A multivariate view on biodiversity multifunctionality relationships reveals functional trade-offs and increased functioning at high diversity

Background/Question/Methods

Can diversity simultaneously affect a wide variety of different ecosystem functions? Despite first studies analyzing effects of diversity on multiple ecosystem functions having revealed stronger diversity effects than studies looking at single functions this questions remains little studied. As an often used approach, the number of species contributing to functioning has been shown to increases with the number of studies, years, or functions considered. Yet, conclusions based on the number of contributing species have been criticized and recent methodological advances have proposed alternatives.

Here we used multivariate statistics to investigate the relationship between plant diversity and multifunctionality. We based our analysis on more than 100 functions measured along an experimental gradient of grassland plant diversity ranging from 1 to 60 species. The set of ecosystem functions included various above- and below-ground processes, e.g. cover, LAI, plant biomass, soil nutrients, and abundance data of plant-associated invertebrates such as earthworms, pollinators and herbivores. Using principle component analysis based on the value of each function in each plot we investigated (1) correlations and trade-offs between functions, (2) the functional fingerprint and (3) the overall level of expressed functioning of each plot and (4) the relationships of all these parameters to plant species richness.

Results/Conclusions

Relationships between the investigated functions spanned the whole spectrum from strong positive correlation (association) to almost perfect negative correlation (trade-off). Intermediate between these extremes, a large number of functions were independent from each other, thus showed correlations around zero. Consequently, a large number of axes (23) was needed to explain at least 75% of the variation observed in the multifunctional space. Plant diversity correlated strongly with the first principle component axis (r=0.67, p<<0.001) while axes of higher order did not show any relationships with plant diversity. This indicates that especially the functions associated with the first principal component axis change along the plant diversity gradient. To calculate an index of multifunctionality, we extendend the “averaging approach” from single functions to a multivariate measure by summing scores for the first 30 principle component axes. The resulting multifunctionality index increased highly significantly with plant diversity from predominantly negative values at low diversity to positive values at high diversity (F_1,76=8.26; p=0.005). Thus, plots of high diversity supported more functions at above average levels than low diversity plots. Results from our multivariate approach are compared to other proposed approaches to measure multifunctionality.