In the autumn of 1927, the biologist Francis Sumner spent two months in Florida and Alabama trapping wild oldfield mice for studies of skin pigmentation. With the advice of local farmers, Sumner managed to shoo hundreds of rodents out of their burrows.

Sumner also took the opportunity to document the complexity of the vacated mouseholes, detailed in an article published two years later (F. B.Sumner and J. J.Karol J. Mammal. 10, 213–215; 1929 ). The creatures’ burrows included a long entrance tunnel leading to a nest and, in case of an invading snake, a secondary escape tunnel that didn’t quite reach the surface, a metre or so from the entrance. Other ethologists have since characterized the humble mouse burrow, the structure of which is seen as a model of complex animal behaviour.

Fast-forward almost a century, and a team led by Hopi Hoekstra of Harvard University in Cambridge, Massachusetts, has elegantly unpicked the genetic basis of this behaviour using a cross-breeding design and cutting-edge genotyping methods. Such work, published in this issue (see pages 284 and 402), should appeal to more than just mouse fanciers. Like few papers before, the work shows how long-forgotten field observations, evolutionary theory and molecular genetics can all be brought to bear on a single question.

We have learned much about the physiology of behaviour from model organisms such as laboratory mice — for example, the discovery of genes that determine circadian rhythms, which revealed important mechanisms underlying behaviours such as sleep. But decades of selection for convenient traits such as docility have made laboratory models less than ideal for studying the evolution of complex behaviours. They tell biologists little about the vast behavioural differences that can exist between closely related animals, probably as a result of natural and sexual selection.

Scientists interested in probing the behaviour of wild animals can follow Hoekstra’s lead and pick animals and behaviours with a rich history of observation and striking differences between close relatives. Decades-old observations of ant behaviour, including those by Edward O. Wilson, culminated in the discovery, published online in Nature today, that a social chromosome explains why some red imported fire ant colonies have one queen, whereas others accept multiple queens.

There are, of course, risks to tackling behavioural genomics in wild animals. The ultimate proof of any gene’s role in a specific behaviour involves knocking in or out the gene to remove or endow that behaviour. Such experiments are a challenge even in model organisms, and so far few precedents have been set in non-model species.

Model organisms, imperfect as they are when it comes to studying some behaviours, have focused attention on a handful of organisms. If every interesting animal becomes fair game, there is a risk that behavioural genetics will be fragmented. “If everyone does it in their own species, it will not be a very productive type of enterprise,” says Laurent Keller, a geneticist at the University of Lausanne, Switzerland, who led the ant research. He suggests that scientists converge on a set of wild animals in which to intensely study behaviour. If the latest research is any indication, such animals will be no strangers to historians of biology.