The metaecosystem framework has been proposed to conceptualize the interactive effects of dispersal and resource flows on the structure and functioning of communities in a heterogeneous environment. Here, we model a two-patch metaecosystem where two species with a trade-off in resource requirements compete for two limiting resources—generalizing the so-called gradostat experimental setup. We study the competition outcome in dependence of resource heterogeneity and between-patch diffusion for different combinations of resource supply ratios. Our numerical simulations show that community composition and local and regional diversity are determined by the interplay of resource heterogeneity, resource supply stoichiometry, and diffusion rate. High resource heterogeneity increases regional diversity, with species coexisting due to spatial segregation, whereas low resource heterogeneity favors local diversity, as species coexist mainly by local resource partitioning. Regional diversity averaged across a gradient of resource ratios decreases monotonically with diffusion rate, while local diversity follows a unimodal dependency. However, these dependencies become bimodal for high resource heterogeneity because various bistable states occur at intermediate diffusion rates. We identify three kinds of bistable states with species priority effect: (i) bistability between the dominance of one or the other competitor, (ii) bistability between one species dominance or species coexistence, and (iii) two alternative coexistence regimes differing in species-relative abundances. Most bistable states appear at high resource levels when biomass fluxes strongly interact with resource fluxes. Our analysis provides new insights for the potential effects of metaecosystem dynamics on biodiversity patterns.