station while facing into the current or slowly moving their pectoral fins to maintain position, allowing cleaner fish to approach. After passing over the cleaning station, they would circle around and allow the current to carry them back to their initial position, before turning again and repeating the behaviour.

In contrast, spotted eagle rays and blacktip sharks were observed passing by the cleaning stations without actively engaging in interactions with cleaner fish. They were observed swimming in mid-water or cruising close to the reef but did not exhibit cleaning behaviour when in close proximity to the reef. This behavioural pattern raises questions about the cleaning behaviour of the two species, prompting further investigation into their preferred cleaning sites. In French Polynesia, ocellated eagle rays (Aetobatus ocellatus) were observed chafing on patches of sandy bottom, which was suggested as their primary cleaning behaviour or an addition to cleaning station visits (Berthe et al., 2016). Research by Oliver et al. (2011) on pelagic thresher sharks (Alopias pelagicus) in the Philippines provides valuable insights into shark cleaning behaviour at seamounts, suggesting potential avenues for exploring the observed differences in the utilisation of cleaning stations by various elasmobranch species. Seamounts and other offshore cleaning sites likely look fundamentally different from coastal cleaning stations in terms of cleaner fish assemblages, as different reefs may have co-evolved with different cleaning requirements and client species. Deploying camera systems strategically at different cleaning sites in different environments would help to illuminate the overall function of cleaning stations for different elasmobranch species and whether species that were not observed cleaning in the present study visit other reefs for cleaning purposes. Addressing questions regarding the use of cleaning station by different species requires a comprehensive assessment of reef topology and structure, cleaner fish abundance and composition, as well as considerations of currents and the risk of predation at the studied reef and other nearby cleaning stations. Further research into the cleaning ecology of elasmobranchs may yield valuable insights into the complex dynamics of cleaning interactions in marine ecosystems, thereby facilitating more effective conservation strategies for these vulnerable species.

The documented intra- and interspecific interactions observed at the studied reef provide novel insights into the complex social dynamics of rare and little-known elasmobranchs. The frequent presence of both species of manta ray and smalleye stingrays at the studied reef underscores the significance of the Praia do Tofo region for threatened elasmobranch populations in the Inhambane province, despite their dramatic documented decline in sightings (Venables et al., 2024). Observations of multiple individuals of the same species, as well as individuals from different species, utilising the same cleaning station concurrently, suggest a level of tolerance and cooperative behaviour among these marine organisms. The behavioural patterns exhibited by reef and oceanic manta rays, such as alternately swimming over the cleaning station, imply a degree of social organisation. Similar cooperative behaviours have been documented in other elasmobranch species including lemon sharks (Negaprion brevirostris), scalloped hammerhead sharks (Sphyrna lewini) and great white sharks (Carcharodon carcharias) (Klimley, 1983; Guttridge et al., 2011; Jacoby et al., 2012; Micarelli et al., 2023).

The behaviour observed between a reef manta ray and a bowmouth guitarfish, where they turned and swam away in different directions upon encountering each other at the cleaning station, raises questions about resource partitioning at cleaning stations. Further research into the ecological niches and behavioural repertoires of different elasmobranch species is warranted to elucidate the mechanisms underlying such interactions and their implications for community dynamics. Future research should focus on exploring how the presence of a combination of species simultaneously soliciting cleaning services affects the quality of cleaning received by each individual, whether there is a hierarchy among client species, or whether cleaner fish preferentially clean certain ray species over others.

The observation of a smalleye stingray and a pink whipray travelling together, with the pink whipray closely trailing the smalleye stingray in and out of frame, supports findings from the Great Barrier Reef in Australia, where pink whiprays have been observed piggybacking on smalleye stingrays (Meekan et al., 2016). This is the only such observation in this study, and further research is required to better understand this behaviour and its ecological significance.

