Empirical studies of fish movement behaviour and their application in spatially explicit models for marine conservation

This thesis investigates the movement behaviour of South African two coastal fish species and evaluates the effectiveness of marine protected areas (MPAs) in their protection and management. Its primary focus is on resolving the movement patterns of roman Chrysoblephus laticeps (Sparidae) in and around the Goukamma and Castle Rock MPAs in the Western Cape province of South Africa. A pilot study of the methodology investigated the movement behaviour of spotted grunter Pomadasys commersonnii (Haemulidae) in the sheltered East Kleinmonde Estuary in the Eastern Cape province. The application of different tagging methods was tested in a controlled tank experiment. Tagged roman were monitored over a 198-day period. Barbed dart, t-bar anchor and Visible Implanted Fluorescent Elastomer (VIFE) tags were compared. Application techniques and underwater visibility of VIFE tags were tested on roman and on fransmadam Boobsoidia inornata in a pilot study. Needles of gauge 25 were found to be optimal for VIFE tag application. Whereas VIFE tagging caused fin rot in fransmadam, it had no negative effect on roman. VIFE tagged fish could be identified by divers from a distance of three metres under ambient light in an observation tank in five metres water depth. There was no significant difference in growth rate between groups of roman with different tags and controls after 198 days. High tag loss rates were experienced for barbed dart and t-bar anchor tags, although barbed dart tags performed better than t-bar anchor tags. Although some of the VIFE marks had deteriorated, all VIFE tagged fish were individually recognised at the end of the study. Conventional dart and VIFE tags are feasible methods to tag roman. However, the high tag loss rate of conventional tags must be taken into account in the design of a tagging study. Previous mark and recapture studies on roman are beset with a number of problems. Poor experimental design and low precision of capture positions resulted in equivocal results of limited value. A tagging experiment was designed to eliminate ambiguity in data interpretation and to produce a dataset that could be used to model roman residency and dispersal. A combination of conventional barbed dart tags and Visible Implanted Fluorescent Elastomer tags was used to tag roman in the Goukamma Marine Protected Area (GMPA) on the temperate South African south coast. Sixty one percent of roman were recaptured within 50 m of the tagging position. A small proportion moved considerable distances of up to four kilometres. The extent of these movements was not dependent on fish size or sex. Data from this experiment and from a previous tagging study in the Tsitsikamma National Park (TNP) were used to model the resident behaviour of roman. The model suggests a probability of 91% (GMPA) and 94% (TNP) of residency within a 10000 m² cell. This result suggests that individual roman will benefit from protection in small MPAs. A different experimental approach was required to investigate the exact home range of this species. Firstly the feasibility of using acoustic telemetry to study the movement of coastal fish in South Africa was investigated. The telemetry equipment comprised two VEMCO V8 transmitters and a VEMCO VR60 receiver linked to a directional hydrophone. A tank experiment was conducted to examine the effects of the transmitter implantation. A tracking experiment was conducted on spotted grunter Pomadasys commersonnii in the East Kleinmonde Estuary. Operated fish recovered quickly and, with respect to swimming behaviour and growth rates, no differences were found between fish with implants and controls. The maximum detection range in the estuary was 400 m. Interference between different transmitter frequencies was negligible. Transmitter location recordings were found to be accurate within five metres. Two fish were tracked over a seven-day period. The fish preferred the lower reaches of the estuary where they made repeated and prolonged use of specific areas. The success of the initial experiments allowed this method to be used to investigate the spatial utilisation and activity patterns of roman Chrysoblephus laticeps. Surgically implanted VEMCO V8, V13 and V16 transmitters were used to track 13 roman inside the Castle Rock MPA in False Bay. Transmitters implanted into C. laticeps in tanks had no apparent effects on growth and physiology. Manual boatand diver-based tracking experiments covered a 17-month period. A VEMCO VRAP radio acoustic positioning system was used over two one-month periods during and after the spawning season of roman. Analysis of data using a 95% fixed kernel algorithm suggests that roman are resident throughout their adult life, occupying home ranges between 1000 and 3000 m². Activity was lower at night. During periods of cold-water upwelling, fish retreated into caves. During the spawning season, females extended their home ranges, possibly to mate with different males. These results confirm that this species is well suited for protection and management with small MPAs. The effect of two MPAs on the South African south coast on the population of C. laticeps was simulated with a spatially explicit individual based model (IBM). Life history parameters determined in recent studies and the effect of fishing on the size of sex change was taken into account. Fish densities and size frequencies were based on recent underwater visual census. The distribution of suitable habitat in the study areas was also incorporated. The results show a rapid recovery of the fish size frequency spectrum and sex ratio to pre-exploitation levels inside both MPAs. Little 'spillover' of fish into the fished areas occurred resulting in negligible improvement of catches. The results suggest that for resident species like roman, even small MPAs offer sufficient protection