The relationship between diversity and functioning (DF) has intensively been studied at the plot-scale using artificially established plant communities. A general finding was that more diverse communities are more productive and productivity temporally more stable. However, whether similar DF relationships apply in natural and human-dominated real-world landscapes remains to be tested. Real-world landscapes differ from experiments in important respects including additional scales of space and ecological organization that may alter response patterns through emergent mechanisms. A key methodological challenge in real-world studies also is that diversity often correlates with other factors that may also affect ecosystem functioning. Observational studies therefore often don’t lack data but its structure complicates statistical isolation of relationships of interest.
Here, we studied diversity-functioning relationships in two large-scale test areas in Europe and North America. Our first objective was to evaluate how design principles from experimental research could help mitigating typical issues with the analysis of observational data. The second objective was to test for diversity-productivity relationships at large scales of space and at organizational levels above species and communities.
We constructed landscape-richness gradients by systematically selecting landscape plots with different land-cover composition. Using probabilistic techniques, we assured that composition and diversity were orthogonal to each other and to other potential drivers of landscape functioning. We also maximized spatial spread of plots, thereby minimizing spatial auto-correlation. Vegetation activity and land-surface phenology in these landscapes was inferred based on satellite-sensed (MODIS) vegetation indices, yielding a time-series of 20 years at a spatial resolution of 250 m.
Results/Conclusions
In both data sets, we found that landscapes with a higher land-cover type richness had a higher landscape-level productivity. In Europe, landscapes composed of 3 to 4 land-cover types were 15-30% more productivity (P=0.001) relative to single land-cover landscapes, and productivity was temporally more stable over the 20-year observation period (P=0.05). Effects similar in magnitude and statistical significance were also found for North America. At least in the European study, landscape-level effects appeared to be independent of local plant species diversity, suggesting that emergent mechanisms not found at smaller scales were at play. Such effects may represent a novel class of mechanisms that underpin diversity effects at the landscape level. We contend that landscape-level diversity–functioning relationships deserve increased attention, not least because they underlie the delivery of ecosystem services to humans.