Morphological Scaling, Behavior, and Evolution: computational and experimental approaches to the genotype-phenotype problem.

Bradley Alicea, Cognitive Science and Quantitative Biology and Modeling, Michigan State University

This research addresses two interrelated domains of investigation with relevance to the genotype-phenotype problem: the interplay between functional anatomy and physiological adaptation, and the role of biological variation in these systems. Both of the research domains presented here can be investigated using computational methods. The former can be probed using virtual environments, while the latter can be probed using various statistical techniques. Recent approaches to the genotype-phenotype problem have involved using fitness landscapes to infer discrete phenotypes such as RNA sequences and metabolic flux analyses to determine muscle adaptation. The work presented here provides both an alternative and a complement to these approaches, as this work focuses more on complex anatomical and behavioral characteristics.

The interplay between functional anatomy and physiological adaptation can be understood using virtual environments. Virtual environments, as introduced in this research, are both a conceptual and technological tool. Conceptually, virtual environments can provide us with a wide range of environmental and physiological manipulations. The degree of isomorphy between these two type of systems, which can be partially manipulated in silico, determine features such as environmental selection and adaptive responses such as changes in gene expression or anatomical compensation. Experimentally, this allows us to create environmental knockdowns, which can be coupled with genetic knockdowns to more fully understand the effects of changes in behavior or environment on organismal diversity and evolution.

The role of standing biological variation in these processes can be explored using statistical tools. In this research, the assumption is made that variation at the molecular level is governed by a series of emergent regularities that are reflected in the scaling relationship between different phenotypic proportions and behavioral- and physiological-related variables called performance scalings or performance allometries. The methods of getting at these phenomena from morphological, behavioral, and physiological observations include nonlinear regression and machine learning methods. The conclusion of this talk will include revisiting experimental methods such as phenotypic switching, and considering a special instance of phenotypic capacitance to explain the results.