The role of spatial structure in determining the strength of the relationships among climate, landscape and limnological properties
Although the nature of the relationships between organisms and their environments has been shown to be underlain by their respective spatial structure, spatial analyses have rarely been used to understand the large-scale patterns in biogeochemical and physical variables at the base of ecosystem processes. It thus remains challenging to disentangle the effect of environmental pressures occurring at various spatial scales on ecosystem functioning. Here we use a large, multi-themed database spanning 17 US states (> 4,000 lakes and landscape units across a > 1,000,000 km² area) to describe the patterns in spatial structure of climate, atmospheric, terrestrial and aquatic variables using Moran’s I (i.e., how similar are sites within a certain distance in terms of a variable “x”). We quantify the strength of spatial auto-correlation of nearby sites as well as the spatial range, i.e., the distance at which spatial auto-correlation disappears. We then assess the links between the spatial structure and the strength of the relationships between pairs of variables.
Most variables were spatially-autocorrelated, showing a distance-decay pattern in their spatial structure, and the strength of autocorrelation for nearby sites tended to correlate with the spatial range of a given variable. For example, climatic and atmospheric deposition variables were the most strongly structured over space, whereas lake and catchment morphometry showed little to no spatial structure across the study extent. Land cover and land use (e.g., % forest and % agriculture in the catchment) as well as lake water quality had intermediary spatial structure, where sites tended to be autocorrelated within a distance of roughly 500 km across the study extent. We found only weak correlations between variables that were structured on very different spatial scales, but the strength of correlations between variables that were structured on similar scales ranged from weak to strong. Hence, we infer that similar spatial structure may be a necessary condition to find strong correlations between ecological variables. Using a large-scale dataset, we empirically demonstrate the importance of spatial scale in understanding the relationships between a diverse suite of environmental characteristics and discuss the implications for the understanding of the effect of environmental pressures on ecosystem functioning.