Plant functional traits are dynamic predictors of ecosystem functioning in variable environments

Abstract

Abstract A central goal at the interface of ecology and conservation is understanding how the relationship between biodiversity and ecosystem function (B–EF) will shift with changing climate. Despite recent theoretical advances, studies which examine temporal variation in the functional traits and mechanisms (mass ratio effects and niche complementarity effects) that underpin the B–EF relationship are lacking. Here, we use 13 years of data on plant species composition, plant traits, local‐scale abiotic variables, above‐ground net primary productivity (ANPP), and climate from the alpine tundra of Colorado (USA) to investigate temporal dynamics in the B–EF relationship. To assess how changing climatic conditions may alter the B–EF relationship, we built structural equation models (SEMs) for 11 traits across 13 years and evaluated the power of different trait SEMs to predict ANPP, as well as the relative contributions of mass ratio effects (community‐weighted mean trait values; CWM), niche complementarity effects (functional dispersion; FDis) and local abiotic variables. Additionally, we coupled linear mixed effects models with Multimodel inference methods to assess how inclusion of trait–climate interactions might improve our ability to predict ANPP through time. In every year, at least one SEM exhibited good fit, explaining between 19.6% and 57.2% of the variation in ANPP. However, the identity of the trait which best explained ANPP changed depending on winter precipitation, with leaf area, plant height and foliar nitrogen isotope content (δ15N) SEMs performing best in high, middle and low precipitation years, respectively. Regardless of trait identity, CWMs exerted a stronger influence on ANPP than FDis and total biotic effects were always greater than total abiotic effects. Multimodel inference reinforced the results of SEM analysis, with the inclusion of climate–trait interactions marginally improving our ability to predict ANPP through time. Synthesis. Our results suggest that temporal variation in climatic conditions influences which traits, mechanisms and abiotic variables were most responsible for driving the B–EF relationship. Importantly, our findings suggest that future research should consider temporal variability in the B–EF relationship, particularly how the predictive power of individual functional traits and abiotic variables may fluctuate as conditions shift due to climate change.