Web-based Decision Support System for broodstock management of siganus guttatus (bloch, 1787) in open fish cage

Abstract

This paper presents a Decision Support System (DSS) for broodstock management of Siganus guttatus - a high valued herbivorous fish species cultured in the Philippines which has a promising commercial potential. The DSS helps aquaculture experts and farmers in monitoring water quality of the fish cages of the breeders known as broodstock. The system predicts future water quality values based on the past and current values; models present and future water quality parameters through graphs; recommends tasks on broodstock management based on the current water quality and provides an early warning for possible fish kill occurrence based on predicted water quality. The algorithm used for the forecasting module of the DSS is Artificial Neural Network (ANN); forecast error was computed by comparing actual and predicted values, to measure the forecast accuracy; and Test-Retest method was used to assess the reliability of the system. The accuracy rate of the system in predicting future water temperature, salinity, and dissolved oxygen are 91.05%, 92.67% and 72.58% respectively. The forecast accuracy for dissolved oxygen is significantly lower than the forecast accuracy for temperature and salinity because of insufficient training data for dissolved oxygen. The overall accuracy of the system in prediction is 85.44%. The test-retest reliability of the water quality shows consistency between values for each water parameter, hence the system prediction is considered reliable.