We call on bioinformaticians, taxonomists and ecologists to collect, store and share new types of data for creating general ecosystem models (GEMs) that will eventually be used for predictive modelling of the biosphere (see G. Mace Nature 503, 191–192; 2013). Funding bodies also need to recognize the importance of this work.

Crucial requirements include: a database of functional traits for different species, which would allow modelling to take advantage of existing information associated with species-based databases; biotic data that indicate how organisms interact through space and time; and census data that quantify the number or weight of organisms in an ecosystem organized by functional traits.

These comprehensive functional data will speed up the development of GEMs by enabling uncertainties to be reduced and predictions to be evaluated (see D. Purves et al. Nature 493, 295–297; 2013).

We therefore appeal to taxonomists, who are usually concerned with morphological traits, to collect information on life histories and behavioural traits, especially with respect to species interactions. Ecologists can also contribute to these enhanced species descriptions by treating individuals as part of assemblages of the same species and as communities of different species, thereby providing valuable collective information.

The technological capacity to store and share trait information is being developed by the Encyclopedia of Life (eol.org/traitbank). This can be scaled up to provide more-comprehensive data, such as those to describe interactions between organisms.