Whether they like it or not, developmental biologists are acutely aware that most genes have more than one function — that is, they are pleiotropic. Population geneticists, on the other hand, have largely ignored this virtually universal feature of genes, principally because of the difficulty of incorporating it into mathematical models: if a gene influences many traits, how does one of its functions evolve without this process disrupting its other roles? By developing a model that quantifies the effect of pleiotropy on selection, Sarah Otto has now come up with some answers.

The new model is designed to address the following question: what is the evolutionary fate of an allele that improves a trait of interest if this allele also has several effects on other traits? As with all models, this one makes several assumptions and simplifications: the crucial assumption is that pleiotropy has an overall negative impact on fitness — that is, the success of an allele that improves the trait of interest is hindered by virtue of that gene having other functions. To simplify the model, Otto then focused on conditions in which the negative effects of pleiotropy are strong relative to the allele's favourable effect on the trait of interest. From this model, some surprisingly general results emerged.

The model predicts that pleiotropy halves the total selection on alleles that spread within a population relative to the case in which evolution proceeds unencumbered by pleiotropy — that is, for every two steps in fitness that evolution takes a population forward when a favourable allele spreads, pleiotropy takes it one step back. And this just describes the effects of pleiotropy on those alleles that do succeed in spreading within the population. A large fraction of alleles that favour a trait of interest do not even spread because their benefit is overwhelmed by deleterious side effects.

The main problem with any model that involves pleiotropy is that no-one knows what pleiotropy is really like. For example, what distribution does the size of pleiotropic effects follow? Despite our ignorance, Otto believes that some inferences about the process of evolution with pleiotropy are possible as, remarkably, certain calculations hold true no matter what shape this distribution takes.

Finding the right conditions to test the model's prediction will not be simple as natural selection is impossible to reproduce. However, previously published artificial selection experiments seem to hint that Otto's numbers are telling the truth.