New possibilities for research on reef fish across the continental shelf of South Africa

Subtidal research presents numerous challenges that restrict the ability to answer fundamental ecological questions related to reef systems. These challenges are closely associated with traditional monitoring methods and include depth restrictions (e.g. safe diving depths for underwater visual census), habitat destruction (e.g. trawling), mortality of target species (e.g. controlled angling and fish traps), and high operating costs (e.g. remotely operated vehicles and large research vessels).1 Whereas many of these challenges do not apply or are avoidable in the shallow subtidal environment, the difficulties grow as one attempts to sample deeper benthic habitats. This situation has resulted in a paucity of knowledge on the structure and ecology of deep water reef habitats around the coast of South Africa2,3, and in most marine areas around the world4,5. Furthermore, the inability to effectively survey deep water benthic environments has limited the capacity of researchers to investigate connectivity between shallow and deep water habitats in a standardised and comparable fashion.6

New possibilities for research on reef fish across the continental shelf of South Africa Subtidal research presents numerous challenges that restrict the ability to answer fundamental ecological questions related to reef systems.These challenges are closely associated with traditional monitoring methods and include depth restrictions (e.g.safe diving depths for underwater visual census), habitat destruction (e.g.trawling), mortality of target species (e.g.controlled angling and fish traps), and high operating costs (e.g.remotely operated vehicles and large research vessels). 1Whereas many of these challenges do not apply or are avoidable in the shallow subtidal environment, the difficulties grow as one attempts to sample deeper benthic habitats.This situation has resulted in a paucity of knowledge on the structure and ecology of deep water reef habitats around the coast of South Africa 2,3 , and in most marine areas around the world 4,5 .Furthermore, the inability to effectively survey deep water benthic environments has limited the capacity of researchers to investigate connectivity between shallow and deep water habitats in a standardised and comparable fashion. 6th the recent advent of sophisticated and cost-effective remote sampling methods suitable for deep water research, 1 ecologists have been able to describe finer-scale patterns of reef ecosystems in both deep and shallow waters.][9][10][11][12][13][14][15][16] The baited remote underwater stereo-video system (stereo-BRUVs; Figure 1) has been at the forefront of these developments [17][18][19] , and has emerged as the most comprehensive, precise and cost-effective tool to measure the ecosystem effects of fisheries 14,20 , and patterns in fish abundance 7,8,15,21  For the first time in South Africa, through a collaborative project of the Elwandle Node of the South African Environmental Observation Network (SAEON), the South African Institute for Aquatic Biodiversity (SAIAB), Rhodes University, the University of Western Australia, and Curtin University (Australia), stereo-BRUVs research on reef fish assemblages is being conducted within the Agulhas Ecoregion from the shallow subtidal area to the edge of the continental shelf.The purpose of this piece is to place in context the necessity of standardised research on the populations of reef fish across the continental shelf of South Africa, to put forward the case for employing stereo-BRUVs in this research, and to introduce the South African marine science community to the research possibilities available with stereo-camera systems.
operated vehicles (ROVs) and autonomous underwater vehicles (AUVs) equipped with video cameras are being used more frequently for research.However, both ROVs and AUVs are expensive and logistically complicated to operate and have been largely restricted to oceanographic (AUVs) or exploratory (ROVs) research.Both RUVs and BRUVs are deployed from a research vessel with the system resting on the sea floor or in the water column, while tethered to a surface marker buoy or to the research vessel.The only distinction between these two approaches is that BRUVs use a food-based attractant to draw fish into the field of view of the camera.DOV is synonymous with underwater visual census (UVC), except that the survey is recorded and the researcher identifies and counts fish post sampling from the video footage, as opposed to identifying and counting the fish while underwater (as in UVC).Stereovideo sampling is a variation of mono-video camera techniques (RUVs, BRUVs and DOVs) that allow for fish lengths to be measured and the survey area to be quantified, greatly increasing the data output per sample and the value of the data for studying the effects of fishing. 14nce the mid-1990s there has been an exponential increase in the number of research articles published in peer-reviewed scientific journals that employed BRUVs, RUVs, DOVs or their stereo-video equivalents to collect fish assemblage data from subtidal reefs across continental shelves (Figure 2).Stereo-video sampling techniques first emerged in the early 1980s 28 ; however, it was only in the mid-1990s that development of the remote stereo-video approach was initiated by researchers at Otago University and Melbourne University. 24Only in the last 4 years has the method expanded globally, resulting in a rapid increase in the number of publications based on data collected with this approach (Figure 2).Further growth in the publication rate can be expected over the next few years as the tools become more available and awareness of the new research possibilities they provide increases.
