Catch structure and dynamics of Antillean "Z" traps on a coral reef: interaction between trap characteristics and fish behaviour

This study experimentally evaluated a number of factors affecting the structure, dynamics and variability in catches from Antillean "Z" traps deployed on fringing coral reefs at Orpheus Island (central Queensland coast). The aim was to develop methodologies to improve the use of traps as a fishing device and a sampling tool for different target species. The effects of trap characteristics (mesh size), presence of bait, bait type and frequency, soak time and the prior residence of individuals in traps were assessed, in terms of total catches and species composition. Using direct diver observations and video technology, the effects of these manipulations on the attractiveness of traps to different species, the number of fish entering the traps and escapement were distinguished. The generality of the patterns observed was assessed repeating experiments at two locations. Cattle Bay (Trial 1) and Pioneer Bay (Trial 2). Catch rates and final catches were consistently higher in large-meshed traps (42 mm hexagonal mesh) compared with small-meshed traps (12.5 mm square mesh), and for traps baited with pilchards (Sardinops neopilchardus) compared with the use of fish oil. The same pattern was observed during both trials, although overall catches were twice as high at Cattle Bay, compared with Pioneer Bay. During the second trial, baited traps did not perform better than traps with no bait. Also, similar total catches were registered in rebaited traps and traps baited only once, suggesting a minor role of bait on catches at this location. The magnitude of the effects of these different trapping procedures on catch rates and final catches was species-specific. Catches were dominated by four species: Siganus doliatus (Siganidae), Lutjanus carponotatus (Lutjanidae), Plectropomus maculatus and P.leopardus (Serranidae). These were affected differently by the mesh size, with S.doliatus caught almost entirely in large-meshed traps, L.carponotatus caught mainly using pilchards for bait, and the coral trout species not significantly affected by any of the treatments. There appeared to be a trend in the composition of fish families trapped. Herbivorous fish appeared at the beginning of the soak period when large predatory species were absent. More predatory species could have progressively entered traps to prey on the smaller fish species that are being accumulated over the soak period. Differences in catches were the result of different behavioural responses of fishes to traps at different stages in the trapping process. Diver-observations and video analysis indicated that different catch rates in different treatments were not closely related to the number of fish attracted to the traps, but reflected species-specific differences in the likelihood of fish entering or escaping from traps. An experiment in which the presence of S.doliatus in traps was manipulated, indicated that catches were affected by visual communication between fish inside and outside large-meshed traps. Greater numbers of S.doliatus entered large-meshed traps in which S.doliatus had been placed experimentally. On the other hand, L.carponotatus was more likely to enter traps baited with pilchards, and the presence of small fish in traps at the beginning of the soak period resulted in a greater catchability of coral trout. Knowledge of these species-specific fish-fish and fish-trap behavioural responses provides considerable scope for designing trapping protocols for targeting particular species. High variability in catches may be an unavoidable consequence of species specific behaviour patterns and may limit their utility as a sampling tool for measuring changes in relative populations densities