Multi-objective optimization for ecological model assessment and theory development
The motivation for building ecological process models is to synthesize observations and explore hypotheses for system functions. In the course of ecological research it is common to choose a trajectory for a research program in order to explain an observed phenomenon, often at the exclusion of other possible explanations. In particular, in optimality studies a single function is often the focus of the optimization. Organisms are rarely optimal for a single measure, and more often represent tradeoffs among competing requirements for growth and survival. We require multiple objectives to be optimized simultaneously in order to make substantial progress in the process of ecological model assessment, and in the use of ecological models in optimality theory development. The multi-objective optimization utilizes Pareto optimality, wherein the non-dominated set of Pareto optimal solutions is generated.A dynamic simulation model is developed to examine the process of branch morphogenesis in P. menziesii, and multi-objective optimization conducted to assess the model structure and evaluate how the morphology compensates for size constraints observed in old-growth systems. The parameters and independent variables in the model structure represent postulates for the process of morphogenesis in P. menziesii. The theoretical optimality objectives show the tradeoffs among major branch growth constraints and functions; the observed morphology emerges in the optimization results as a compromise between hydraulic constraints and foliage display.The model of plant form and function presented demonstrates how, through the synthesis of multiple phenomena, multi-objective optimization for process models can be a key tool in theory development for ecological systems. The conclusions of the process model are bounded by the context of the model structure, parameters and objectives. The model shows that the old-growth branch morphology is not optimal for any single constraint; the more likely explanation is that it compensates for multiple constraints acting simultaneously in the system. These results are tempered by the bounds of the model structure and they are presented in that context. The modeling results should be integrated with empirical investigation of the physiological processes relevant to branch growth in this and related morphologies.