Monitoring marine fishes using underwater video techniques in the Mediterranean Sea

The use of underwater video techniques has expanded rapidly in ecological studies and is particularly desirable in protected areas since the method does not impact the habitat or remove fish. The Mediterranean Sea is a biodiversity hotspot under high anthropogenic pressure and consequently, non-destructive and non-extractive techniques for fish monitoring are advantageous. Here, we review 110 publications that used underwater video in fish-related studies in the Mediterranean basin. The most common technique used in the Mediterranean Sea was Remotely Operated Vehicles (ROV) (33%), followed by Remote Underwater Video (RUV) systems (20%), Diver Operated Video (DOV) systems (20%) and Baited Remote Underwater Video (BRUV) systems (19%). Approximately one third of the studies used an additional sampling method, such as fisheries-based or molecular methods with the aim to compare the surveying effectiveness or produce complementary data. The most frequent objectives of the reviewed studies were related to fish community structure, i.e., focusing on community wide metrics such as abundance and biodiversity, or behavioral analyses, while the most commonly studied environments were those of the western Mediterranean and shallow waters, usually involving sandy or rocky reef habitats. Sampling protocols differed widely among studies with transect lengths, soak times and baits all varying. Future research should focus on the least studied parts of the region, such as the eastern and southern Mediterranean Sea and deep-sea habitats. Finally, the development of standardized sampling protocols is recommended to ensure that data are comparable among studies.


