Unleashing the Potential of the 360° Baited Remote Underwater Video System (BRUVS): An Innovative Design for Complex Habitats

Abstract

Coastal ecosystems are vital for numerous demersal and benthopelagic species, offering critical habitats throughout their life cycles. Effective monitoring of these species in complex coastal environments is essential, yet traditional survey methodologies are often impractical due to environmental constraints like strong currents and high wave regimes. This study introduces a new cost-effective Baited Remote Underwater Video System (BRUVS) design featuring a vertical structure and 360° cameras developed to overcome limitations of traditional BRUVS, such as system anchoring, overturning, and restricted frame view. The new design was compared against a previous one used on the northwest Iberian coast. Key performance metrics included species detection, habitat identification, and operational efficiency under complex hydrodynamic conditions. Findings reveal that the two designs can effectively identify the common species typically observed in the study area. However, the new design outperformed the previous by significantly reducing equipment losses and anchoring issues. This enhancement in field operations’ simplicity, operability, portability, and resiliency underscores the new system’s potential as a cost-effective and efficient tool for demersal and benthopelagic ecological surveys in challenging coastal seascapes. This innovative BRUVS design offers advanced monitoring solutions, improving habitat assessment accuracy and responsiveness.

1. Introduction

Baited Remote Underwater Video Systems (BRUVS) have become a commonly used method for non-intrusive sampling of marine life in various marine environments. Essentially, a BRUVS unit consists of a waterproof video camera affixed to a frame, with a bait bag positioned at the front of the frame to lure fish into the camera’s field of view. Once the unit is deployed into the water from a boat, it is left on the seabed to record for a predetermined time before being recovered. This methodology typically uses regular action cameras with a restricted field view (130° ≤ 170° angle view), and the physical structure is formed by a metal or Polyvinyl Chloride (PVC) [1,2,3,4,5,6]. This method is beneficial when traditional sampling techniques, such as diver surveys or trawling, are not feasible due to environmental or area conditions (e.g., high wave and wind regime, the vulnerability of habitats, marine protected areas (MPAs) survey, presence of strong currents) and eliminates sampling biases that may arise from diver presence responses [7,8,9,10,11]. BRUVS have been used for various research objectives, including capturing spatial and temporal variation in species abundances [4], monitoring population abundances [12], biodiversity [13], or assessing fish-habitat associations to understand species distributions [10,11,14]. However, as we found in our previous surveys and field campaigns [15], BRUVS structures can easily anchor, become damaged, or even be entirely lost due to the different contact points with the complex rocky bottoms, especially in highly hydrodynamic coastal areas. One of the most sensitive moments to anchoring or losing the BRUVS occurs during recollection. This sensitivity is due to the technique and angle at which the system is pulled, which, combined with the wave and boat movement, bottom currents and the presence of rocky bottoms, creates the ideal scenario for the system to anchor, resulting in the loss of essential data. Additionally, the typical cameras used in BRUVS have restricted fields of view and record a small portion of the environment, thereby limiting and biasing the data produced. Therefore, it is crucial to be aware of these issues and take steps to facilitate sampling campaigns to minimise their impact on data collection.

In light of these challenges, the emergence of 360° video cameras has revolutionised filming. As camera technology improves, costs decrease, making it more accessible to the scientific community to use this type of camera for different purposes [10,16,17,18,19,20,21] since they significantly expand the field of view. Leveraging this technology, we propose a new BRUVS design using 360° cameras. The new design should allow additional visual data collection, such as surrounding habitat characterisation, while reducing the necessary logistics, time spent, and monetary cost. Most importantly, it should minimise the risk of entanglement and loss, thereby enhancing the efficiency and effectiveness of data collection.

With this study, our primary goal is to develop and validate a suitable BRUVS structure design and configuration for complex coastal habitats. This development is crucial for optimising surveys of demersal and benthopelagic assemblages in environments such as the northwest (NW) Iberian Coast, characterised by complex heterogeneous habitats and exposed coastlines [10,22,23,24]. The specific aims of this study are: (i) to introduce a new BRUVS design (new design, ND) that utilises low-cost 360° cameras with a structured design less prone to entanglement and (ii) to compare its performance with a previous structure design and configuration (previous design, PD), previously used by our team in the same study area. We aim to simplify the deployment and survey process with the new structure and configuration, resulting in an optimisation of the data collection and a substantial reduction in equipment loss and field sampling time. This simplification aims to contribute to the optimisation of benthopelagic demersal species surveys in complex habitats, making the processes more efficient and effective.

