1 2 Integration of Bayesian inference 3 techniques with mathematical modelling 4 5 6

Skeptical views of the scientific value of modelling argue that there is no 13 true model of an ecological system, but rather several adequate descriptions of 14 different conceptual basis and structure. My study addresses this question using a 15 complex ecosystem model, developed to guide the water quality criteria setting 16 process in the Hamilton Harbour (Ontario, Canada), along with a simpler plankton 17 model that considers the interplay among phosphate, detritus, and generic 18 phytoplankton and zooplankton state variables. Predictions from the two models 19 are combined using the respective standard error estimates as weights in a 20 weighted model average. The two eutrophication models are used in conjunction 21 with the SPAtially Referenced Regressions On Watershed attributes (SPARROW) 22 watershed model. The Bayesian nature of my work is used: (i) to alleviate problems 23 of spatiotemporal resolution mismatch between watershed and receiving waterbody 24 models; and (ii) to overcome the conceptual or scale misalignment between 25 processes of interest and supporting information. The lessons learned from this 26 study will contribute towards the development of integrated modelling frameworks. 27 28