Introduction
Fish monitoring underpins conservation and management and is particularly critical given the declining biomass trends that characterize many commercial fish species worldwide (Costello et al. 2012;Palomares et al. 2020).Fish monitoring can utilize either fishery-dependent (commercial or recreational) or fishery-independent data.Fishery-independent techniques include experimental fishing (Priester et al. 2021), remote sensing (Santos et al. 2000), acoustic surveys (Rountree et al. 2006), telemetry (Hammerschlag et al. 2011), underwater visual census (Samoilys and Carlos 2000), genetics (Blower et al. 2012) and underwater video (Mallet and Pelletier 2014).The use of non-destructive and non-extractive techniques, such as those based on video, is particularly desirable in Marine Protected Areas (MPAs) in order to maintain the integrity of the ecosystem and avoid placing further pressure on fish populations (Collie et al. 2000;Sciberras et al. 2018;Murphy and Jenkins 2010).Underwater video use in ecological studies mainly aims to monitor fish populations (Marra et al. 2016), study the behavior of fish (Ajemian et al. 2016) or develop and compare methodologies (Stobart et al. 2007).
The first underwater video system for biological studies was described in 1952 (Barnes 1952) and since then, underwater video has become an increasingly popular ecological tool due to its versatility and the technological advances during the last two decades (Mallet and Pelletier 2014).In the Mediterranean Sea, the first underwater video study was conducted in late 1970s (Fedra and Machan 1979) with a rapid increase in the use of underwater video after 2000.Most underwater video techniques, in contrast to Underwater Visual Census (UVC), enable scientists to sample without time and depth limitations and reach locations inaccessible to divers (Cappo et al. 2007; Unsworth et al. 2014).Video-based methods provide a permanent record of standardized replicates with detailed optical view, which minimizes observer bias regarding species identification, fish length estimates and sample unit area (Langlois et al. 2010;Unsworth et al. 2014).Moreover, sampling can be performed by non-taxonomy experts and video footage can be examined several times and by different observers in the lab (Langlois et al. 2010).Contrary to fisheries-based methods and experimental fishing, video techniques also provide information on habitat (Collins et al. 2017) and animal behavior (Ayma et al. 2016) and are less selective in terms of species and sizes (Murphy and Jenkins 2010).The use of video techniques is however constrained under poor visibility conditions (Sward et al. 2019) and the presence of bait in baited camera systems may introduce unnatural behaviours (Harvey et al. 2007).When bait is used, the area extent from which fish are drawn is unknown and thus, abundance is relative rather than a real estimate (Heagney et al. 2007).
Underwater video-based fish studies can utilize a range of methodologies, depending on the study objectives and available resources.Underwater video equipment can be permanently or temporarily deployed, in shallow water or abyssal depths, powered by batteries or cable, stationary or deployed along transects, and target pelagic or benthic habitats.Systems may be deployed with single cameras or in stereo-configuration to allow length measurements.Worldwide, commonly used methods include the remote underwater video technique (RUV), baited remote underwater video (BRUV) systems, remotely operated vehicle (ROV), diver operated video (DOV) systems and towed video (TOWV) Techniques.
Remote underwater video (RUV) systems consist of a video recording device being placed underwater, on the seafloor or in the water column, and do not require human presence to operate.The systems can be set by a diver or deployed from a vessel (Mallet and Pelletier 2014) and can be either linked to a monitoring station by cable (Aguzzi et al. 2011), usually used for long term monitoring, or deployed autonomously operating by battery power (Galasso et al. 2015).Bait can be placed optionally in front of the mounted camera(s) creating an odor plume in order to attract organisms from a broader area, in which case, the systems are referred to as Baited Remote Underwater Video (BRUV) systems (Cappo et al. 2003).Fish monitoring studies increasingly use BRUV systems (Bailey et al. 2007), especially in Australia (Mallet & Pelletier 2014;Whitmarsh et al. 2017;Harvey et al. 2021) and their use is strongly recommended for mobile and predatory species (Cappo et al. 2003).Similar to RUV, BRUV systems can be either linked to an onboard personal computer to provide real time underwater footage (D'Onghia et al. 2015a) or autonomously function underwater without cables (Torres et al. 2020).
Vol.: (0123456789) Remotely Operated Vehicle (ROV) is an unmanned underwater submersible operated from the surface.ROVs are available in various sizes and can perform at different depths (Sward et al. 2019).They can carry a large range of tools, such as jaw grabbers, multiple cameras, laser beams and multiparametric probes (Bo et al. 2012a;D'Onghia et al. 2015a).The main advantage of ROVs is that they can survey at deep waters and cover large areas (Sward et al. 2019).However, careful consideration is needed as to the appropriateness of the technique as certain fish species can react negatively to ROV presence (Ayma et al. 2016).Another limiting factor for ROVs use in fish studies is their high cost (Sward et al. 2019) but during the last decade, market demand has led to the creation of mini-ROVs, tools that are low-cost, reliable and easy to operate (Rubin 2013).
Diver Operated Video (DOV) systems is a form of underwater visual census where the transect is filmed by a diver.DOV surveys can be conducted along predesigned transects (Marra et al. 2016), by stationary point count (Mascolino et al. 2019) or by following the focal organisms (Goverts et al. 2021).The main advantage of DOV systems is that the diver-operated camera can be maneuvered within inaccessible areas such as overhangs, and thus record cryptic species (Watson et al. 2005).Limitations of DOV techniques include their constraint to relatively shallow depths and the impacts of the diver's presence on fish (Langlois et al. 2010).
Towed video (TOWV) systems are survey tools consisting of an underwater camera attached on a frame linked to a vessel such that the frame is drawn behind the vessel, either in the form of a sled (Spencer et al. 2005) or a structure hovering above the seafloor (Barker et al. 1999).The main advantage of a TOWV system is its suitability for high current areas as it is driven by the vessel's horsepower (Barker et al. 1999).A disadvantage is that a sled's operation is challenging on rocky and irregular substrates (Rooper 2008).
In the Mediterranean Sea, all these methods have been used.In addition to applications to fish ecology and monitoring, video recording techniques have also been used for commercial purposes in aquaculture and fisheries.Aquaculture studies have used such techniques to estimate abundance and biomass of bluefin tuna (Costa et al. 2009;Mariani et al. 2014) or behavioral changes of captive fish under different conditions (Sarà et al. 2007;2010).Fisheries studies use video techniques to estimate the catchability of fishing gears (Dremière et al. 1999;Papadopoulou et al. 2015) and record the effect fish aggregating devices can have on fish community structure (Addis et al. 2016) or fish behavior (Sinopoli et al. 2015).During the last years, sea users are also contributing to ecological fish knowledge by producing public amateur underwater videos.Fish videos by recreational fishermen or scuba divers often help scientists identify the appearance of non-indigenous fish for the first time (Stern et al. 2019), track their expanding distribution in the Mediterranean (Deidun et al. 2011) or provide information about the spatio-temporal distribution of important species, such as groupers (Sbragaglia et al. 2021).
This review focuses on underwater video techniques used in the Mediterranean Sea, which have been either primarily or secondarily used to study marine fishes.Underwater video techniques have a promising future in the study of fish given their ongoing evolution in performance and their non-destructive and non-extractive nature, making their review timely.The Mediterranean Sea hosts over 17,000 marine species (Coll et al. 2010), 750 of which are fishes (Dimarchopoulou et al. 2017), and has been highly impacted by anthropogenic activities over the last decades (Halpern et al. 2008;Micheli et al. 2013).Fisheries (Piroddi et al. 2020), climate change (Barnett et al. 2001) and invasive species (Katsanevakis et al. 2014) have all contributed to the declining health of the Mediterranean Sea by leading directly and indirectly to substantial declines in species richness, diversity, density and biomass levels of both vertebrates and invertebrates (Claudet & Fraschetti 2010;Coll et al. 2010;Prato et al. 2013).This is particularly the case for species targeted by fisheries, the majority of them being overexploited (Tsikliras et al. 2015).Reflecting the need to better understand the fish of the Mediterranean, this review also provides recommendations for fish monitoring.

