Integration of Observations and Models for an Improved Understanding of Marine Ecosystem Dynamics

• Main Author: Kuhn Córdova, Angela M.
• Other Authors: Department of Oceanography, Doctor of Philosophy, Dr. Marion Gehlen, Dr. Christopher Taggart, Dr. Michael Dowd, Dr. Hugh MacIntyre, Dr. Marlon Lewis, Dr. Katja Fennel, Not Applicable, Yes
• Language: English
• Published: 2017
• Subjects: marine ecosystem models; parameter optimization; North Atlantic Region; Phytoplankton; model complexity; Red Sea; Nitrogen > Fixation; North Atlantic
• Online Access: http://hdl.handle.net/10222/73354

While both observational and modelling approaches can improve our understanding of ocean ecology, each type of approach has intrinsic limitations. Direct observations have limited temporal/spatial coverage and many quantities are not easily measured. Models rely on assumptions and parameters that are not always based on direct observation. My thesis research systematically combines observations and models through the use of parameter optimization, and further investigates model behavior with the help of sensitivity analyses and hypothesis-oriented experiments. I apply the optimization formalism in three case studies that revisit paradigms in biological oceanography including drivers of the phytoplankton spring bloom, the importance of trophic interactions in determining rates of primary production, and the biogeochemical role of nitrogen-fixing organisms. The first case study juxtaposes bottom-up and top-down hypotheses to explain the initiation of the phytoplankton spring bloom. Realistic and idealized model simulations reveal that the conceptual bases of both hypotheses are ecological truisms. A spring bloom can develop in the absence of mixed layer fluctuations, and both its magnitude and timing are strongly dependent on nutrient and light availability. Changes in zooplankton grazing modulate phytoplankton biomass but do not produce significant shifts to explain bloom initiation. In the second case study, I compare ecosystem models of different trophic complexity. I found that models of low complexity can accurately respond to bottom-up drivers of phytoplankton phenology; however, aspects like the spring bloom termination, accurate simulation of primary production, and partitioning of nitrogen cycling pathways require a higher degree of complexity that is insufficiently constrained by presently available observations. In the third case study, I demonstrate that the inclusion of specific planktonic traits, such as heterotrophic diazotrophy, is necessary to explain biogeochemical characteristics at certain geographical locations. Despite the regional scope of these study cases, my conclusions provide insights that can be extrapolated to large-scale applications.