Background/Question/Methods Plant ecology has two dominant models of interspecific competion, each taking precedence at a different spatial scale: Local scale, single-patch models with dynamic resources, and spatially heterogeneous regional models that do not explicitly consider resource dynamics. Each framework has strengths and weaknesses. Local, dynamic resource models provide a mechanistic understanding of competition, but ignore broader spatial contexts. Regional models allow growth-rates to be understood in terms of contributions from multiple mechanisms, but often treat competition phenomenologically. Because of these differences, it is often unclear how these models relate to one another.

Results/Conclusions I show that if there is spatial heterogeneity in resource supply rates among patches, simple resource competition models coupled by dispersal can be scaled up to derive approximate analytical relationships for regional species coexistence. These regional scale relationships can be partitioned into the variation-dependent and variation-independent coexistence mechanisms (i.e. the spatial storage effect, non-linear competitive variance, and growth-density covariance). I find that if establishment is resource dependent, the R* of a species in a single patch depends on the amount of inter-patch dispersal it experiences, but the equilibrium density in patches is unaffected by dispersal. A species that disperses a large proportion of its offspring from natal patches will leave more resources behind in relatively rich patches and drive resources lower in relatively poor. This can allow the invasion, and coexistence of a species that is a worse resource competitor, but experiences less inter-patch dispersal. Similar results have been reported for coupled two-patch models of resource competition. Under my derivation of these results, I can attribute coexistence to the contributions from different variation-dependent mechanisms. If local growth rates of species are linear functions of resources, the low-density regional growth-rate of the inferior resource competitor is facilitated by growth-density covariance. Contributions from the spatial storage effect and spatial non-linear competitive variance tend to be small and cancel one another. The relative contributions of these mechanisms are likely to depend on the details of the relationship between local growth-rates and resource concentrations. This work extends theory derived from coupled two-patch resource competition models by considering an arbitrary number of patches, passive dispersal and partitioning the regional scale growth-rate into contributions from different coexistence mechanisms.