The dependence of ecosystem function on biodiversity has been the topic of intense research in recent decades. This field was sparked in part by a provocative grassland experiment conducted at Cedar Creek, MN in the late 80s/early 90s. This experiment led Tilman & Downing (1994) to propose that species richness had a positive, saturating effect on the drought resistance of communities. However, details of this study were criticized, including the effects of a ‘hidden’ nitrogen addition treatment, which could affect the richness and productivity of communities (Huston 1997). The debate surrounding this early study raises a challenging issue for studies of biodiversity/ecosystem function: How do we untangle the ecosystem effects of diversity when richness is not independently controlled, but rather emerges from community assembly following an environmental change? We revisit data from the Cedar Creek experiment, applying techniques based on the Price equation (Fox & Kerr 2012). This allows us to disentangle the intertwined effects of community assembly, changes in species richness and functioning, and environmental perturbations (nitrogen addition and drought). We resolve prior critiques of the experiment, while emphasizing how our analyses could be applied more generally to understanding variation in ecosystem function across natural communities.
Nitrogen enrichment prior to the drought had strong effects on the richness, composition, and function (above-ground biomass production) of plant communities. Enrichment drove communities away from a saturating biomass/richness relationship observed in control plots, lowering richness while increasing total biomass. Enriched communities experienced significant turnover driven by losses and gains of species. The divergent assembly of these communities set the stage for the subsequent drought (1986-88) effects, which were central to Tilman & Downing’s analyses. Accounting for differences among communities due to nitrogen treatment, we find no effect of richness on drought resistance (change in biomass over the drought; p-value = 0.4377). Instead, nitrogen treatment was the primary driver (p < 0.0001), as Huston suggested. Our Price equation analyses also provide detailed insights into how communities changed: low diversity/high N communities lost significant biomass, but few species, while higher diversity/low N communities lost many low functioning species. Our results offer clear examples of how assembly produces non-random patterns of community richness and composition, which in turn affect ecosystem function. Conceptual and analytical approaches need to account for assembly’s effects, as we attempt to relate the significant discoveries of experimental biodiversity/ecosystem function research to understanding how ecosystems function in nature.