Description: Metaecosystem theory addresses the link between local (within habitats) and regional (between habitats) dynamics by simultaneously analyzing spatial community ecology and abiotic matter flow. Here we experimentally address how spatial resource gradients and connectivity affect resource use efficiency (RUE) and stoichiometry in marine phytoplankton as well as the community composition at local and regional scales. We created gradostat metaecosystems consisting of five linearly interconnected patches, which were arranged either in countercurrent gradients of nitrogen (N) and phosphorus (P) supply or with a uniform spatial distribution of nutrients and which had either low or high connectivity. Gradient metaecosystems were characterized by higher remaining N and P concentrations (and N∶P ratios) than uniform ones, a difference reduced by higher connectivity. The position of the patch in the gradient strongly constrained elemental stoichiometry, local biovolume production, and RUE. As expected, algal carbon (C)∶N, biovolume, and N-specific RUE decreased toward the N-rich end of the gradient metaecosystem, whereas the opposite was observed for most of the gradient for C∶P, N∶P, and P-specific RUE. However, at highest N∶P supply, unexpectedly low C∶P, N∶P, and P-specific RUE values were found, indicating that the low availability of P inhibited efficient use of N and biovolume production. Consequently, gradient metaecosystems had lower overall biovolume at the regional scale. Whereas treatment effects on local richness were weak, gradients were characterized by higher dissimilarity in species composition. Thus, the stoichiometry of resource supply and spatial connectivity between patches appeared as decisive elements constraining phytoplankton composition and functioning in metaecosystems.

Description: Identifying stable coexistence in empirical systems is notoriously difficult. Here, we show how spatiotemporal structure and complex system dynamics can confound two commonly used stability metrics in empirical contexts: response to perturbation and invasion rate when rare. We use these metrics to characterize stable coexistence across a range of spatial and temporal scales for five simulated models in which the ability of species to coexist in the long term is known a priori and for an empirical old field successional time series. We term the resulting multivariate distribution of metrics a “stability fingerprint.” In accordance with a wide range of classic and recent studies, our results demonstrate that no combination of empirically tractable metrics or measurements is guaranteed to “correctly” characterize coexistence. However, we also find that heuristic information from the stability fingerprint can be used to broadly characterize dynamic behavior and identify circumstances under which particular combinations of species are likely to persist. Moreover, stability fingerprints appear to be particularly well suited for matching potential theoretical models to observed dynamics. These findings suggest that it may be prudent to shift the focus of empirical stability analysis away from quantifying single measures of stability and toward more heuristic, multivariate characterizations of community dynamics.