Enhanced functionality of the redesigned hybrid evolutionary algorithm HEA demonstrated by predictive modelling of algal growth in the Wivenhoe Reservoir, Queensland (Australia)

This paper presents the functionality of the newly designed hybrid evolutionary algorithm (HEA) applied for synthesizing predictive rules from complex ecological data by providing the options for: (a) modelling single or multiple rules and (b) optimizing model parameters by Hill Climbing (HC) or Differential Evolution (DE). The effectiveness of the improved HEA is tested by predictive modelling of chlorophyll-a and the tropical cyanobacteria Cylindrospermopsis monitored in the Wivenhoe Reservoir in Queensland (Australia) from 1998 to 2009. The paper validates results of the alternative optimization algorithms and model structures, and provides insights into ecological relationships captured by the models by means of sensitivity analyses.