Literature survey and data selection
Searches in Google Scholar and Scopus were conducted for all articles published prior to 31 December 2021.We used the keyword string "ROV or RUV or VIDEO or DOV or TOWV or BRUV or RECORD-ING and MEDITERRANEAN and FISH".The first 1000 findings from Google Scholar were viewed, while all results from the extended field of natural sciences in Scopus were viewed, including the results "Agricultural and Biological Sciences", "Biochemistry", "Genetics and Molecular Biology", "Environmental Science", "Earth and Planetary Sciences", "Engineering", "Multidisciplinary", "Computer Science", "Physics and Astronomy", "Chemistry", "Mathematics", "Chemical Engineering", "Energy", "Veterinary".We included only publications in peer reviewed journals and conference proceedings in the English language.Materials in the grey literature (i.e., technical reports, theses etc.) were excluded from the analysis.With the above selection criteria, 110 publications were gathered (102 peer reviewed journal articles and 8 conference papers) and were further reviewed.Included studies were strictly limited to marine waters, the use of underwater video recording techniques and pertained to fish, both as their main researched organisms and as unintentional occurrences in another research framework.The reviewed studies were also carried out exclusively by scientists and were held in field environments.Therefore, citizen science studies and those in the fields of aquaculture and fisheries were not included.
Consistent information was recorded for each included study, including the year, the study objective, the studied organisms, the sampling site and technique and the data analysis.Publications were assigned to categories according to the study objective: (1) studies related to fish behavior ("behavioral"); (2) studies focusing on community analysis and community wide metrics such as abundance and biodiversity ("community related"); (3) studies making comparisons between different techniques and describing new methods ("methodological"); and (4) studies with main focus usually on benthos and not fish, but mentioning fish species in the results as species associated with the studied system ("auxiliary").Category 3 included only studies that described a new technique rather than studies using a modified technique.For each publication, the focal study organisms were recorded and assigned to one of the following categories: (1) single fish species; (2) groups of fish species; (3) the entire fish community; or (4) other organisms such as benthic invertebrates.The information extracted from the videos was noted, including, for instance, abundance, biomass and fish size.The habitat type was also noted based on the substrates included in the sampling regions.Habitat types included coralligenous habitat, coral habitat (alive, dead or rubble), rocky reef or algae covered rock, seagrass, artificial substrate and grains.Grains were further categorized by increasing grain sizes and are hereafter referred in the text as "sand" (clay/ silt/mud/sand), "gravel" (gravel/pebble/cobble) and "bare rock" (boulder/bare rock).Artificial substrates included artificial reef, hard substrates on pillars of gas platforms or power stations and shipwrecks.The depths at which the video recording systems were deployed in each publication were recorded as well.
Sampling technique was assigned as one of 5 categories: (1) "RUV" included all studies in which a recording device was fixed underwater; (2) "BRUV" was assigned to all baited RUV studies; (3) "ROV" included studies using remotely operated video; (4) "DOV" included those studies that used diver-held underwater cameras to record; and (5) "Other" category consisted of TOWV, manned submersible, techniques that did not fall in any of the other descriptions e.g.camera fixed on a kayak recording in transects or camera fixed on a trawl, and all other sampling methods that were not clearly described.RUV and BRUV techniques ranged from custom-made structures equipped with cameras (Torres et al. 2020) to benthic landers (Capezzuto et al. 2012).Lastly, information on additional sampling methods apart from the video techniques, such as nets and longline was recorded.
Percentages in the results section do not sum up to 100% since most studies were conducted in multiple areas and countries, covered multiple habitats, used more than one sampling gear etc.

