No description available.

Adult fish that belong to the family Poeciliidae cannibalise juveniles, both in the wild and under captive conditions, but this behaviour has only been partly investigated in the Poeciliidae in some of the commercially valuable species. The objective of the research is to develop an understanding of intercohort cannibalism and parturition-associated behaviour in captive-bred swordtail (Xiphophorus helleri), with applications to industry and future research of other poeciliids. Experiments investigating the effect of adult stocking density and sex ratio on the production of juveniles were used to determine if cannibalism occurs under culture conditions. The average rate of intercohort cannibalism ranged from (5.5 to 53.9%), and was positively density dependent and independent of sex ratio, indicating that males and females were probably equally cannibalistic. The highest number of juveniles (1725.7±141.4) produced per tank over 70 days was obtained from two males and eight females. To develop a better understanding of adult and juvenile behaviour during parturition, fish were observed under controlled laboratory conditions using video and behaviours such as attack (burst of speed by an adult in the direction of a juvenile), escape (avoidance of cannibalism after attack) and cannibalism (predation of a live juvenile by an adult), for example, were identified. Under laboratory conditions most young escaped in downward direction after attack (49%) and most utilised the refuge made that was made available. Furthermore, most attacks (62-65%) and cannibalism (57-84%) occurred at the bottom. Since the presence of refuge significantly increased the rate of juvenile survival under culture conditions, it was hypothesised that the rate of cannibalism could be reduced under farming conditions if juveniles were protected when they escaped downwards. This hypothesis was accepted as it was found that refuge at the bottom of the water column or the inclusion of a false-bottom reduced the rate of cannibalism by 49% and 72%, respectively. Similarly, the hypothesise that the rate of cannibalism could be reduced if juveniles where given protection when escaping sideways (32% of juveniles escaped sideways in the laboratory) was also accepted when tested under farm-scale conditions because a false-side reduced the rate of cannibalism by an average of 45%. Since males and females were equally responsible for cannibalising juveniles in the laboratory, it was hypothesised that the rate of cannibalism would decrease proportionately with the removal of males (Le. 20% of the cannibals) from the population; the removal of males under farming conditions resulted in a 19.5% reduction in the rate of cannibalism. Since older juveniles were better able to escape cannibalism than neonates and since adults habituate to stimuli that previously resulted in attack behaviour, it was hypothesised that the rate of cannibalism would remain unaffected by the length of time that juveniles were exposed to adults in the breeding tanks. This hypothesis was also accepted when tested under farm conditions. However, some hypotheses based on laboratory observations were not accepted. For example, a constant low light intensity did not appear to decrease the rate of cannibalism under farm conditions; also, the occurrence of dead and deformed juveniles went unnoticed in the laboratory, and under farm conditions, where adults did not have access to the bottom of the tank, 10% of the harvest consisted of dead and deformed juveniles. It is concluded that technologies, such as bottom-refuge or a false-side, that increase the size of the liveharvest and allow for the removal of potentially less viable offspring are recommended for the commercial production of poeciliids. The overall similarity of X. helleri behaviour between the laboratory experiments and the farm-scale trials suggests that the post-partum behaviour of X. helleri remains consistent under these different conditions; thus, behaviour under one set of conditions may be used to predict behaviour under other conditions. The application and significance of extrapolations to industry and future research of X. helleri and possibly other poeciliids were discussed and the most applicable laboratory observations with the highest extrapolation capacity were proposed. Furthermore, techniques were developed to aid industry and future researchers in making predictions relating to behaviour of X. helleri under different conditions based on laboratory observations. The results were used to develop a model indicating that selection pressures against cannibalism are not likely to exist at the rate of cannibalism observed here since the potential genetic gain through kin survival and inclusive fitness was shown to be greater than any potential genetiC loss experienced by a victim of cannibalism. The model was successfully tested under a range of social conditions. Other possible explanations for cannibalism in poeciliids, such as parental manipulation, nutritional advantages, opportunistic predation and the recovery of energy are discussed. It is suggested that the most likely proximate cause of cannibalism under captive conditions is opportunistic predation. The theory that cannibalism ensures that only viable genes of the victim are expressed, through inclusive fitness, is a possible ultimate cause of cannibalism, which may have been inherited from feral ancestors of captive-bred X. helleri.