CHOICE AND ACCURACY OF BACK - CALCULATING LENGTHS MODEL IN FISHERIES ECOLOGY - DEVELOPMENT OF A COMPUTER-PROGRAM

Models to back-calculate lengths measurements from hard parts are often used without precautions in studies of population dynamics and fisheries ecology. The choice of a model must be justified and its validity has to be evaluated. A computer program is proposed to fit most of the models found in literature (linearizable, polynomial) and to back-calculate the lengths according to different methods. Some criteria to choose the best model and to estimate its reliability are given: scatter plots, statistical tests to check the model's significance and to evaluate the normality and independance properties of the residuals, the confidence limits for back-calculating lengths. The program is first checked on a 239 young migrating sea trout (Salmo trutta L.) trapped in a Normandy river and then on a subset of the total sample.