Understanding the Genotype-Phenotype Map: Contrasting Mathematical Models

Abstract

In this chapter, we review and compare existing theoretical models of the relationship between genetic and phenotypic variation or genotype-phenotype map (GPM). By doing that, we introduce the reader to concepts and assumptions of evolutionary genetics and contrast them with concepts and models coming from developmental biology, specially its most systems biology side. Although these two approaches can be regarded as complementary to study the same underlying problem, phenotypic variation and evolution, they contradict each other in a number of ways.The simplicity of the pure genetic models has been used to argue (Wagner 2011) that they are the most general. We argue, in contrast, that epigenetic factors are crucial to understand the GPM. We understand epigenetic factors as non-genetic factors that are instrumental in building the phenotype during development (Waddington, 1968).We argue that models including epigenetic factors exhibit features found in real GPMs that are not found in purely genetic models. Since these features are widely found in real GPMs no model can be considered general without those and, then, models including epigenetic factors are more general than purely genetic models, even when the details of the former are specific of certain types of phenotypes.

Beyond genotype‐phenotype maps: Toward a phenotype‐centered perspective on evolution

“…However, without specifying a developmental mechanism one cannot tell if and when particu-lar phenotypes are more or less likely to arise within a population (which cannot be predicted by any previous measurement of the G matrix). While this is acknowledged to limit inference about phenotypic novelty, [10,[20][21][22] it is also relevant to the evolution of quantitative phenotypes since lack of mechanistic representations limits the ability to understand how developmental processes structure the phenotype distribution available to selection, and how those biases themselves evolve. [10,[22][23][24] We suggest that an understanding of the evolutionary causes and consequences of how phenotypes form, persist, and get transmitted will benefit from a generic representation of development that reflects the causal parity of genetic and non-genetic factors in development.…”

“…While this is acknowledged to limit inference about phenotypic novelty, [10,[20][21][22] it is also relevant to the evolution of quantitative phenotypes since lack of mechanistic representations limits the ability to understand how developmental processes structure the phenotype distribution available to selection, and how those biases themselves evolve. [10,[22][23][24] We suggest that an understanding of the evolutionary causes and consequences of how phenotypes form, persist, and get transmitted will benefit from a generic representation of development that reflects the causal parity of genetic and non-genetic factors in development. [25,26] By causal parity, we mean that, although genetic and non-genetic factors play different roles during development, those factors are equally indispensable for phenotypic determination.…”

“…[3,38] Such mechanistically derived GP-maps have certain advantages over statistical descriptions of phenotypic variation (e.g., the heritable co-variation between number and width of segments captured by a G-matrix). [22,23,39] First, the putative GP-map is not limited to the variation realized in nature, but encompasses all phenotypic variation accessible through mutation, including novel traits. [40][41][42] This makes it possible to disentangle the appearance of phenotypic variants (i.e., which morphologies are more likely to arise by genetic mutations) from their persistence (i.e., which morphologies are maintained because of their functionality).…”

“…However, without specifying a developmental mechanism one cannot tell if and when particular phenotypes are more or less likely to arise within a population (which cannot be predicted by any previous measurement of the G matrix). While this is acknowledged to limit inference about phenotypic novelty, [ 10,20–22 ] it is also relevant to the evolution of quantitative phenotypes since lack of mechanistic representations limits the ability to understand how developmental processes structure the phenotype distribution available to selection, and how those biases themselves evolve. [ 10,22–24 ]…”

“…While this is acknowledged to limit inference about phenotypic novelty, [ 10,20–22 ] it is also relevant to the evolution of quantitative phenotypes since lack of mechanistic representations limits the ability to understand how developmental processes structure the phenotype distribution available to selection, and how those biases themselves evolve. [ 10,22–24 ]…”