In general, results show that the stereo-BRUVs technique outperforms all other available methods with the data characterised by low levels of variability, high levels of species richness, high abundances of species targeted by fisheries and accurate information on the size structure of these populations. 14,18,19,24,30,36mote video sampling techniques are flexible in that the systems can be deployed with or without bait, or with the bait type varying between units or deployments, and can be tailored to address a multitude of questions by targeting specific components of the fish community.As such, remote video sampling techniques reduce the research footprint and provide data capable of addressing research and management questions across geographical and depth gradients, within special management areas (e.g.no-take marine protected areas, MPAs) and across different habitats.Importantly, the feasibility of long-term standardised monitoring will be increased significantly by the cost-efficiency of the method.
One of the major advantages of stereo-video over mono-video techniques is that it allows for precise length measurement of fish underwater.
This advantage means that the potential for observer error is avoided (compared with UVC) and the need to remove fish from the water, which frequently leads to barotrauma, is eliminated (compared with controlled angling and fish traps). 17,37,38The size distribution of a fish community is a more sensitive measure of fishing pressure than abundance, as larger individuals of a species are typically more aggressive and caught first when fishing. 1 Research from Western Australia has demonstrated that the size data collected with stereo-BRUVs are comparable to fisheries-dependent data sources and are therefore potentially useful for informing management of stocks. 36,39In addition, size-based indicators, such as mean length in a population or community, mean maximum length in a community, and the slope and intercept of size spectra, are particularly useful for long-term ecosystem-based fisheries management (EBFM). 40,41Furthermore, the use of stereo-video also allows the distance from the cameras to a fish to be measured and hence the area sampled can be standardised. 42er the last 4 years, single camera remote underwater video systems (RUVs and BRUVs) have been successfully employed to survey the reef fish communities in the Tsitsikamma National Park, Still Bay National Park and Table Mountain National Park MPAs in South Africa. 31,43,44The results suggest that BRUVs are highly suitable for obtaining relative abundance data for most of the reef fish species occurring in the Agulhas Ecoregion.Furthermore, data collected in the Tsitsikamma National Park MPA, using six different techniques, show that BRUVs detected 92% of all bony fish and 71% of all cartilaginous species recorded by the six methods, compared with 68% and 43%, respectively, for the second best method -RUVs (Table 1). 45Interestingly, in the same study UVC recorded zero cartilaginous species and only 60% of the bony fish species, even though it had three times the sampling effort compared with both RUVs and BRUVs. 45These results further emphasised the efficiency of BRUVs, with the method requiring only 21 samples per annum to detect a 10% increase in the abundance of important commercial fisheries species, compared with 49 samples required by RUVs and 72 samples required by UVC. 45 During 2013, stereo-BRUVs were deployed for the first time in South Africa (and Africa), to collect fish assemblage data within the Tsitsikamma National Park MPA.A total of 194 samples, equating to 388 h of video footage, was collected from inside and outside the MPA using four units during 17 sampling days.Sampling depth ranged from 6 m to 80 m, encompassing the deepest extent of the reef habitat inside the MPA and equivalent habitat outside the MPA.All sampling was conducted off an 8-m semi-rigid ski-boat fitted with a simple capstan winch to retrieve the weighted systems.Preliminary analysis of 36 samples has resulted in the detection of 41 species and a total of 3644 size measurements, showing that the stereo-BRUVs not only provide the same benefits as the mono-camera BRUVs, but also produce a considerable amount of length frequency data (±100/sample).