Year and geographical characteristics
The first underwater video study in the Mediterranean was published in 1979 (Fig. 1).A 20-year research gap followed with little to no use of video-based methods.From approximately 2010 and onwards, the use of video techniques in fish research has increased rapidly, with more than half of all studies published since 2016 (Fig. 1).Across 110 publications 115 records of video techniques were identified.Most studies used ROV (33%) while RUV, BRUV and DOV systems were almost equally used in around 20% of studies each.In a small number of studies (8%) the description of the video technique was inadequate and thus, could not be assigned to a specific group.Articles using manned submersible vehicles and TOWV techniques were published once and twice respectively.
The most studied area in the Mediterranean Sea using underwater video techniques was the western part of the basin, surveyed in 63% of the reviewed publications (Fig. 2).Specifically, almost half (49%) of the studies were conducted in three GSAs, all located in the western Mediterranean: northern Spain (GSA 6), southern and central Tyrrhenian (GSA 10) and Balearic Islands (GSA 5).The central Mediterranean and Adriatic Sea were the geographic locations of 19 and 16% of the studies, respectively, while only 7% of the studies were conducted in the eastern Mediterranean Sea (Fig. 2).The least studied area was the southern part of the Mediterranean basin with no studies being conducted in the Alboran Island (GSA 2), Algeria (GSA 4), northern Tunisia (GSA 12), Gulf of Gabes (GSA 14), Malta (GSA 15) and southern Ionian Sea (GSA 21).By country, the majority of the  Approximately one third of the publications (35%) stated that they were conducted within MPAs, while four studies took place in locations that were candidates to become MPAs at the time.The rest of the publications did not specify whether they were conducted in protected areas or not, however the possibility of more studies being conducted in MPAs but failing to mention it cannot be ruled out.The prevailing sampling methods used in MPAs were RUV (37%), DOV (26%) and BRUV (21%) systems.In MPAs, many studies using underwater video explored the protection effect on fish populations and fish behavior.However, the use of MPAs as sampling sites has not been restricted to protection effect studies.Populational changes on spatial and temporal level and habitat use based on the season were also questions addressed within MPAs.Lastly, biodiversity characterization has taken place in regions candidates for MPAs (Table S1 in Supplementary Material).

Depth & habitat
The areas that were mostly covered (more than 50%) by underwater video methods were shallow waters of less than 500 m depth.The maximum depth at which video systems were deployed was greater than 5000 m depth (Fig. 3A).Studies using DOV and RUV systems were restricted to shallow depths (Fig. 3B) up to 62 and 45 m, respectively (Fig. 3B).All other video systems sampled a wider range of depths from shallow waters (0-20 m) down to the bathypelagic (ROV) or even the abyssopelagic zone (BRUV).
In most studies (93%), video techniques were used to study fish close to the seafloor, as the camera systems were set on the seafloor (RUV, BRUV), operated on the seafloor (TOWV) or above it (DOV, ROV) usually at around 1.5 m from the bottom.Exceptions were specific studies in which the camera was set in the water column to monitor pelagic predators, set at a depth gradient along fish farm cages or under the bow of a kayak in a study of fish availability as prey for birds.Also, in certain DOV and ROV studies, divers and ROVs deviated from the seafloor to perform the survey by following the distribution pattern of the studied fish.

