Functional traits as predictors of community assembly and ecosystem function in restored prairies

Background/Question/Methods Human land use increasingly threatens biodiversity and functioning of natural ecosystems. Given the rate of habitat loss, restoration –reinstatement of biological communities and ecosystem functionality – will be critical to conservation, yet, restoration outcomes are notoriously variable. Trait-based approaches may provide mechanistic generalizable predictions of how restoration goals can be met, but links between environment, traits, and ecosystem function are poorly understood. In 29 prairies restored from former agricultural lands, we asked how environmental conditions determine trait composition, how trait composition is related to ecosystem function, and if traits that respond to environmental conditions or affect ecosystem function are correlated. We determined community weighted mean trait values using plant species abundance to weight traits: plant height, vegetative mass, seed mass, and specific leaf area. We collected data on environmental conditions thought to determine trait composition: fire frequency, surrounding landscape composition, land-use history, site age, and soil characteristics. We also considered whether environmental conditions may directly influence site to site variation in ecosystem function. Finally, we collected ecosystem function data for each site: aboveground biomass, floral cover, decomposition, and predation of arthropods and seeds. We used structural equation modeling to determine relationships between environment, traits, and ecosystem functions.

Results/Conclusions Our ability to explain variation in ecosystem function using traits and environmental conditions ranged widely (R­­­­²=0.2 to 0.76). We found evidence for trait-based community assembly, as all functional traits were explained by environmental conditions (R­­­­²=0.17 to 0.37). For example, fire frequency and plant height were positively correlated. Some functions were strongly influenced by trait composition, such as plant height and floral cover, which were negatively correlated. Traits also ranged in the degree to which they were correlated with one another (r=0.7-0.92) and this provided insights into variation in function. For example, sand content and vegetative mass were negatively related, vegetative mass was highly correlated with plant height, and plant height was negatively related to floral cover.  Certain functions were explained by environmental conditions alone (e.g., sand content was negatively correlated with decomposition).  Together, these results suggest that both trait composition and environmental conditions play a role in shaping ecosystem function during restoration, and the importance of each is dependent on the function of interest. A trait-based approach to restoration can afford a mechanistic understanding of community assembly and ecosystem function. By using management techniques to alter environmental conditions, we might predictably influence restoration outcomes.