The aim of the study was to evaluate the performance of Litopenaeus vannamei in an intensive photo-heterotrophic hypersaline system with minimal seawater replacement, and establish relationships between parameters of a stochastic production model and relevant water quality variables. Six experimental 1000 m 2 lined ponds were stocked at a density of 120 shrimp m À2 for a 105-day trial. Salinity increased from 37 to 45 AE 2 g/L, and the water level was maintained with the weekly addition of filtered seawater, equivalent to 1.6% per day. The stochastic model predicted that, at harvest, there is 95% confidence that the system produces between 12.1 and 14.7 t/ha with a mean final individual weight of 13.1 g and a mean survival of 84.2%. Sensitivity analyses showed that dissolved oxygen and individual final weight of shrimp were the main variables influencing yield variance.Nitrogenous compounds were maintained between optimal cultivation levels (NH 3 -NH 4 + = 0.73 AE 0.43 mg/L, N-NO 2 À = 0.09 AE 0.05 mg/L, N-NO 3 À = 3.22 AE 0.11 mg/L). Heterotrophic bacteria (6.6 AE 3.4 9 10 5 CFU/ml) and chlorophyll-a concentration (108.5 AE 80.2 lg/L) showed a similar development pattern, indicating a strong relationship between bacteria and microalgae during cultivation. Vibrio spp. concentrations were low (1.24 AE 1.42 9 10 3 CFU/ml). It was shown that the photo-heterotrophic system could be used in hypersaline conditions, typical of semi-arid regions, to consistently produce between 12.1 and 14.7 t/ha in 15 weeks.
K E Y W O R D Shypersaline, intensive cultivation, Litopenaeus vannamei, minimal seawater replacement, photo-heterotrophic, stochastic modelling