Objectives
The objective of the studies utilizing video recording techniques varied widely (Table S1 in Supplementary Material) with some studies having multiple ones.In total 126 objective records were identified.Most studied topics were at fish community level (44%) focusing mainly on differences in community wide metrics, such as abundance.ROV (31%) and BRUV systems (29%) were the most popular techniques in community analysis studies, followed by DOV systems (21%).Secondary to this objective were behavioral studies (34%) focusing on feeding or mating behaviour for example.Differences in community wide metrics or behavior were tested against season, location, depth, level of protection, habitat complexity or habitat type, environmental factors, such as CO 2 concentration and temperature, time of the day, tourism impact or alien algae invasion (Table S1 in Supplementary Material).Behavioral hypotheses were usually addressed using RUV and DOV systems (38 and 35%, respectively).Apart from the community related and behavioral studies, some studies (17%) were methodological and others (20%) included fish as organisms of secondary interest associated with other animals, such as corals (Table S1 in Supplementary Material).These studies were conducted almost entirely (91%) by ROV, whereas BRUV systems were frequently used in methodological studies (42%).
Most methodological publications compared exclusively the effectiveness of different techniques to study fish (72%), while some focused only on the description of a new sampling method (22%) and one study described a new methodology and also compared it to another method (Table S1 in Supplementary Material).Comparative studies were related to the effectiveness of video techniques against other fish sampling methods such as UVC, eDNA, nets, longline and trawling.
More than half of the studies (71%) that used recording devices focused exclusively on fish without references to other organisms such as invertebrates, which might have been recorded.Few studies examined both fish and benthos concurrently (9%) and some (20%) used video to mainly study benthic macrofauna while mentioning the occurrence of fish when present.RUV and BRUV systems were almost entirely used in fish studies.In contrast, more than half of the ROV surveys included fish in their results only as an incidental occurrence in benthic studies (Table S1 in Supplementary Material).Additionally, 6 ROV studies incorporated also the human impact in the form of marine litter, in addition to fish and benthos (Table S1 in Supplementary Material).

Combination of multiple techniques
More than two thirds of the publications (69%) used exclusively a single video technique, while others (26%) combined a video technique with another method, such as fisheries data.Three studies (3%) used two different video techniques and two studies (2%) combined two video techniques with additional sampling methods (Table S1 in Supplementary Material).
Generally, the use of multiple techniques aimed to combine their results and make inferences for different metrics (59%), compare methodologies (12%) or both (29%).UVC was the most common additional method (35%) and was mostly combined with DOV (in 42% of the DOV studies).The use of fisheries-based methods was also frequent and included long-lines, nets, trawls and traps (Table S1 in Supplementary Material).These techniques were usually combined with BRUV and rarely with ROV.Behavioral studies related to foraging, besides video techniques, included the collection of specimens for gut analysis, the calculation of their Body Condition Index or stable isotope and fatty acid analyses.The consumed biomass was also measured in an herbivory experiment.Further methods included theoretical modelling and eDNA analysis (Table S1 in Supplementary Material).

