Replenishment of benthic marine populations typically involves ''settlement'' from pelagic larval to benthic juvenile habitats. Mortality during this transition has been unknown because of the difficulty of measuring propagule supply in open water. For three weeks, we compared the nocturnal passage of presettlement fishes across the barrier reef encircling Moorea Island (French Polynesia) with the abundance of benthic recruits in the back-reef lagoon on the following morning. During this time, Ͼ40,000 presettlement unicornfish, Naso unicornis entered our study area of ϳ1 km 2 with half arriving on just two nights. Using coupled Beverton-Holt functions to describe the decay of each cohort, we were able to predict the daily abundance of recruits and their final age structure from the presettlement inputs. The best model estimated that ϳ61% of the potential settlers were lost between their nocturnal arrival and the following morning, independent of cohort size. Postsettlement mortality was density dependent, varying between 9% and 20% per day. We attribute all mortality to predation and suggest that high risk associated with settlement has shaped colonization strategies. Because fishing targets the survivors of this population bottleneck, aquarium fisheries may be more sustainable when sourced from pelagic juveniles.

Diffcrcnces in Mediterranean fish communities of two rocky coastal arcas, one inside an integral marine reserve and thc second outside the reserve, ncar Banyuls-sur-Mer, France, wcrc compared using underwater visual census aftcr a 12-year interval. In 1980, thc community structure in and outside the marine reserve was studied at two depths (Bell, 1983). In 1992, WC uscd the same methodology to asscss if variations occurred in this fish coinmunity between these two samplings. In both 1980 and 1992, water depth affected the relative abundance of fish communities and the number of species was roughly maintained in both sites. However, there were several other important qualitative and quantitative variations in the fish fauna over this period: (1) The abundance of specics has decreased in the integral rcscrve whereas it has been maintained outsidc the reserve; (2) Nine species arc more abundant in the integral rcscrvc and nine others are more abundant outsidc the reserve. Among the specics vulnerable to fishing, such as Lubrus merulu, Symphodus (inca, Mullus surmuletus, Dip1odu.c. sargus, il. vulgaris, Scorpavnu porrus, Ohlada melanuru, 6 of them are more abundant inside the integral reserve and 4 othcrs arc more abundant outside; (3) Thc demographic structure of vulnerable spccics inside the reserve has changcd: only Ihe proportion of large fish (30 to 40 cm length) compared to medium (15 to 30 cm) and small (0 to 15 cm) fishes was higher in the integral reserve. The prohibition of recreational and professional fisheries and scuba diving has the fish community in the integral marine reserve in 1992 enhanccd less than in 1980. The impoverishment of the fish density in the integral reserve between these two periods is difficult to understand in regard to the stability of the fish density outside the reserve. These results stress the need for a more regular and more extensive survey of the fish assemblage in and around the marine reserve of Cerbère-Banyuls.