2. Materials and Methods

2.1. Study Site

We conducted the sampling campaigns in Viana do Castelo, Portugal, on the NW Iberian coast, during the spring-summer months of 2019 and 2020. This coastal area is characterised by a complex mixture of sandy and rocky substrates, mainly composed of schists and quartzites. These complex fractured rocky bottoms mixture associated with variable depths creates a challenging environment and topographic complexity for typical BRUVS structure designs. Although the rocky part mainly comprises compact, non-fragmented rock platforms, this area’s erosion led to a topographic heterogeneity (e.g., canyons and crevices). This complex topography associated with the presence of rigid substrates facilitates device anchoring. Alongside with its unique geological features, the vertical and westward-facing orientation of this coastline subjects it to different phenomena such as (a) intense-wave energy regimes, especially during the winter where wave heights can go up to 11/12 m, (b) intense groundswells, (c) the southward surface “Portugal Current”, (c) intense wind patterns and (d) seasonal upwelling events [24,25,26,27]. These factors significantly generate high-productivity waters that lead to low visibility and demanding diving and nautical conditions for most of the year [10,23,28]. These combined factors are the primary reason for the scarcity of studies on this region.

2.2. Sampling Design BRUVS

In order to cover the existing main structural habitats, 100 BRUVS were randomly deployed at depths between 5 and 30 m using the two designs along the same sampling area: 50 of the previous design (PD) and 50 of the new design (ND). Deployments occur on different days across the two spring-summer periods (2019 and 2020) to ensure that the data encompasses a representative sample of the variability in environmental conditions and species presence.

2.2.1. Previous Design (PD) BRUVS

The previous design was developed by adapting the setup used by Roberson et al. [5], due to its simplicity of construction. In this design, a stainless-steel tube is welded into the base horizontally (1 m long), and another crosses the middle (0.5 m). The two vertical arms stay connected diagonally by another tube and are equipped with steel plates (24 and 35 cm) to hold the camera (G-EYE 900 (4K/Full HD), Decathlon) and the bait container. Bait containers with sardines are positioned in front of the video system to attract animals close to the camera, thus improving the identification of smaller species. The camera’s platform is 30 cm above the system base and connected to a surface buoy (Figure 1). PD BRUVS deployments were separated for at least 200 m to avoid overlap of bait plumes and reduce the probability of animals moving between sites within the sampling period [29]. The cameras were active for 45 min each.

2.2.2. New Design BRUVS Deployment

The previous BRUVS design proved impractical in our study area due to its complexity. Aiming to develop a more suitable BRUVS, we created a new design using Insta 360 One X cameras (Arashi Vision Inc., Irvine, CA, USA), instead of the regular action cameras. Due to the physical structure of this type of camera, the new BRUVS platform could be simplified. In this design, the camera stays threaded to a steel plate (10 × 7 cm²) attached to a 2 kg diving weight and connected to a 50 cm steel chain. The bait container (with sardines) is positioned 50 cm from the camera, on the other end of the chain, with a small floater, allowing it to stay suspended in the water, away from the bottom. The weight below the camera forces it always to be upright (Figure 2). The cameras were active for 45 min each.

2.3. BRUVS Design Performance

The data collected from the field deployments and video analyses were systematically compared to evaluate the performance of the two designs. Metrics included deployment time, equipment resilience, species detection, and habitat visualisation capabilities.

2.3.1. Time Consumption and Design Resilience

Deployment time per camera was calculated for both designs, considering the entire operational cycle (deployment, recording period of 45 min and recollection). This allowed for a comparison of field efficiency between the designs. The resilience of each BRUVS design was assessed based on the following criteria:

(a)

System anchoring: The stability of the BRUVS on the seabed was observed (e.g., overturning or drifting).

(b)

Equipment Damage: Any damage to the BRUVS components after recollection was documented.

(c)

Equipment Loss: Instances of equipment loss, such as cameras or other critical components, were documented.

2.3.2. Video Analysis

The video footage from each BRUVS deployment was analysed to compare the performance of the two designs in capturing marine life and habitat information:

(a)

Angle of View: The field of view provided by the cameras in each BRUVS design was assessed to understand the type of information that could be collected.

