The Application of an Artificial Neural Network to Quantify Anthropogenic and Climatic Drivers in Coastal Phytoplankton Shift

Abstract

The overlapping effect of anthropogenic activities and climate change are major drivers for a shift in coastal marine phytoplankton biomass. Linear regression analyses are not sufficient to detect the nonlinear relationship between complex environmental factors and phytoplankton shift. Here, an Artificial Neural Network (ANN) model is applied to quantify the relative contribution of pearl oyster farming, temperature and rainfall on phytoplankton increases in Cygnet Bay, Australia. The result shows that increased oyster farming ranks among the most important factors for phytoplankton increases, with a relative importance of 54% for diatoms and 74% for dinoflagellates; temperature plays a second important role with a positive impact on diatoms (relative importance of 25%) but a negative impact on dinoflagellates (relative importance of 19%); rainfall is the least important which enhances diatom biomass only (relative importance of 21%). Our ANN analysis provides a useful approach for quantifying the complex interrelationships affecting phytoplankton shift.