Predicting the growth of gilthead sea bream (Sparus aurata L.) farmed in marine cages under real production conditions using temperature- and time-dependent models

The growth of gilthead sea bream ( Sparus aurata ) has been studied considering five multiple exponential regression models using data from 20 lots of gilthead sea bream growing in 20 marine cages from a Mediterranean commercial fish farm. The final weight ( W _f) of fish was predicted in relation to the initial weight ( W _i), time ( n ) and temperature ( T ), or the sum of effective temperatures (∑ T _ef). The estimated weight results from the simulation using the five models have been compared with the real final weight values using the mean of the absolute values of the prediction errors in short and long term (the precision value). All models presented a high determination coefficient, above 96%, and good prediction values in the short term. Regression models were tested using data from six new cages. The best models for predicting the growth of sea bream long term were the ones where final weight is expressed in relation to the initial weight and the sum of effective temperature,     and     obtaining long-term prediction errors 12.9% and 10.7% respectively.