The aim of this thesis was to study and quantify the dynamics of bivalve communities and their influence on the pelagic system. To achieve this, an individual-based population model for bivalves (based on the Dynamic Energy Budget theory), was developed and coupled to a hydrodynamic and biogeochemical model (MOHID Water Modelling System). The result is a process oriented modelling tool that integrates physical, biogeochemical, ecological and physiological factors governing bivalve populated marine ecosystems. The originality of this work lies, among others, in the integration of several fields of knowledge to achieve a better understanding of the relative importance of the processes. The integrated modelling tool was successively tested throughout its development and it was implemented in a real ecosystem (Balgzand, Wadden Sea, The Netherlands). The structure of the thesis mirrors the steps towards the final goal of building and implementing the integrating modelling tool.
Chapter 2 deals with detailing feeding processes in bivalves, particularly on mussels. A new mechanistic description of bivalves feeding was developed as an extension of the standard DEB model. Filtration, ingestion and assimilation are assumed as three different steps and pseudofaeces production is computed as the difference between filtered and ingested fluxes. The concept of Synthesizing Units described by the DEB theory was used to develop generic formulations to account for different types of food, with type-specific ingestion and assimilation efficiencies. Necessary parameters were estimated and the model performance was evaluated by comparison with literature data for the blue mussel for a wide range of experimental conditions, with good results.
A new set of DEB parameters for the blue mussel Mytilus edulis is presented in Chapter 3. The new estimation is based on the covariation method that consists on the simultaneous minimization of the weighted sum of squared deviations between data sets and model predictions in one single procedure. Different data sets, obtained from the literature, were used in the estimation procedure and model results using the new and the previous estimations were compared with the observations. For the present model configuration and for the tested datasets, the study concludes that the parameter set obtained by the covariation method leads to a better fit between model and observations, and is therefore potentially more consistent and robust. In Chapter 4, the individual model, including the feeding model extension presented in Chapter 2, and the new parameters set presented in Chapter 3, was tested against field observations. At 4 locations in the North Sea, labelled mussels were kept under natural condition. Shell length and and dry weight was determined for some mussels and environmental properties were measured next to the experiment sites. Results revealed that the individual model was able to reproduce perfectly the pattern and reasonably the average growth of the mussels in the four places. This result implies that the main metabolic processes at the individual level are well described by the model. By performing different model scenarios we conclude that inorganic particles can have an important effect in the individual performances, even more than seasonal changes in food composition.
The upgrading of the generic individual model from Chapter 4 into a population model is de- scribed in Chapter 5. An individual based population model (based on DEB) was built, meaning that the population dynamics is represented by several cohorts' trajectories. Each cohort consists of identical individuals born at the same time and showing identical properties throughout their life and that interact with other cohorts through food competition. Other population processes included are initial egg mortality, background mortality, and predation (including cannibalism). Model properties were studied through the analysis of theoretical scenarios. Next, the model was used to schematically simulate a mussel bed located in the intertidal area of the Balgzand. Major simplifications were made on the loss of larvae by dispersion, the tide effect, and the feedbacks to lower and upper trophic levels. Besides the great amount of available data, important information on predators' diet is still missing, namely on what fraction of the predators diet consists of mussels. As the role of predation is an important question, multiple scenarios were explored, with simple variations of the predation parameters. Criteria were then developed to select well fitting results, narrowing the parameters combinations. The selected modelling scenarios were able to reproduce the timing of some peaks in mussel abundances. They also produced similar size distributions, but the absolute number of individuals was not well predicted. Possible causes were identified and suggestions were made to overcome the discrepancy, including a better description of food availability and larvae/food dispersion potentially given by an ecosystem model. Chapter 6 firstly describes the main features of the model and the concepts used in coupling the individual based population model with MOHID Water Modelling System. This is followed by a fully integrated implementation in the Balgzand are, The Netherlands. The model simulates, in a fine resolution domain, hydrodynamics (currents and water elevations), waves, heat, salt and sediment transport, biogeochemical cycle of nutrients, primary production and bivalve population dynamics. It is initialized and forced by an extensive observations data set during a period of two years (2009/2010). Model results for a reference scenario are in good agreement with observations, and provide a consistent quantitative description of local hydrodynamics and biogeochemical cycles. The mussel spawning season in the Balgzand is long and almost continuous and larvae dispersion is quite important. The study strengthens that there is no single mortality factor responsible for the population dynamics regulation. Early stage mortality (top-down) can control the persistence of new cohorts, in particular cannibalism and shrimp predation, although starvation (bottom-up) is the main process responsible for bivalve loss over the year in terms of biomass. By performing a scenario considering the nonexistence of mussel beds the study shows that in general bivalves' activity intensifies the seasonal patterns of phytoplankton and nutrients in areas close to the mussel beds, but they do not change their overall spatial distribution. The Balgzand acts as a sink of phytoplankton, due to bivalves' filtration. Without bivalves it would export phytoplankton. It also acts as a source of ammonia, exporting about 40% more than the input flux, suggesting high ammonia regeneration. Thus, the study confirms and quantify that bivalves do have the potential to influence ecosystem functioning due to their role in nutrient cycling.
As future work, more scenarios could be performed to provide insight in the relative importance of a process. As an example, it could be interesting to test the influence of oysters and cockles on mussel growth; test the influence of temperature in prey-predators relations between mussels and shrimps; estimate possible locations for artificial mussel beds or study the ecosystem response to different environmental scenarios. As the first integrated modelling study that focus on the mussel' beds in the Balgzand, the main difficulties on model design, setup and results analysis were overcome and it can now be further used, tested and improved. The model is general enough to allow its application to any ecosystem with similar processes and multiple species. The first step was taken, but only more tests, implementations and improvements will give the model, and the scientific community using it, the desired experience to serve as an effective and reliable management tool.