Empirical dynamic model deciphers more information on the nutrient (N)—phytoplankton (P)–zooplankton (Z) dynamics of Hooghly-Matla estuary, Sundarban, India

Abstract

Ecological models are considered as snapshots of ecological processes that are frequently expressed by mathematical equations. Parameters that explain the rate of the processes in the system are included in mathematical equations of the ecological process.The empirical dynamic model (EDM) is an equation-free, non-parametric model that is based on time-series data to recover the dynamic attractor of the system that governs the system dynamics. The model selection is made through the measure of forecast skill and its optimization. The non-linearity of the Sundarban mangrove ecosystem is hard to define by the limited parameters. Thus, the application of EDM in such a super-dynamic system is considered to be a need of current research to understand the complex system dynamics. The cross convergent mapping identifies the causal variables of the system. The EDM showed that the salinity of downstream and nutrient input from mangrove forests determines the nutrient availability; and the water temperature, surface solar irradiance, and salinity are the main drivers behind phytoplankton abundance in the estuary of the Sundarban mangrove ecosystem. The higher values of the forecast skill and lower values of root mean square error (RMSE) ensure better predictability of the non-parametric model over the parametric model.