Technical information: sampling & video analysis
Clupeid species and more specific European pilchard (Sardina pilchardus) and round sardinella (Sardinella aurita), were the most common bait used (38%) in BRUV sampling either individually or as part of a mixture in pellets.Mackerels (Scomber spp.) was also a common bait (29%), while baits mentioned only in one study (8% each) included a mix of fish scraps, cephalopods, and/or cetacean flesh and oil, tuna pieces and phytodetritus in the form of a live diatom culture deposition.Only three studies failed to mention the type of bait used.
The sampling effort in stationary techniques (RUV and BRUV systems) was the soak time (Fig. 5A).Many publications (28%) using RUV or BRUV systems provided only the range of time analyzed, or the minimum or maximum deployment duration.It was common for BRUV systems to sample for 30 min (19%), while in RUV studies sampling lasted for even less than 30 min (27%).Five publications using stationary techniques (12%) did not specify the sampling effort, while five of the RUV studies (23%) mentioned the rate in which they extracted picture frames from their video footage.
In transect techniques the sampling effort was estimated according to the transect length (Fig. 5B).Most of the ROV studies (56%) included no clear information regarding the transect length sampled.In some cases (22%) transects were of unequal length and only a range of the size was included.A small number of publications sampling with ROV used transects of "Range" includes studies mentioning the minimum, maximum or range of deployment duration, "unspecified" refers to non-provided information and "pictures" represents the cases in which frames were extracted from the video footage.b The transect length (m) in Diver Operated Video (DOV) and Remotely Operated Vehicle (ROV) studies."Range" includes studies using various transect lengths, and studies in which no information was given are placed in the "unspecified" group 100 m and shorter (14%), while in three studies (8%) the ROV sampled for transects greater than 100 m.Usual transect lengths in DOV studies (Fig. 5B) were 25 or 50 m (36%), while more than half of the publications (64%) did not specify the transect length.
Depending on the study goal different data could be derived from the videos processed (Table 1).Data types included community wide metrics, behavior, such as mating and feeding behavior, and functional traits, such as trophic guilds.A common metric was species abundance, that can be estimated in multiple ways depending on the technique type used and author preferences (Table 1).In stationary techniques, besides the traditional metrics such as biodiversity, evenness and species richness, time of first arrival and the time at which the maximum number of individuals per species is seen (tMax) were measured (Table 1).

