Much of the early history of evolutionary genetics was focused on understanding the relative contribution of genes and the environment to observed levels of phenotypic variation. Chief in this pursuit was Ronald A. Fisher who, amongst his many achievements, developed a statistical framework for partitioning these sources of phenotypic variance in a population. Underlying this framework was the idea that genetic and environmental sources of phenotypic variance in a population could be summed as long as they act independently, providing a simple method to statistically partition the relative effects of these sources of variation in phenotype. This logic is easy to follow if (as Fisher believed) the environment has negligible effects on phenotype and is distributed at random across individuals (and genotypes) in the population. Other researchers at the time (led most notably by Lancelot T. Hogben), however, argued that this framework under-estimated the importance of the environment and also missed a third and important source of phenotypic variation: that which arises from the combination of a particular genetic constitution with a particular kind of environment. Nowadays, we refer to this differential response of genotypes to environmental variation as genotype-by-environment interactions (GEIs) and know that this source of phenotypic variance is almost ubiquitous in most animal and plant populations. Unfortunately, most researchers in the early part of the twentieth century viewed GEIs as an annoying departure from Fisher's additive framework. This view was particularly evident in agricultural genetics where the presence of GEIs often meant that a good genotype (or crop variety) in one environment may perform poorly in another environment. In such instances, the predictive power of genotypes across environments is greatly reduced, which has obvious consequences for the efficiency of selective breeding programs. It was not until the mid-1980s, however, that the explicit role of GEIs in the evolutionary process ...