By Mike Cappo, AIMS.
For decades scientists have directly counted reef fish with SCUBA in depths to 30 metres, but the deeper fauna has been invisible -- and surveys there have been subject to the selectivity inherent in trawl, trap and handline fishing techniques. Meanwhile, the image quality of underwater television and video has increased markedly, and costs and size have decreased to such an extent that underwater video is now an integral tool of marine research.
Researchers at AIMS have adapted cheap Hi8 "HandiCams" in underwater housings made of PVC sewer pipe to launch a fleet of Baited Remote Underwater Videos (BRUVS), to capture the benefits of both underwater visual counts and trapping in surveys of demersal fish. Tested to 100 metres, the BRUVS must be robust and cheap enough to meet the high risk of loss due to snagging, shark attack and strong currents.
Like traps, the BRUVS are baited with 1 kg of minced pilchards in front of the camera, and deployed in a string of six – each with rope for hauling and a string of surface buoys. The cameras record for 90 minutes or 3 hours and are retrieved with a pot hauler. Like direct underwater counts, the immediate habitat type is visible when the tapes are interrogated, and the timing and abundance of fish visits is recorded using the time code stamped on the tape. Replay, pause and zoom functions allow the researchers to closely examine visitors to the BRUVS for identification, and – in case of unknown species – still images can be "grabbed" and emailed around the world to taxonomists. The tapes offer a permanent record that can be shown to the public and fishermen alike for their own interpretation.
Applications to date have focussed on waters 30–90 metres deep, at the bases of reefs, on inter-reef shoals and under the baitfish schools in the famous Cape Bowling green billfish grounds. The most profitable approach has involved habitat mapping and classification by towed video, by Dr Roland Pitcher's team from CSIRO Division of Marine Research, Cleveland Laboratories, followed by deployment of BRUVS in areas with or without specific features, such as sponge and sea whip gardens. This approach has been used to characterise the nature of specific "reds grounds" located for us by Capt. Jim Dalling and other operators.
What's for dinner?
A sea-snake makes a lucky escape when a tiger shark feeds on the bait canister.
General information about sharks
Shark attack fatalities- a few facts
The range of fish, sharks, rays, sea snakes and other animals sighted on tapes has been remarkable – over 300 species to date. From 3 centimetre leatherjackets to 3 metre hammerhead sharks, the videos non-destructively record aspects of the biodiversity present in previously unknown habitats. Most fish are just passing the BRUVS, and are visibly indifferent to the bait, but others vigorously attack the bait canisters and create a berley plume that, in turn, attracts other species. Deep sets to 91 m revealed a charted shoal on the shelf edge off Townsville is, in fact, a huge mound (or "bioherm") of dead Halimeda algae, overlain by a veneer of living plants, and inhabited by rosy jobfish, red emperor and iodine bream. Some of the productive "reds grounds" have proven to be bare gravely sand, pock-marked with large craters of unknown origin – perhaps "wonky holes" in some cases. Sets under the baitfish schools off Cape Bowling Green have revealed high densities of juvenile red emperor associated with sparse patches of sponges and sea whips, although the reasons for the baitfish aggregations are still unknown.
Although approximate measurements can be made using single cameras when fish swim past scale bars, developments are underway to precisely and accurately measure the fish seen anywhere in the field of view, by using the principles of airborne topographic mapping. An informal consortium of engineers and biologists, led by Dr Euan Harvey of the University of Western Australia, has developed software to measure length and other features using paired images from stereo-video cameras. Field trials have shown errors to be about 0.5%, or 5mm on the length of a 1000 mm fish. The next step is full automation of tape analysis, where computer software identifies targets and measures them.
These developments were the subject of an FRDC workshop on "Video in Australian fisheries and mariculture habitats".
For more information, contact
Mike Cappo
