Nonlinear relationships between phytoplankton nutrient utilization traits and environmental factors

Abstract

Relationships between the functional traits of phytoplankton and environmental factors remain insufficiently explored. However, these dependencies could be very useful for trait-based ecological modelling of phytoplankton communities. The purpose of this study is to estimate possible connections between nutrient utilization traits of marine phytoplankton, cell volume, temperature of the experiment, irradiance and taxonomic category of the species collected in Ecological Archives E096–202 (Edwards et al., 2015).The search for patterns was made using the author's technique of polynomial multilayered neural network (NN) building. The optimal set of neurons' connections with the preceding layer of the network was obtained by means of a genetic algorithm run for each neuron. This approach gave good results given the sparse data sets and small samples. Categorical predictor “taxonomic group” was parameterized by an integer.We found approximations for maximal uptake and growth rates, half-saturation constants, maximum and minimum nutrients quotas as nonlinear functions of cell volume, taxon number, temperature and irradiance. Mean-square errors were lower than the errors in respective relationships to the cell volume only. These functions were used to fill in the gaps in the ecological archive. Verification of the obtained relationships was performed using an energy balance model and new data on nutrient utilization traits not used for their estimation.