The presence of “hitchhiker” species alongside smalleye stingrays and manta rays suggests potential symbiotic relationships or ecological strategies (Becerril-García et al., 2020; Nicholson-Jack et al., 2021; Pate et al., 2021). Oceanic manta rays, in particular, were frequently observed traveling with up to eight remoras attached to their ventral surface, while reef manta rays were also observed with remoras but less frequently and in lower numbers. These remoras, which feed primarily on zooplankton, attach themselves to manta rays and other megafauna species to be carried to feeding areas (Nicholson-Jack et al., 2021). In the process, they reduce the hydrodynamic swimming efficiency of their host, but it is possible that they also dislodge larger parasites (Nicholson-Jack et al., 2021). Smalleye stingrays and reef manta rays were also observed travelling with cobia on multiple occasions. The association of cobia especially with smalleye stingrays could indicate commensalism, where cobia benefit from protection or access to food resources provided by the larger rays (Michael, 1993). Further research is required to fully understand the nature and extent of these associations and their ecological implications.

The use of remote underwater video cameras has proven to be a non-invasive research method that holds significant potential for studying the undisturbed, unbaited, natural behaviours of some of the rarest and most threatened marine megafauna species. Remote underwater video cameras, especially LL-RUV cameras, offer a range of advantages over traditional research methods. RUV allows researchers to observe and document the occurrence and behaviour of marine species without the need for human interaction, thus minimising disturbance to the animals and reducing observational bias caused by animals altering their behaviour in the presence of divers (Becker et al., 2010; Dickens et al., 2011; Lee et al., 2016; Zarco-Perello and Enríquez, 2019). Previous RUV studies primarily used action cameras and therefore faced limitations due to their short battery life and constrained recording duration (Oliver et al., 2011; Barr and Abelson, 2019; Zarco-Perello and Enríquez, 2019; Peel et al., 2020).

This study demonstrates that LL-RUV cameras provide more comprehensive data on the occurrence and undisturbed behaviours of rare elasmobranch species than direct human observation techniques by offering uninterrupted video footage over extended periods (Bilodeau et al., 2022; Cordier et al., 2022). The extended recording duration of LL-RUV cameras also minimises disturbances caused by the presence of divers, their bubbles, or boat noise, which can deter elasmobranchs from cleaning stations, leading to them being missed during short-term surveys (de Vincenzi et al., 2021; Bilodeau et al., 2022; Cordier et al., 2022; Mickle and Higgs, 2022). Continuous monitoring has the potential to provide valuable insights into diurnal and seasonal variations in habitat use, undisturbed behaviours of various species, predator-prey interactions and other ecological dynamics, while also capturing activity during sunrise and sunset periods when divers are absent from the reefs. In contrast, the limited recording window of GoPro™ cameras means that video is only captured within 90 minutes of diver disturbance, which could lead to certain species avoiding the area, thus highlighting the advantage of using LL-RUV cameras for continuous, uninterrupted monitoring during all daylight hours.

Despite their numerous advantages, LL-RUV cameras have some limitations that should be acknowledged. The quality of video footage can be significantly affected in areas with low water visibility or high turbidity, such as near estuaries or river mouths (Figueroa-Pico et al., 2020; Jones et al., 2021). Under these conditions, animal sightings may be missed, particularly for species that do not approach the camera closely. While this issue also affects diver-based surveys, it highlights the importance of deploying LL-RUV cameras under favourable environmental conditions to avoid skewed results. For shark research, the absence of bait in LL-RUV setups can pose a challenge. BRUV systems are more effective for attracting sharks closer to the camera, enabling easier species identification and behavioural observations (Whitmarsh et al., 2017; Prat-Varela et al., 2023). By contrast, the LL-RUV camera used in this study is particularly well-suited to studying the natural cleaning behaviours of rays, such as manta rays and smalleye stingrays, which frequent cleaning stations. These species are less likely to respond to bait, making LL-RUVs a more appropriate tool for documenting their behaviour. Furthermore, baiting can introduce biases, as different species respond differently depending on bait type, and some, such as filter feeders, are entirely unaffected. The non-baited nature of LL-RUV systems allows for the observation of undisturbed, natural behaviours, providing a broader ecological perspective. Additionally, LL-RUVs avoid ecological disturbances caused by bait, which can disrupt local food webs or attract predators unnaturally. Unlike BRUVs, which often favour carnivorous or scavenging species, LL-RUVs capture a broader representation of marine communities, including herbivorous and filter-feeding species. The extended recording durations of LL-RUV cameras facilitate long-term, continuous monitoring, capturing behaviours and interactions across various times of the day, including periods when divers or baited systems are absent. Finally, by minimising diver presence, LL-RUV cameras reduce human-induced disturbance, making them an invaluable tool for studying sensitive habitats or species.

