Biodiversity in agricultural landscapes can be increased with conversion of some production lands into 'more-natural' – unmanaged or extensively managed – lands. However, it remains unknown to what extent biodiversity can be enhanced by altering landscape pattern without reducing agricultural production. We propose a framework for this problem, considering separately compositional heterogeneity (the number and proportions of different cover types) and configurational heterogeneity (the spatial arrangement of cover types). Cover type classification and mapping is based on species requirements, such as feeding and nesting, resulting in measures of 'functional landscape heterogeneity'. We then identify three important questions: does biodiversity increase with (1) increasing heterogeneity of the more-natural areas, (2) increasing compositional heterogeneity of production cover types and (3) increasing configurational heterogeneity of production cover types? We discuss approaches for addressing these questions. Such studies should have high priority because biodiversity protection globally depends increasingly on maintaining biodiversity in human-dominated landscapes.