Trait identity and functional diversity co-drive response of ecosystem productivity to nitrogen enrichment

Background/Question/Methods: Exploring the mechanisms underlying the change in ecosystem productivity under anthropogenic nitrogen (N) inputs is a fundamental ecological question. It has been advocated that functional traits rather than environmental factors and plant species diversity are the keys to linking ecosystem productivity with environmental changes. However, very few studies have considered the joint effects of environmental factors, species richness and functional traits on ecosystem productivity under increasing N inputs. We conducted a N-manipulation experiment in a Tibetan alpine steppe since 2013, and constructed structural equation models to assess the effects of N-induced changes in environmental factors, species richness and traits metrics (mean, variance, skewness and kurtosis) on gross ecosystem productivity as well as three resource use efficiencies (water, light, and phosphorus (P) use efficiencies), based on actual measurements during the peak growing season in 2016.

Results/Conclusions: Our results showed that both light and P use efficiencies decreased under N enrichment, largely due to the N-induced decline in functional diversity of leaf P concentration (illustrated by the increasing kurtosis-leaf P concentration). However, both gross ecosystem productivity and water use efficiency exhibited initial increases and subsequent slight decreases with increasing N additions. The nonlinear pattern was closely associated with both the increased morphological trait (i.e., mean-leaf area) and the decreased diversity of leaf P concentration. These results demonstrate that N-induced changes in functional traits may have dual effects on ecosystem productivity: the stimulating effects of the dominant trait identity via increasing canopy light interception vs. the inhibiting effect of decreasing trait diversity via declining resource use efficiencies. Inclusion of functional traits is crucial to enhance the robustness of land surface models for predicting the response of ecosystem functions to increasing N inputs.