Discussion
The application of video recording techniques in fish monitoring has been increasing worldwide (Mallet and Pelletier 2014) and this trend is also followed in the Mediterranean Sea, as described in the present work.The use of video recording techniques is expected to increase in the future, particularly for monitoring fish populations and habitats in MPAs and sensitive ecosystems as they are non-destructive and fisheries independent methods (Murphy and Jenkins 2010).
The hydrography of the Mediterranean Sea, a semi-enclosed basin, contributes to the rapid warming up due to climate change, which is more intense in the south than the north (Marbà et al. 2015).At the same time, local environmental knowledge has already indicated alterations in species abundance and distribution in the northern parts of Africa, as thermophilic species are undergoing abundance rises (Azzurro et al. 2011) and other species distributions are shifting northwards (Azzurro et al. 2019).Additional stress is present in the eastern Mediterranean that is impacted by invasions of non-native species more than the western (Galil et al. 2018).Even though rapid changes are occurring in the southern and eastern parts of the Mediterranean, the northwestern regions are the most well studied areas as shown by this work and former studies (Marbà et al. 2015).Considering the above, the need to conduct new surveys in these quickly changing parts of the southern and eastern Mediterranean arises.Future research could focus on gathering data of southern Mediterranean fish to determine the current state of species inhabiting these waters, but it could also focus on comparative studies between the south and north, to identify and highlight the observed changes.Regarding the expansion of alien species, studies could be conducted not only to track their distribution, but also to explore their behavior in the new ecosystems (D' Agostino et al. 2020) and the potential altered behavior of the native species as a response to the invasion (Alós et al 2018;Vivó-Pons et al. 2020).
Many studies in the western Mediterranean were mainly conducted in Italy, Spain and France.While the prosperity of countries in the western part of the basin could explain the skewness in the distribution of publications in the Mediterranean Sea, the large number of ROV studies in Italy could also be linked to local ROV production (Global Electric Italiana).Meanwhile, Spain run a large number of RUV studies, half held at a permanent observatory system (OBSEA), a cabled system established in Barcelona, providing permanent visual material of the surroundings (Aguzzi et al. 2011).
A rather positive finding in the present review was the extended use of underwater video techniques in MPAs.Underwater video methods form an effective alternative to traditional sampling techniques, such as commercial or experimental fishing, and especially in the Mediterranean Sea, a biodiversity hotspot (Coll et al. 2010) with only 6% of the total area under protection (Claudet et al. 2020), such initiatives need to be further encouraged and become well established to secure habitat and fish population integrity.
The use of ROV has been the most common video method in the Mediterranean Sea, particularly for benthic studies, despite the high purchase and operational costs (Sward et al. 2019).Such studies usually focused on describing coral (Fabri et al. 2019) or other benthic habitats (Bo et al. 2012b) combined with ecosystem-associated fish.This method will probably become even more popular in the upcoming years with more widespread use of mini-ROV (e.g., Nalmpanti et al. 2021); due to its low-cost and easy operation (Rubin 2013).RUV and DOV systems were also common techniques, especially in behavioral studies.For example, in experimental settings a fixed camera is beneficial to allow recording the fish reaction to an altered condition (e.g. in Paxton and Smith 2018) and in studies where the fish need to be followed, a diver is a more appropriate option (e.g. in Buñuel et al 2020).
Besides the research objectives that determine the technique choice, experimental depth is also an aspect that needs to be considered when selecting the appropriate technique.
Underwater video can reach a wide range of depths (Sward et al. 2019;Whitmarsh et al. 2017), but most studies in the Mediterranean Sea have been conducted above 500 m depth.Especially DOV techniques were restricted in depths above 62 m due to diver capability limitations, as for depths extending the limits of conventional SCUBA special equipment and specialized divers are required (Soldo and Glavičić 2020).RUV systems were also limited to shallow waters, but BRUV and ROV techniques extended deeper.The choice of BRUV is appropriate in greater depths, where the occurrence of motile fauna is sporadic, since the bait plume can attract individuals from extended areas (Cappo et al. 2003).Indeed, in our review bait was always included when deploying RUVs in depths greater than 45 m.All BRUV systems Vol.: (0123456789) deployed in depths greater than 350 m were landers (Capezzuto et al. 2012) or submersibles (Jones et al. 2003) fixed on the seafloor.These apparatuses are designed to withstand high pressure (Liang et al. 1998) and operate for long periods of time (Kononets et al. 2021).In contrast, simple apparatuses, which consist of a camera with short battery life, placed inside its housing and then fixed on a steel frame, were deployed in shallower depths (Díaz-Gil et al. 2017;Cattano et al. 2021).Similar to landers and submersibles, ROVs are also capable of reaching deeper regions (Sward et al. 2019).Indeed, their usage in the Mediterranean Sea extended in a wide range of depths, although the nature of many ROV studies, which focused on specific habitats, set them usually in deeper locations.Such environments were cold-water coral ecosystems (Taviani et al. 2017;Moccia et al. 2019), submarine canyons (Fabri et al. 2014;Ayma et al. 2016) and bathyal seamounts (Bo et al. 2020a).
With the exception of one study focusing on pelagic fish, all other publications were restricted to recording fish close to the sea floor or right above it.Fish studies carried out in the water column are scarce probably due to the challenging nature of the sampling technique, as very high sampling effort is required to record few specimens (Torres et al. 2020).The absence of pelagic video samplings does not help with enhancing data collection on large pelagic fish; their assessment is still limited despite worldwide population declines (Letessier et al. 2015).
Other sections of the Mediterranean that were understudied besides pelagic systems were the deep-sea habitats; the reviewed studies were mostly restricted in waters above 500 m.Nonetheless, fish in the deep-sea ecosystems have slowly started to be harvested and thus, their status, ecosystem function and services should be examined (Armstrong et al. 2012) despite the financial constraints of exploring these regions.Additionally, studies in coralligenous habitats, the second most important habitat in the Mediterranean Sea for species diversity after Posidonia oceanica meadows (Boudouresque 2004), were also limited.The increase of video studies in these habitats is not only essential due to their high ecological value, but also because these habitats are vulnerable to destructive sampling techniques, such as fisheries-dependent methods.

