Spatial variation, synchrony, and loss of stability in ecosystem variables

Background/Question/Methods

Loss of temporal stability in an ecosystem disrupts the constancy of services and taxes its components (e.g., populations).  Causes of such variability are numerous and include isolation due to habitat fragmentation, synchronized fluctuations among connected systems, and variation of weather or overexploitation.  Specific stability outcomes arise from local causes within ecosystem patches (e.g., density dependent growth) as well as regional processes that operate across the landscape (e.g., dispersal).  Regardless of causal mechanism, however, these processes alter spatial patterns in the landscape.  We hypothesized that spatial pattern - an index of how local dynamics are distributed across space - would be a predictor of stability at the regional, landscape scale.  We investigated contributions from local and regional sources of stability in a wide range of ecosystem variables from (i) aquatic microcosms (composed of invertebrates, phytoplankton and microbes) in which spatial connectivity and exchange of materials were varied, and (ii) 14 years of data from natural Jamaican rock pool ecosystems. 

Results/Conclusions

General Linear Model results strongly indicated that most observed patterns of stability (measured as mean/standard deviation), whether in different variables or replicate systems, could be accommodated by just two dimensions:  the magnitude of differences (gradients) among local patches and the synchrony or coherence among patches.  Increasing disparities among patches (e.g., differences in population densities, nutrient concentrations) were detectable as high spatial variance, and generally accompanied instability of a system or variable.  Meanwhile, increasing regional synchrony tended to compound the negative effect on stability.  These patterns were surprisingly consistent in the two data sets for abiotic factors, biotic aggregate metrics (e.g., primary production) and individual species, implying generality of the model. These results, along with analytical findings, clarify the link between spatial and temporal variability and may yield new opportunities for understanding how landscape processes affect the stability of ecosystems.