Importance of standardised monitoring
7][48][49] The single or multi-species approach to traditional management is considered by many to be outdated, particularly for reef fish, and there is a drive to implement holistic EBFM. 3,47,48,50Similarly, the assumption that managers can rely on resource users to abide by regulations has been repeatedly disproved, compelling managers to adopt approaches that are feasible to manage and enforce, such as notake MPAs. 3,48,49osystem-based fisheries management requires sound knowledge on the ecology of the natural systems being managed. 50,51Because of the vulnerable nature of populations of over-exploited fish species, monitoring methods need to be non-destructive to remove the possibility of additional impacts on stocks.][54] In South Africa, this applies for most of the important endemic reef dwelling species targeted in the line-fishery industry, with some occupying a depth range in excess of 150 m. 52,53 This is true for species such as carpenter (Argyrozona argyrozona), dageraad (Chrysoblephus cristiceps), red stumpnose (C.gibbiceps), roman (C.laticeps), black musselcracker (Cymatoceps nasutus), blue hottentot (Pachymetopon aeneum), hottentot seabream (Pachymetopon blochii), red steenbras (Petrus rupestris), scotsman (Polysteganus praeorbitalis) and panga (Pterogymnus laniarius), many of which are considered over-exploited and are listed by the South African Sustainable Seafood Initiative as species of concern or as unsustainable. 48,49,55Adequate monitoring across these broad distributional ranges in a standardised and comparable fashion is required to provide data that promotes effective management of such species.The remote underwater stereo-video system, whether baited or unbaited, offers an opportunity to meet this need.

Conclusions
There is a growing global recognition that high-resolution, nondestructive and in-situ stereo-video techniques can provide improved understanding of fine-scale ecology on deep and shallow reef habitats, and deliver data that support effective EBFM.Preliminary work in South Africa has proved that the method can be cost-efficiently employed by small research teams working off small vessels.The stereo-video research platform developed at SAIAB and SAEON will be able to operate down to depths of 250 m, covering the entire depth range of the continental shelf of South Africa, and will open an extensive array of new research possibilities to scientists based at tertiary education facilities and research institutes in South Africa.The next step is to implement priority research projects that, for example, provide data necessary to develop an understanding of the ecology of shallow-and deep-water habitats and their connectivity, and to determine the spatial patterns of abundance and biomass distributions of vulnerable and endemic fish species.This will support the implementation of effective EBFM and form the basis of long-term monitoring programmes that inform adaptive EBFM.
Beyond the scientific value, hours of video footage will be available for educational purposes to raise awareness regarding the vulnerability of reef fishes and the role of no-take MPAs in protecting reef ecosystems.This material can be used to stimulate interest in marine biology amongst the younger generation and inform communities on the importance of MPAs and fisheries regulations.

Table 1:
Fish species sampled by means of controlled angling (CA), fish trap (FT), remotely operated vehicle (ROV), remote underwater video systems (RUVs), baited RUVs (BRUVs) and underwater visual census (UVC) in the Tsitsikamma National Park Marine Protected Area during a simultaneous method comparison. 45Each method was allocated two randomly selected days for data collection within a 24-day sampling period.

Method (number of samples)
Common name (total number of species recorded with all six methods) CA ( 16) FT ( 25

Figure 1 :
Figure 1: (a) Schematic of a baited remote underwater stereo-video system (stereo-BRUVs) showing stainless steel frame (A) with pins to mount additional weights (B) and rigid centre bar (C) that holds the housed digital high-definition cameras (D).Extending perpendicularly from the centre bar is a pole that holds the synchronising diode (E) and the bait container (F).The system is linked to the surface by a buoy and rope system that attaches to the stainless steel frame (G).(b) Stereo-BRUVs deployed at a depth of 20 m on Rheeders Reef off Storms River, Tsitsikamma National Park Marine Protected Area (photo: Steve Benjamin).

Figure 2 :
Figure 2: Number of articles published in peer-reviewed scientific journals (a) from 1952 to June 2013 that included use of video sampling techniques and (b) from 1994 that included use of stereo-video sampling techniques.The bars indicate the number of articles published by July 2013, while the open circle indicates the expected number of published articles by the end of 2013.The articles were collated by means of academic publication search engines (Academic Search Premier, Cambridge Journals, CSIRO, ESA, JSTOR, Google Scholar, Inter-Research, Science Direct, Sci-Verse Scopus, Springer Link, Web of Knowledge) and reference list reviews of key literature.Search terms entered included: BRUVs, RUVs, DOVs, remote video and fish.Studies using remotely operated vehicles and autonomous underwater vehicles and those with data collected beyond the continental shelves were excluded.