While the use of video technology has been widespread due to its relatively low cost, other technologies which complement or often outperform video technology are used in aquaculture. This section will review some of the aquaculture research carried out using these technologies outlining their advantages but also discussing their limitations. The studies in this section will be outlined individually, in turn, to highlight applications.
2.5.1
Acoustics and Acoustic Telemetry
The effective performance of acoustics (sound) in the water makes the technol- ogy a natural choice for use in aquaculture research. This is especially important for salt water, where the effectiveness of radio telemetry is severely degraded (Section 2.5.2). In tank based research there are two main types of such work: tracking of individuals using acoustic tags, and detecting movement (or noise) of the whole tank population. B´egout Anras and Lagard`ere (2004) used small
ultrasonic tags on rainbow trout to monitor fish swimming behaviour at differ- ent densities. They tracked nine fish in three tanks of 116, 333 and 583 fish, three tagged fish per tank. The tracking system sampled every 5 seconds with an accuracy to < 10cm producing data on distance travelled, turning angles, swimming trajectories, space utilisation and activity patterns. In other work La- gard`ere et al. (2004) recorded the feeding sounds of brown trout, rainbow trout and turbot using hydrophones. The aim of the experiment was to identify sounds characteristic of feeding, specifically suction feeding, which is commonly associ- ated with teleost fish. While the research identified some unique sounds related to feeding, the authors observed noise generated by aquaculture systems (water flows, pumps, etc.) and meteorological noise.
Fitting between the above mentioned areas of studies was a non-invasive way of tracking fish in tanks using acoustics (Conti et al., 2006). The problem however was scattering of acoustic signals within a tank. Conti et al. (2006) developed a set of equations that would allow a system to differentiate between a true signal and its echo. They suggested this technique may be suitable to track the activity of fish in tanks and fish growth, and be used as an alarm if unusual events are detected.
Sea cage aquaculture usually involves larger fish to which acoustic tags could be attached. Cubitt et al. (2003) used acoustic tags to track the movement of 39 chinook salmon within 3 cages (with an average of 6000 fish in each) - 13 fish per cage. Using 8 hydrophones in total, they were able to provide three-dimensional data of fish movement. Their results showed differences in fish movement between day and night, and also change in activity during feeding. The influences of var- ious sources of noise (eg. boats, feeders, generators and natural noise within the environment) were identified, as noise affected the possible frequency of sampling as well as the accuracy. In addition the positioning of hydrophones, which had an effect on the triangulation, influenced the signal, with a recommendation that arranging hydrophones in a cube gives better accuracy than a rectangular box arrangement.
More recent research on tracking fish in open sea cages was conducted as part of the Open Ocean Aquaculture program at the University of New Hampshire (Rillahan et al., 2009). The project investigated the use of acoustic technology to monitor fish behaviours in sea cages. A four hydrophone system capable of tracking hundreds of tagged fish was deployed to allow Atlantic cod to be tracked in a submerged cage. It was observed that the environmental noise affected the sampling rate. A daily activity pattern, similar to the one reported by Cubitt et al. (2003), was observed - higher activity during the daytime than during night time. One important feature of this research project was the very small size of the
tag used, minimising the effect on physical attributes of fish. The tag operated at a different frequency to that which is traditionally used, which resulted in a smaller detection range (not a problem due to the size of the cage) but the accuracy of the system allowed fine movements of tracked fish, like turns and accelerations to be recorded - something not achievable with traditional systems where tags are larger, and also made the system applicable to juvenile fish.
While tagging systems can provide accurate three dimensional data, there are draw-backs especially when considering widespread commercial use. These systems are expensive and they are difficult to set-up (especially compared with video systems) because they require careful positioning in three dimensions and calibration. The implantation of tags is an invasive procedure and while the above mentioned research has not found evidence that the procedure was detrimental to the fish, it is possible that in larger scale commercial settings stress levels could increase. This would mean that tagged fish would behave differently from non- tagged fish, e.g. tagged fish which are stressed may not feed as much or occupy the same area of the cage as non-tagged fish. In a commercial system this has obvious problems as a management tool because the fish being used as a representative sample of the population, are not actually representing the behaviour of the majority of the fish in the whole system. Also tags have a limited life span and would require continuous maintenance and costly replacement. While the cost of the technology will decrease over time, currently such systems are restricted to research-oriented deployments.