(b)

Species Detection: The presence and identification of species were compared between the two designs. Species were identified based on distinctive morphological characteristics visible in the video, such as body shape, coloration, and behavior by the same observer. Reference materials, including regional identification guides and online databases (e.g., FishBase), were used for cross-referencing and verification. Individuals with ambiguous features were excluded to ensure accuracy.

(c)

Habitat Visualisation: The ability to visualise and understand the surrounding habitat during the video processing’s was evaluated.

(d)

Video processing facilities: The capability of video software to processing, including multiview and zoom facilities, was accessed for each design.

3. Results and Discussion

3.1. BRUVS Performance Comparisons

BRUVS deployments and field campaigns resulted in approximately 300 h at sea and 4500 min of video footage collected throughout several months, allowing the capture of various environmental conditions and species interactions.

Table 1. Field and video analyses performances of PD and ND. Average deployment time/per camera—deployment + recorded (45 min) + recollection time; Lost cameras—cameras lost in the field; An-choring—BRUVS setup anchored but later recovered; Common species—identification of the fre-quent species found in the area; Surrounding habitat—surrounding habitat perception; Angle of view—view processing angle; Zoom in/zoom out ≠ angles—possibility to zoom during video pro-cessing in different angles in the same frame or zoom out (tiny planet view).

	PD	ND
Average deployment time/per camera	≈1 h 45 min	≈1 h
Lost cameras	6	0
Anchoring	25	0
Common species	yes	yes
Surrounding habitat	challenging	straightforward
Angle of view	fixed frame < 180°	variable frame—360° up/down
Zoom in/zoom out ≠ angles	no	yes

compares the performances of both designs in the field and during video analyses. The common species in the study area could be detected with both designs. However, the ND consistently outperformed the PD across several key metrics. The ND demonstrated significant advantages in deployment efficiency, requiring approximately half the time per camera compared to the PD (PD ≈ 1 h 45 min vs. ND ≈ 1 h).

Furthermore, with ND, we did not experience any equipment damage or loss, which starkly contrasts with the PD, with which we lost six cameras and encountered anchoring issues in 25 deployments. The significant difference in the field and time deployment of BRUVS is a direct result of the simplification of the BRUVS structure in the ND which also enhanced its portability. In terms of video analysis capabilities, the ND’s enhanced angle of view (up to 360° compared to PD’s fixed < 180° frame (Figure 3)) facilitated a more comprehensive understanding of the surrounding habitat (Figure 4 and Figure 5b).

Furthermore, the ND offers superior flexibility in video processing software, allowing for zooming in/out at different angles, viewing the same frame in multiple ways, and utilising multiview functionality (such as tiny planet view, standard view, and natural flat view) [10]. These features allow for a better representation of the area and identification of smaller or camouflaged species (Figure 4) and significantly enhance habitat categorisation and species identification during video processing. This comprehensive representation and identification of the study area and its inhabitants allow for a more direct association between the habitat and species with the ND.

3.2. Discussion

Our results indicate that both PD and ND designs effectively capture the common species of the study area. The slight differences observed in the number of species detected can be attributed to natural variability during the sampling days (

Table 2. List of species identified along the 100 deployments between the two BRUVS designs. (√—specie identified, X—specie not identified).

Species	PD	ND
Ctenolabrus rupestris	√	√
Symphodus bailloni	√	√
Labrus bergylta	√	√
Coris julis	√	√
Symphodus melops	√	X
Symphodus roissali	√	X
Pollachius pollachius	√	X
Trisopterus luscus	√	√
Diplodus vulgaris	√	√
Spondyliosoma cantharus	√	√
Diplodus sargus	X	√
Conger conger	√	√
Dicentrarchus labrax	√	√
Serranus cabrilla	√	√
Loligo vulgaris	√	√
Pomatoschistus pictus	√	X
Parablennius gattorugine	√	√
Parablennius pilicornis	√	√
Trachurus trachurus	√	√
Trachurus picturatus	√	√
Chelon auratus	√	√
Chelon labrosus	√	√
Phycis phycis	X	√
Chelidonichthys lucerna	X	√
Belone belone	√	√
Mullus surmuletus	√	√
Hyperoplus lanceolatus	X	√
Zeugopterus punctatus	X	√
Raja undulata	√	X
Balistes capriscus	√	X
Pseudocaranx dentex	√	X
Scomber colias	√	X
Necora puber	X	√
Carcinus maenas	√	√
Octopus vulgaris	X	√
Total species identified	28	27

). However, the ND design significantly reduces equipment losses and anchoring issues, making it more advantageous. The simplified and optimised logistic of the new design enhance the quality of field campaigns, particularly in high-energy and complex coastal areas like the NW Iberian coast.

