Safeguarding genetic diversity is one of the Convention on Biological Diversity's main targets, because genetics underpins ecosystems. General models of whole ecosystems (D. Purves et al. Nature 493, 295–297; 2013) therefore need to incorporate genetic data if they are to represent natural systems and guide conservation policy.

Yet it is important to understand that species diversity and genetic diversity do not always correlate; that the interplay between phylodiversity and functional diversity can be highly complex and regulated by interactions between cornerstone species; and that ecosystem resilience is tied to evolutionary history and genetic diversity. An ecosystem's full genetic potential, as represented by ancient lineages and maximally diverse taxa and key species, must be realized.

Drew Purves et al . reply: It could be useful to incorporate genetics into general ecosystem models (GEMs), along with complexities such as stoichiometry or long-range migrations. Genetically based GEMs could interface with the increasing amounts of genetic data available to capture the effects of individual-scale adaptation, although this can be approximated without explicit genetics (for example, traits mutate in the Madingley model we describe in our Comment). However, such complexities bring greater computational demands and increased model freedom, which might lead to the model producing almost any output unless properly constrained with data.