Abstract
Selection mapping is a family of methods that use genome-wide genetic data to detect signatures of selection in the genome. The methods used to quantify these signatures range from relatively simple diversity statistics to complicated population genetic models or machine learning classifiers. Selection mapping of domestic and feral animals have identified signatures of selection in regions of the genome known to be associated with simple traits, large-effect alleles for complex traits, but also many regions of unknown significance. Selection mapping can be thought of as part of a broader move towards "natural history of the genome". As large amounts of genomic data are becoming available, there is a space for descriptive accounts of genome-wide data that aid hypothesis-generation and draw attention to striking features of the genome. However, selection mapping gives a particular, biased view of selection, and has methodological issues including dependence on population genetic parameters that are often unknown, and a bewildering array of analysis options. Therefore, we would do well to think hard about how and why we apply it.