Adaptation is and has always been a slippery concept. Even among contemporary evolutionary biologists there is no consensus about how to identify adaptive traits. Most definitions of adaptation incorporate elements of history. Among these, some require that adaptive traits be built by natural selection for their current roles. Others take a phylogenetic perspective and require that adaptations be derived relative to traits Of antecedent taxa. Still other definitions require evidence of functional design in the biological machinery underlying the trait, implying a history of selective fine-tuning. The historical definitions are most useful for addressing questions about evolutionary history like, ''What was the phylogenetic trajectory of a phenotype with current utility?,'' or ''How does a trait's current use relate to its original function?'' Historical definitions are less useful for addressing questions about phenotype existence like, ''Why do certain phenotypes predominate over others in nature?'' This is because history-laden definitions often fail to classify as adaptations traits that are maintained by natural selection. Obviously it is important to employ a definition that is appropriate for the research question being pursued. Indeed, the (mis)application of history-laden definitions to questions of phenotype existence has created the illusion that nonadaptive traits abound in nature and caused widespread doubt about the importance of natural selection in molding phenotypes. To circumvent these problems we propose a nonhistorical definition of adaptation. In our view, an adaptation is a phenotypic variant that results in the highest fitness among a specified set of variants in a given environment. This definition treats adaptation as a relative concept, sets forth operational criteria for identifying the phenotypic traits to which the concept applies, and decouples adaptations from the evolutionary mechanisms that generate them. Natural selection theory predicts that among a specific set of alternatives the most adapted phenotype will be the one that predominates in a given environment. This is a testable proposition. When it is false the frequency of the trait must be due to nonselective processes such as drift, migration, selection acting on a correlated trait, or recent changes in the environment. Teleonomic demonstrations of adaptation are a special case of our suggested analysis because inferring a trait's ''function'' implies the application of a specific fitness criterion (e.g., optimal design) to rank alternative phenotypes. The logical structure of our definition allows us to confront five recent challenges to adaptationism, namely that: (1) it is impossible to identify adaptive traits without a knowledge of phylogeny, which can reveal constraints on adaptation, (2) genetic correlations among different traits usually prevent or retard ascent toward adaptive peaks, (3) ''developmental constraints'' limit the operation of natural selection, (4) the complex and highly ordered phenotypes studied by adaptationists can be explained most parsimoniously as manifestations of simple mechanisms, and without reference to natural selection, and (5) it is impossible to study the adaptive significance of human behavior because natural selection has not had time to operate on humans in our ''changed'' environments. We show that these challenges rest on inadequate specification of the components of adaptation.
