Towards a verified mechanistic model of plankton population dynamics

Abstract

Plankton are a signicant component of the biogeochemical cycles that impact on the global climate. Plankton ecosystems constitute around 40 % of the annual global primary productivity, and the sinking of plankton to the deep ocean (the so-called biological pump) is the largest permanent loss of carbon from the coupled atmosphere-surface ocean-land system. The biological pump need only increase by 25 % to cancel the anthropogenically-released ux of CO2 into the atmosphere. Mechanistic models of atmosphere-ocean dynamics have proved to have superior predictive capabilities on climate phenomena, such as the El Ni~no, than empirical models. Mechanistic models are based on fundamental laws describing the underlying processes controlling a particular system. Existing plankton population models are primarily empirical, raising doubts to their ability to forecast the behaviour of the plankton system, especially in an altered global climate. This thesis works towards a mechanistic model of plankton population dynamics based primarily on physical laws, and using laboratory-determined parameters. The processes modelled include: diusion and convection to the cell surface, light capture by photosynthetic pigments, sinking and encounter rates of predators and prey. The growth of phytoplankton cells is modelled by analogy to chemical kinetics. The equations describing each process are veried by comparison to existing laboratory experiments. Process-based model verication is proposed as a superior diagnostic tool for model validation than verication based on the changing state of the system over time. To increase our ability to undertake process-based verication, a model of stable isotope fractionation during phytoplankton growth is developed and tested. The developed model has been written to complement other process-based models of biogeochemical cycles. A suite of process-based, biogeochemical models, coupled to an atmosphere-ocean circulation model, will have superior predictive capabilities compared with present global climate models.