Although this study focused on the advantages of the LL-RUVs, it is important to also recognise the strengths of actions cameras like GoPro™ cameras in capturing high-resolution footage with a wider field of view. The superior video quality of GoPro™ cameras facilitated the identification of elasmobranch species, as their distinctive visual markings were visible more clearly, resulting in fewer unidentified sharks and rays in the dataset for GoPro™ recordings compared to the OOCAM. GoPro™ cameras were able to capture fevers of devil rays of up to 55 individuals in contrast to the OOCAM, which recorded only individual devil rays or fevers of up to six individuals. This disparity is likely due to the GoPro™ cameras’ wider field of vision, which enabled them to capture mid-water sightings as well as documenting cleaning stations dynamics. Depending on the research focus, four GoPro™ cameras spaced 90° apart can provide a more comprehensive view of reef activity and reduce the risk of missing rare species or behaviours. Alternatively, 360°cameras, which offer a comprehensive environmental view, could also be considered, though they share the same short battery life challenge as GoPro™ cameras (Auster and Giacalone, 2021; Hemery et al., 2022). Future studies should seek to combine the benefits of both camera systems, with LL-RUVs’ extended battery life and continuous monitoring capabilities complementing the high-resolution video and broader field of view of action cameras. This combination could provide an even more comprehensive approach to studying marine megafauna, enabling researchers to capture both cleaning behaviours and pelagic species occurrences while maximizing the strengths of each system in documenting diverse ecological dynamics.

Further analysis of temporal trends in sightings, ideally using LL-RUV cameras that capture all hours of the day, could provide additional insights into observed differences in species occurrence and behaviour. Elasmobranch visitations to the studied cleaning stations may be more frequent during the hours when SCUBA divers are present on the reefs (08:00 until 14:00), which is also when GoPro™ deployments are possible. A more continuous, non-invasive monitoring system like the OOCAM could help to enrich the data by capturing these temporal patterns over longer periods, providing a clearer understanding of species distribution and behaviour.

The main challenge of RUV-based research, especially long-term or continuous monitoring, is that data processing can be time-consuming, creating a bottleneck in data analysis (Erickson et al., 2023; Magneville et al., 2023). However, recent advancements in machine learning and computer vision algorithms have enabled automated species recognition, behaviour analysis and object tracking in large video datasets (Ditria et al., 2020, 2021; Marrable et al., 2022). These technological advances in artificial intelligence and machine learning have the potential to significantly reduce the time and effort required for data processing and analysis, thereby facilitating comprehensive, long-term monitoring of marine environments. Consequently, the use of RUV and associated technologies could open new avenues for efficient marine research, especially in the age of big data, leading to more timely development for conservation strategies. Future research should focus on developing automated video analysis and species recognition techniques to streamline the data analysis process, making it feasible to continuously monitor different reefs. This would enable comparisons across reefs and provide deeper insights into ecosystem dynamics. An example of this advancement is FishID, an AI-system that identifies, counts and measures aquatic animals and plants (FishID, 2024). Developed as part of The Global Wetlands Project at Griffith University, FishID exemplifies how technology can enhance marine conservation (FishID, 2024). Other organisations are also integrating AI solutions for environmental monitoring, showcasing the potential of technology to improve our understanding of marine ecosystems and conservation efforts.

In lieu of an automated data processing technique, contributions by trained citizen scientists in the analysis of LL-RUV footage were instrumental in this study. Previous research has shown that, if trained well, citizen scientists can acquire the knowledge and skills essential for contributing to scientific research (Norman et al., 2017; Keeping et al., 2021; Magson et al., 2022). Kosmala et al. (2016) and Chin and Pecl (2018) outline potential issues with the quality of data that can arise when using citizen scientists instead of trained scientists, but issues relating to the accuracy of data can be minimized with training and supervision of citizen scientists as well as expert validation of records (Kosmala et al., 2016; Keeping et al., 2021). In the present study, citizen scientists were trained on shark and ray identification through presentations and in-water training by AOA staff with a minimum of M.Sc. degree in a related field. All records of megafauna sightings on remote video cameras were confirmed by the researchers, who re-watched all video segment included in this study. In addition, a random subset of 10% of the video footage analysed by citizen scientists was reviewed by AOA staff to ensure sightings were recorded correctly.