Abstract: Biodiversity and ecosystem function (BEF) studies aim to understand how ecosystems respond to a gradient of species diversity. Diversity-Interactions models analyse the BEF relationship by relating an ecosystem function response of a community to the identity of the species in the community, the initial proportions of the species and their interactions, and community richness. It can also adjust for the density of individual species or overall. A strength of the modelling approach is its ability to describe the BEF relationship using a broad definition of species diversity that goes beyond single dimensional captures of species diversity such as richness alone. The recently developed ‘DImodels’ R package provides a user-friendly way to implement this modelling approach.
In this work, we aim to: 1) introduce the DImodels R package and highlight its capabilities, 2) compare the Diversity-Interactions modelling approach to other commonly used BEF approaches, 3) present novel visualisations developed to identify patterns in ecosystem function across the multiple facets of species diversity and 4) illustrate our approaches by applying them to data from a Norwegian grassland biodiversity experiment.
In the Norwegian experiment, a species diversity gradient was established across 60 plots from a pool of seven agronomic species, including five grass and two legume species. There were monocultures of each of the seven species and mixtures of up to seven species arranged according to a simplex design. The species diversity gradient was repeated across a regular and low nitrogen fertiliser supply. Applying the modelling approaches to the Norwegian data, we showed strong benefits of mixing the agronomic species for biomass yield, in particular when species from each of the functional groups (grass and legume) were mixed. The strength of the interaction effects were maximised under the low fertiliser treatment.
A major strength of the DI modelling framework is its ability to decompose the various elements of species diversity, including species identity (composition), species richness and evenness, and the effects of species interactions. The DImodels R package is useful for implementing the approach as it can easily compute and compare various forms of the species interactions and can automatically test for non-linearities in the species interactions form.