Using neural networks to predict surface zooplankton biomass along a 50°N to 50°S transect of the Atlantic

Abstract

Four Atlantic transects between the UK and the Falkland Islands were carried out daring spring and autumn as part of the Atlantic Meridional Transect (AMT) programme. These 50°N to 50°S transects cross several ocean regions. An optical plankton counter (OPC-1 L) sampled continuously along the transects from the ship's uncontaminated seawater supply, giving a surface distribution of zooplankton abundance and size. Measurements of underway fluorescence-derived chlorophyll, sea surface temperature and salinity were also taken from the uncontaminated seawater supply. The relationship between zooplankton biomass and these variables was investigated using multiple linear regression and neural network techniques. In the analysis, loge-transformed biomass was used to reduce the influence of extreme values. Two transects were used to develop the models, and two to test the generalization capabilities of the models. Multiple linear regression could explain up to 55% of the observed variation in the transformed biovolume, and demonstrated the association of hydrographic variables and diel migration within the surface zooplankton community. Neural networks, starting with the same set of variables, were able to explain up to 78% of the variability, showing an increased performance over the multivariate analysis. An optimized model accounted for 77% of the variance in the original data. However, it showed greater generalization capabilities (R2= O.47) when applied to new data sets than either the original neural network model (R2= 0.31) or the multiple linear regression model (R2= 0.34). This study highlights the non-linear nature of the parameters' associations with the zooplankton biomass and their variability between oceanographic regions.