The sand substrate used in the tank systems acted as an internal bio-filter (Coman et al., 2005) producing positive effects on water quality by reducing the ammonia concentrations in the low water exchange recirculation systems. As a result lower ammonia concentrations but higher nitrite concentrations were recorded under the low water exchange recirculation system (system R) as compared to the flow-through water system (system F). Consequently, bacteria converted most ammonia in the system R into nitrite. The process to convert nitrite into nitrate was however only speeded up when increasing the water recirculation rate from 280% a day in experiment 1 to 400% a day in experiment 2 and therefore most of the nitrite was converted to nitrate in experiment 2. However, metabolizing bacteria in the bio-filter produce carbon dioxide (CO2) which gradually acidified the rearing water over time; as a result, the alkalinity levels dropped sooner (after 4 weeks) in system R while it was only going down after two months in system F. The reason for this was the high flow-through rate at 200% a day, which did not allow nitrifying bacteria to develop effectively on the sand substrate initially. Possibly the sand substrate in the flow-through water system started to function as an internal bio-filter after the first two months. In this regard Primavera (1983) and Coman et al. (2005) suggested to periodically add sodium carbonate to the recirculation tank, therefore it would be interesting to test a recirculation system at lower flow-through rates for future research work instead of the high rates (200% a day) used in the present study.