The enhanced performance of the ND design carries profound implications for marine research. Researchers can now conduct more frequent and extensive surveys by minimising equipment loss and simplifying deployment processes. This increased survey frequency and coverage can lead to more accurate data on species distribution and habitat use, thereby informing more effective conservation strategies and management policies. Moreover, the comprehensive 360° view offered by the new cameras facilitates better habitat characterisation [10,18], a key factor in understanding the ecological dynamics of the study areas.

Despite the widespread use of traditional BRUVS designs in marine biodiversity studies worldwide, many of the coastal sites where they are used meet specific conditions that facilitate their deployment, such as a stable hydrodynamic regime, a heterogenous depth, a low wave regime, or previous cartographic surveys that aid in the choice and knowledge of suitable BRUVS placement sites [30]. Also, the typical angle range of the regular action cameras used in the typical BRUVS designs makes the identification quite restrictive, and understanding the surrounding habitat is also challenging. Nevertheless, animals and habitats must be detected and identified to effectively monitor their distribution and abundance during surveys, regardless of the study site’s biogeographical and environmental conditions, especially in exposed areas.

In the NW Iberian coast, specifically on the northwest coast of Portugal, the typical survey methodologies—divers surveys and trawling—are not viable options due to the complex topography and environmental conditions [10,23,24,25,27] that compromise the security of the survey and have a more negative impact on the ecosystem. Thus, BRUVS is one of the most suitable and low-impact methods for collecting site data in this area. However, various constraints affect the results of typical BRUVS in areas similar to our study site, and this is why most published studies with BRUVS are developed in the Southern Hemisphere, where guidelines for using these techniques exist compared to the North [30]. Even previous studies developed on our coastline need to be complemented with other methods (e.g., visual census by divers and ROV, extraction methods, fisherman’s surveys), in order to survey demersal and benthopelagic species [31]. These BRUVS limitations can be overcome by implementing 360° cameras in BRUVS and their structure simplification. The performance of ND is mainly due to the fact that in this design, the BRUVS structure has only one contact point to the bottom, making it less prone to anchor, especially during BRUVS recollection (Figure 5a).

Additionally, the 360° view provided by the cameras and the video processing software (Insta360 studio 2021, version 3.5.8) allows for multiview options [10], making it possible to identify smaller, distant, or camouflaged species (Figure 4). Using 360° cameras in BRUVS facilitates a comprehensive view of the sample site, improving the assessment of species and surrounding habitat. Furthermore, even if the BRUVS structure falls into a crevice, the 360° view makes identifying species possible since we can observe in a top-point view, allowing the use of the video samples. During the placement and recollection of BRUVS, the 360° view allows for the assessment of the global habitat and provides a more comprehensive view of the study site (Figure 4 and Figure 5b). These possibilities were previously hard to achieve with standard action cameras (angle view 130° ≤ 170°) [32], like the one used on the PD (Figure 3).

Understanding the surrounding habitat and identifying species with more detail allows us to collect much more information during the same work window for a given sampling point. New possibilities emerge with the 360° cameras’ post-processing software, with multiple angle view options (e.g., tiny planet view, fisheye view and natural view (flat view)), allowing the identification of small, rare and camouflaged species (Figure 4). This is possible because, in the post-processing of 360° videos, we can evaluate the habitat from any given angle and zoom in and out to view the same point from different perspectives, angles and distances. Therefore, the cameras used in ND significantly improve over regular action cameras, as concluded in previous studies [33,34].

Despite these advantages, the ND design may face some challenges in extremely turbulent waters or areas with heavy sedimentation that could obscure the camera’s view. Future research should explore different 360° camera models to identify the most cost-effective and reliable options to standardise the camera used. Testing the ND design in various environments, such as estuaries, rivers or deeper marine habitats, will help assess its versatility. Furthermore, integrating additional sensors, such as those measuring temperature or salinity, could provide a more comprehensive understanding of the surveyed habitats. Investigating the long-term durability of the ND design in diverse conditions will also be beneficial. Moreover, this type of high-definition video produces large data files [35] that can be demanding and time-consuming to process. With the progress of efficient artificial intelligence (AI) that can assist underwater video processing and enable efficient data analysis, this problem can be solved in future years due to artificial intelligence (AI) capabilities, e.g., machine learning or pattern recognition [36].

