Figure 1.

Path diagram illustrating the principle of multivariate genotype–phenotype mapping. For p loci $x_1,\dots ,x_p$ and q phenotypes $y_1,\dots ,y_q,$ the genotype–phenotype map acts via a genetic latent variable ($\mathrm{L}{\mathrm{V}}_{\mathrm{G}}$)—a joint effect of multiple loci—on a phenotypic latent variable ($\mathrm{L}{\mathrm{V}}_{\mathrm{P}}$), which is a combination of multiple measured phenotypic variables. The effects of the loci on the phenotype are denoted by the coefficients $a_1,a_2,\dots ,a_p,$ and the composition of the phenotypic latent variable is denoted by the coefficients $b_1,b_2,\dots ,b_q.$ These latent variables and their differential effects are properties of the genetic-developmental system of the studied organisms. Multivariate genotype–phenotype mapping seeks latent variables with a maximal genotype–phenotype association in the given sample (thick arrow).