Technical considerations
The present review revealed that sampling protocols can differ widely among publications.Bait was not consistent among BRUV studies, although clupeid fishes that are fairly cheap to purchase and contain high amount of oil (Dorman et al. 2012) were a usual choice.Soak duration ranged from a few minutes to hours.RUV systems were used for short deployments (in experimental settings such as in Alós et al. 2015b) or even months in the case of the permanent observatory system OBSEA (Aguzzi et al. 2011), while BRUV systems were usually deployed for around 30 min, a duration that has been proposed in literature as the minimum time required to reach MaxN, the maximum number of individuals recorded in a single video frame (Díaz-Gil et al. 2017).At the same time, maximum deployment duration was defined by battery life capacity (Follana-Berná et al. 2020).In transect studies usual transect lengths ranged between 25 and 50 m for DOV systems, while ROVs were deployed to cover a large variety of transect lengths.
Considering that there is need for comparable data in spatial and temporal scales among different research groups, the multiple sampling settings mentioned before, form an obstacle to the homogeneity of the results.In addition, experiments producing accumulation curves (Stobart et al. 2007) to find the adequate sampling effort are time and money consuming and it is thus, common not to follow this preparatory step before sampling.The above create a need for technical papers focusing on assessing the sampling effort and procedure required to obtain "good enough" and comparable data.Therefore, it would be beneficial for future studies to focus on technical experiments related to the deployment duration (such as Stobart et al. 2007 for BRUV systems), length and width of transects, number of replicates required and type and amount of bait in different habitats, depths and target species.Publications should include reproducible and straightforward sampling protocols with detailed information regarding the methodology, such as details regarding the video system, deployment and sampling design and analysis, as suggested for BRUV studies (Whitmarsh et al. 2017).
In addition, there is need for more methodological studies aiming to standardize the sampling effort per technique, to make possible the complementary use of different techniques in the future.Studies that simultaneously use techniques of different nature, such as fisheries-dependent, sedentary or transect techniques have yielded higher fish diversity observations (Lowry et al. 2011;Schramm et al. 2020).Despite the benefits, when outputs are expressed in different units, for example in a transect and a sedentary sampling, direct comparisons are challenging.

Conclusion
The current review clearly reveals that the use of underwater video in the Mediterranean Sea has been increasing rapidly in fish ecological studies.Their use expands from community analysis to behavioral and methodological approaches.Despite that, the eastern and southern parts of the basin remain understudied.Deep-sea and pelagic habitats, as well as coralligenous ecosystems are also regions of high ecological value with large research gaps.Finally, methodological studies aiming to optimize and standardize sampling effort are generally lacking but are necessary.In the future technological progress is expected to give more opportunities to explore the uses of underwater video and fill research gaps especially in ecologically important regions that need protection.

Fig. 1
Fig. 1 The number of underwater video studies in the Mediterranean Sea for each of the identified techniques published prior to 31 December 2021.Underwater video techniques include Baited Remote Underwater Video (BRUV) and Remote Underwater Video (RUV) systems, Remote Operated Vehicle (ROV), Diver Operated Video (DOV) systems, and other (techniques not included in any of the above groups)

Fig. 2
Fig. 2 Distribution of underwater video publications in the Mediterranean Sea per Geographical Subarea (GSAs) and frequency of underwater video techniques per Mediterranean division as indicated by the pies.Circle sizes indicate proportionally the number of studies in each Mediterranean division.Techniques include Baited Remote Underwater Video (BRUV), Remote Underwater Video (RUV), Remote Operated Vehicle (ROV), Diver Operated Video (DOV), other (techniques not included in any of the above groups).Numbers show the geographical subareas (GSAs) according to FAO (see Materials and Methods)

Fig. 3 a
Fig. 3 a Frequency of underwater video uses in the Mediterranean Sea by depth and b Depth distribution for each video technique.Underwater video techniques include Baited Remote Underwater Video (BRUV), Remote Underwater

Fig. 4
Fig. 4 Habitat types identified in the reviewed publications

Fig. 5 a
Fig. 5 aThe soak time (min) in Baited Remote Underwater video (BRUV) and Remote Underwater Video (RUV) studies."Range" includes studies mentioning the minimum, maximum or range of deployment duration, "unspecified" refers to non-provided information and "pictures" represents the cases in which frames were extracted from the video footage.b The transect length (m) in Diver Operated Video (DOV) and Remotely Operated Vehicle (ROV) studies."Range" includes studies using various transect lengths, and studies in which no information was given are placed in the "unspecified" group

Table 1
Type of information that was derived from the video footages as identified from selected publications Vol:. (1234567890)