We need methods and designs that simplify field logistics and facilitate sampling consistency, which is critical in ecological studies, primarily when BRUVS are used [30]. Surveys with these characteristics are essential to promote future effective conservation measures, create marine reserves (MPAs), and support no-take zones, allowing an easy understanding of the distribution and diversity within an area and their correlation with the seabed and habitat [14,37]. This understanding is more critical in coastal areas since there are many species with different foraging strategies [37]. Also, there is a need for cost-efficient, practical and simplified methods for monitoring demersal benthopelagic assemblages in these areas where trawlers and scuba diving transepts for long-term monitoring are the most used methods but not the most appropriate or efficient [38,39].

During this study, we found several strengths of the ND approach, underscoring its value as a monitoring tool for demersal benthopelagic species and their habitat in coastal marine ecosystems. Although we acknowledge the technical camera-related limits, which we can categorise as the main limitations of the ND design (Figure 6), they should not overshadow the benefits (e.g., portability, 360° angle view). These limitations should be well recognised and managed to promote ND adoption in more environments. The benefit of implementing an ND video survey, especially for monitoring demersal and benthopelagic species in complex areas, remains significant and outweighs the constraints listed (Figure 6).

While this study focuses on the performance comparison between two BRUVS designs, it is valuable to consider the broader context of actual marine research methodologies. BRUVS and ROVs (Remotely Operated Vehicle), for instance, are two commonly used tools that, although serving different primary purposes, can complement each other in assessing fish assemblages and habitat complexity. BRUVS are particularly effective for in-situ observations of demersal and benthopelagic species, providing detailed data on species composition, behaviour, and interactions within specific underwater locations. Bait survey techniques also typically show significant differences in fauna composition. BRUVS allows researchers to obtain high-resolution video footage essential for identifying species at different sites during the same work window and studying their ecological roles in the marine environment with minimal disturbance. In contrast, ROVs are primarily utilised for mobile surveys, offering a flexible perspective of marine habitats. ROVs are adept at navigating complex underwater environments, capturing data on habitat structure, and assessing large-scale spatial patterns. This method is invaluable for mapping habitats, monitoring environmental changes, and detecting broad patterns in habitat use by marine organisms. However, the mobility and components of ROVs can sometimes influence species’ behaviour (combination of light, sounds and speeds), which are factors to consider when interpreting data collected by this method since, together, they can create a complex array of stimuli that would likely elicit a variety of behavioural responses and species distribution.

Schramm and colleagues [40] pointed out avoidance behavioural responses of fish to ROVs as the possible cause of fewer species identifications compared to BRUVS, particularly over soft sediment habitats (where fewer refugees exist). Bond et al. [41] also underscore the need to develop more studies regarding species’ interactions with commercial importance ones and ROVs to understand species’ responses to ROV components and movement stimuli. BRUVS also have been shown to have higher statistical power than the transect methods performed with ROVs, probability because transect methods have higher sample sensitivity in survey habitat heterogeneity within a sample unit compared to BRUVS [42]. Results from Jessop et al. [43] also showed differences in assemblages of BRUVS when compared to transect methods, with higher abundances and species richness, including commercially important ones, identified with BRUVS. These results highlight the importance of understanding the broader context of these two methodologies and their potential integration in marine research. Combining these approaches can yield a comprehensive understanding of marine ecosystems. For instance, ROVs can pinpoint areas with complex habitats, which can be explored in detail using BRUVS to survey the demersal and benthopelagic communities within those habitats.

4. Conclusions

Developing low-cost technologies that allow us to obtain different data using a simplified methodology—promoting continuous and productive research in marine ecosystems—is still necessary. Even though continued research is still required to optimise this system, we have shown that the new design (ND) is a new step in monitoring highly complex hydrodynamic habitats. Eventually, as methods improve and become standardised, researchers and natural resource managers will have access to different data types with much more detail at a fraction of the cost, simplified logistics, and less time-consuming survey methods. However, as with any technique, we recognise that no methodology can provide a high range of detection across various species, habitats and depth conditions. There will always be poor outcomes, gaps in data, and missing information in ecological studies. Nevertheless, if we can reduce logistical constraints on deployment and costs while increasing the capacity to collect information, we will contribute to coherent and effective conservation decisions.