This will be discussed later. Two species, Mansonia uniformis and Mansonia septempunctata, which breed in association with macrophytes such as water hyacinth Eichhornia crassipes, became less common from stage 1 to 2. The saltmarsh species Aedes vigilax was also collected in reasonable numbers at all localities around the reservoir. This species is known for its wide dispersal powers and was undoubtedly blown in from the extensive intertidal wetlands on the coast. Thus on the basis of abundance, two taxa – Culex annulirostris and Anopheles annulipes s.1. – warranted further consideration. The former species is considered to be the major vector of arboviruses in Australia (Russell 1995), transmitting Ross River, Barmah Forest, Kunjin, Kokobera, Alfuy and Edge Hill viruses and Murray Valley encephalitis, as well as dog heartworm. Of these, Ross River is by far the most common arbovirus in coastal northern Queensland, with morbidity approximating 400 cases per 100,000 population. Thus from first principles, this arbovirus and perhaps Barmah Forest, about which little is known, would constitute the greatest hazard to recreational use. Although Anopheles annulipes has previously been implicated in malaria transmission at Sellheim during the Second World War, this species group has returned isolated positives of Ross River and Barmah Forest viruses and Murray Valley encephalitis from other parts of Australia. However, no transmission studies have been done on the population from the reservoir. Thus on the evidence to date, it could not be regarded as a major concern at the Ross River dam. Both Culex annulirostris and Anopheles annulipes were shown to have seasonal peaks of abundance during the late post-wet season (March to May), with populations building up with the onset of spring (September to October). Spatially, the trapping programme was designed to compare mosquito numbers on the foreshore of the stage 1 lake with two localities expected to be on the margins of the stage 2A lake, with two remote localities (and therefore theoretically unaffected by any water resource project activity) as negative controls. Mosquito numbers (i.e. for those species known to breed at the dam) decreased with distance away from the Ross River dam. Both light trapping and human bait collections carried out twice per month were reasonable indicators of broad seasonal trends in mosquito abundance. However, the statistical analysis indicated that occasionally the light traps could miss short periods of high biting activity (Jones et al. 1991). If greater resolution was required, it was recommended that light traps could be supplemented with animal baited traps, although it is probable that this could be rectified by intensifying the light trapping regimen. Cluster analyses of dam breeding species in both 1984–85 and 1991–93 indicated that light trap catches along the northern (Big Bay, Ti-Tree Bay, Round Island) and western sides (Ross River) gave similar patterns, but the profile towards the east (Antill Creek, Toonpan, Oak Valley) was somewhat different (Barker-Hudson et al. 1993; Hearnden and Kay 1995). On this basis, adult mosquito surveillance would therefore need to be based on two localities at either end of the lake.