Data from: Higher-order interactions capture unexplained complexity in diverse communities

Natural communities are well known to be maintained by many complex processes. Despite this, the practical aspects of studying them often require some simplification, such as the widespread assumption that direct, additive competition captures the important details about how interactions between species impact community diversity. On the other hand, more complex non-additive 'higher-order' interactions, are assumed to be negligible or absent. Notably, these assumptions are poorly supported and have major consequences for the accuracy with which patterns of natural diversity are modelled and explained. We present a mathematically simple framework for incorporating biologically meaningful complexity into models of diversity by including non-additive higher-order interactions. We further provide empirical evidence that such higher-order interactions strongly influence species' performance in natural plant communities, with variation in seed production (as a proxy for per capita fitness) explained dramatically better when at least some higher-order interactions are considered. Our study lays the groundwork for a long-overdue shift in how species interactions are used to study the diversity of natural communities.York Gum woodland focal plant fecundity and neighbourhood dataThis file contains estimated seed set (fecundity) data on 773 individuals of six focal annual plant species common to the York Gum woodlands in SW Western Australia. Seed set per focal plant was estimated for each plant by counting up the total number of flowers (or inflorescences) on each plant and multiplying that number by the average number of seeds produced per flower(infor.) using 1-3 flowers(inflor.) per plant. In addition to seed production values, this dataset also includes the identity and abundance of all plants within a 7.5 cm radius around each focal plant. More details are available in the readme file and in the methods of the paper associated with this dataset.complete.Rdat