Plot-level experiments have shown that biodiversity-ecosystem functioning (BEF) relationships occur at organizational levels ranging from genes to communities. Whether BEF patterns can be extrapolated to real-world ecosystems and the landscape level remains unclear, although that is where ecosystem services for humans are provided. Using networks of large field plots that span study regions around the URPP sites, we will analyze BEF relationships at the landscape level using land cover maps, species inventories, eDNA samples, and EF-indicators derived from satellite images. Using statistical modeling, we will test (1) for BEF-effects at landscape scale (species richness-ecosystem functioning relationships) and at landscape level (landscape diversity-landscape functioning relationships), and (2) whether these relationship differ between biomes or landscape types.
Studies in which biodiversity was manipulated directly have provided detailed insight into causal effects on ecosystem functioning and some of the underlying mechanisms. However, these studies have been criticized for their relative simplicity and their small spatial scale. It remains unclear whether experimental results can be extrapolated to real-world ecosystems and the landscape level, even though it is there that ecosystem services of enormous global economic value arise. Real-world ecosystems typically are more complex, closer to steady-state and interconnected at the landscape scale, leading to landscape-level “meta-ecosystems” and thus landscape diversity. In contrast to experiments, real-world BEF relationships also occur in a variable environmental contexts (e.g. heterogeneity, environmental adversity, regional species pool).
Our objective is to apply knowledge gained from experiments to develop new approaches to study BEF relationships at landscape scale and landscape level. Specifically, we ask:
(1) are species richness-ecosystem functioning relationships in the real world at large plot scale comparable to positive BEF relationships in experiments? Does a diverse landscape composed of a mosaic of different ecosystems (i.e. land-cover and vegetation types) perform “better” than a less diverse one? Such patterns may occur because of mechanisms that operate at spatial scales larger than small plots, including the exchange of genes and species between ecosystems and abiotic interactions (e.g. flows of energy and resources).
(2) do BEF relationships in (1) differ between biomes? Under environmentally adverse conditions, community assembly may be more strongly driven by environmental filtering than by niche partitioning. Under benign conditions, limiting similarity may be more important, or, if species have similar fitness, similar species may coexist (“neutral” biodiversity).
(3) does diversity promote the stability of EF? Stability effects in plot-scale experiments have been analyzed with respect to contributions from so called “asynchrony” and “portfolio” effects. We propose to extend these approaches to landscape-level stability in space and time.
Using design principles from field experiments, we will establish networks of large field plots (≥1 km2) systematically spanning gradients in landscape diversity in study regions around the URPP sites. A preliminary measure of landscape diversity will be obtained from the richness and composition of land-cover types within candidate plots. Afterwards, more detailed measures of landscape and species diversity will be derived from sources such as the Global Forest Biodiversity Initiative (>1 million plots; http://www.gfbinitiative.org) and local ground-based inventories. In collaboration with Florian Altermatt, we will further assess the diversity of multiple taxa by sampling eDNA from water bodies that integrate genetic material at plot scale. Proxies characterizing ecosystem and landscape functioning of the same plots (integrated across the different land-cover patches within a plot) will be derived from satellite images (e.g. MODIS EVI).
We will analyze our data applying methods from experimental BEF research. These include mixed-effects and structural equation modelling. The scale-dependency of patterns will be assessed by combining analyses of data that have been aggregated to different spatial resolutions.