Using truss networks to estimate the biomass of Oreochromis niloticus, andto investigate shape characteristics

Truss and length measurements were made on 463 tilapia Oreochromis niloticu s of known mass. These measurements were then used to build mathematical mo dels to estimate tilapia biomass. Two different approaches were used to bui ld the models. The first was to take some simple mathematical models, and t o optimize them using least squares regression. The second approach was to build models using artificial neural network software. Models were built fo r two different but overlapping size ranges and both approaches were succes sful in creating models that estimated the total biomass to within +/- 1%. The truss measurements were also used to assess the principal shape compone nts of different sized tilapia. (C) 2000 The Fisheries Society of the Briti sh Isles.