None of the predators of apple snails in their native ranges have been shown to play a significant role in snail population regulation, although snail kites may be important in this regard (R.H. Cowie, personal observations). In South-east Asia, various fish, birds, rats, lizards, frogs, toads, beetles and ants are known to feed on introduced apple snails or their eggs (Halwart, 1994a). Some of these, especially rats, also cause serious damage to rice, and introduction or promotion of others as biocontrol agents may have unknown environmental consequences. Only ducks and fish have attracted any serious consideration as potential control agents.<br>Rice farmers often breed ducks and herd them into rice fields to eat the snails in the period before transplanting (Cowie, 2002;Wada, 2004). A similar approach has been taken for taro in Hawaii (Levin, 2006;Levin et al., 2006). Various duck varieties have been used (Teo, 2001;Levin, 2006;Levin et al., 2006). Two to four ducks per 100 m² were effective in controlling young snails (Vega 1991;Pantua et al., 1992;Rosales and Sagun, 1997;Cagauan, 1999), but some farmers reject this practice because duck faeces contain fluke cercariae that penetrate the skin, which results in itchiness or paddy-field dermatitis (Cagauan and Joshi, 2003). A density of 5-10 ducks per ha in continuous grazing for a period of 1-2 months significantly reduces the pest density from 5 snails per m² to 1 snail per m² (Cagauan, 1999). As ducks graze on and otherwise damage young rice seedlings, it is appropriate to release the ducks when the transplanted seedlings are 4 weeks old. For direct-sown rice, a longer waiting period of 6 weeks is necessary. Using ducks for control may be more effective against P. canaliculata than using chemical molluscicides because the chemicals become ineffective either due to poor drainage in the plots or because snails are still buried in the soil (Cruz and Joshi, 2001).<br>Fish have also been suggested as biological control (Rondon and Sumangil, 1989;Morallo-Rejesus et al., 1990), but few quantified studies have been undertaken (Cagauan and Joshi, 2003). Cyprinus carpio (common carp) and Oreochromis niloticus (Nile tilapia) are popular species for controlling P. canaliculata, with the former more effective than the latter in removing snails (Halwart, 1994b). C. carpio crack the snail's shell, ingest the soft tissue and spit out the broken shell;thus they can feed on snails up to 12 mm high. In contrast, O. niloticus ingests the whole shell, and can therefore only feed on snails smaller than 3 mm. In Japan, black or Chinese carp (Mylopharyagodon piceus) and C. carpio fingerlings have been released to feed on newly hatched snails (Mochida et al., 1991). Models predicting predation rates are provided by Yusa et al. (2001), Ichinose and Tochihara (2001) and Ichinose et al. (2002). One of the problems with using fish is that the water must be kept deep enough for them, which may not be compatible with other methods (Wada, 2004).<br>Little is known of microorganisms associated with ampullariids that might be useful in control, nor of parasitoids that attack either the snails or their eggs. In the Philippines, twelve bacterial isolates were tested, seven of which were effective against P. canaliculata (Cowie, 2002).<br>Halwart (1994a) recommended that specific natural enemies for P. canaliculata, such as the predatory Sciomyzidae, should be sought in its native home in South America.<br>All deliberate introductions of non-indigenous species, including as biological control agents, should be carefully evaluated prior to introduction in terms of both their positive and negative potential impacts, and monitored after introduction.