Other sonar technologies provide an alternative to tagging. An earlier study as part of the Open Ocean Aquaculture program, undertaken by Michel et al. (2002), examined the feasibility of a multi-frequency/multi-beam sonar capable of detecting objects at up to 50 metres. Several problems were encountered with the system used. Tests of different frequencies proved inconclusive and problems were encountered in detecting fish, especially moving fish as the scan period was too long. While this research looked at cost effective methods of detecting fish in sea cages, another related technology exists which while costly can provide very good results in detecting and tracking of fish. The device, called Dual-frequency IDentification SONar (DIDSON) from Sound Metrics Corporation, is a multi- beam sonar (96 beams) capable of generating video-like sequences of acoustic data. So far its use in aquaculture research has been limited. McKinnon et al. (2008) used DIDSON in Northern Australia to examine the impacts of sea cage farming on wild fish surrounding the cages. Work in Australia, conducted by the New South Wales Department of Primary Industries, on the migration of fish in Australian freshwaters (Baumgartner et al., 2006), produced video sequences of acoustic data generated by DIDSON. The fact that DIDSON generates video means that research undertaken in visual systems may be applicable to the anal-
ysis of acoustic video sequences with only minor modifications. DIDSON could provide a powerful all weather monitoring system capable of tracking fish move- ments in sea cages. While the effectiveness of DIDSON has been praised in the reports, personal communications with researchers revealed that DIDSON is not capable of penetrating through the school of fish and so it may be of limited use. Additionally the cost of the device (which was AU$120,000 in March 2010) means that substantial funding would be required to purchase a unit for research purposes and it is likely not to be feasible for commercial aquaculture use on a per cage basis.
2.5.2
Radio Telemetry
The use of radio and acoustic telemetry has been reviewed by Baras and La- gard`ere (1995). The review covered active tags which use batteries to power circuitry sending out periodic signals. Tagging involves handling fish (Baras and Lagard`ere, 1995) and can have an effect on fish physiology and behaviour of fish (Bridger and Booth, 2003). While radio tags can perform reasonably well in freshwater, salt water operation is limited due to the conductivity of salt water (Koehn, 1999), hence acoustic tags are preferred in salt water (Section 2.5.1). Also the cost of equipment (tags and readers) is somewhat more expensive than video because the price of radio tagging systems is dependent on demand. Higher volumes of production usually mean lower costs. While this is achievable in video technology where consumer demand is high, radio tagging systems are not di- rectly sought by the general public. However the advent of Radio Frequency Identification (RFID) technology is bringing radio telemetry to the mainstream population through supermarkets and other sectors wanting to tag their products for the purpose of identifying and streamlining supply systems. This will increase the demand for radio telemetry and drive the cost down, potentially making it more attractive to researchers and farmers.
Advances in RFID development have produced tags which have no battery of their own but are activated by an energy signal sent from the scanner. This technology is called Passive Integrated Transponders (PIT). Because there is no need for a battery, tags are very small and can be easily inserted into the fish. Also, the effects on fish physiology and behaviour are negligible (Prentice et al., 1990). However the main downside of the tags is the limited detection range (< 30cm), requiring experiments to have strategic points such as near demand feeders (Cov`es et al., 2006) and water inlets, where antennae will be located such that the system will determine the presence of fish close to the point of interest. Fish can also be channelled into narrow passageways for the system to detect
their presence. A narrow channel ensures fish passing through will be registered by the readers. Using this setup, Alan¨ar¨a et al. (2001) examined the passage of Brown Trout between foraging and refuge areas and determined the influence of social rank and temperature on foraging fish behaviours. They noted patterns in foraging behaviours with dominant fish feeding mainly during the night, while subordinate fish either fed in the later part of the night at lower temperatures or during the day at higher temperature, thereby compromising between the risk of predators and the need to feed. A PIT system was used to observe habitat selec- tion by juvenile burbot between four different quadrants with different habitats (Fischer et al., 2001). The use of passageways between habitats allowed fish to swim to a different habitat through an electronic gate, which was used to decide if the transition actually occurred or if fish withdrew to a previous habitat. A study of bottom dwelling Atlantic halibut used PIT tags to examine the surface activity (Kristiansen et al., 2004). The PIT antenna was placed just under the surface of the water and it was used to detect fish swimming on the surface. They found a negative correlation between surface activity and growth rate and suggested that surface activity might be an indication of poor welfare.
PIT systems have also been used to tag a subset of fish in a larger experimental system. Dempster et al. (2008) tagged 400 fish out of a population of 2000 to ensure that these 400 fish could be identified during later stages of the experiment. During transfers, tagged fish were identified and dealt with according to the experiment design. There was no attempt to track or detect fish in water, tags were simply used to identify tagged fish from non-tagged fish. This is the most common application of PIT systems in aquaculture research (Dempster et al., 2008).