A model of channel catfish growth in the partitioned aquaculture system

The Clemson Partitioned Aquaculture System (PAS) is a green water system, in which paddle wheel driven, high rate, algal basins are used to process fish metabolic waste into phytoplankton biomass. The objectives of this study were: (1) To create an equation able to predict individual fish weight and population biogain vs. time under varying water quality and system operational parameters. (2) To allow for optimization of feed application and stocking/harvesting strategies. (3) To provide a basis for the future development of an overall PAS bio-economic model. To meet these objectives, six years of PAS catfish production data from six 1/3 acre units was used to develop a detailed mathematical model describing channel catfish (Ictalurus punctatus) growth under water quality and raceway culture conditions experienced in the (PAS). Catfish stocking densities ranged from 60 to 120 kg/l. Net catfish production ranged from 10,000 to 22,000 kg catfish/ha. System variables impacting the modeled fish growth included: feed rate (90--315 kg/ha d), temperature (27--32°C), morning dissolved oxygen (3--10.3 mg/l), un-ionized ammonia (0--2.96 mg/l), pH (7.0--9.0), fingerling stocking weights (16--80 g fingerlings) and numbers, and harvest weights (600--800 g) and numbers. Two approaches were used to model the relationship between the statistically significant variables, water temperature, morning dissolved oxygen, un-ionized ammonia, and fish average weights, impacting the model's dependent variable feed intake represented as the percentage body weight (PBW) fed to the fish. In a statistical model these variables were correlated to fish growth in a regression model yielding typical R2 values ranging between 0.5--0.9 using six years of pooled growth data. A theoretical model was developed around four sub-models explaining the relationship between PBW and literature based feed intake models, water temperature, dissolved oxygen, and NH3 inhibition. The theoretical model was able to predict feed intake more accurately than the empirical model.