Karl Andraczek (@KarlAndraczek), from the German Centre for Integrative Biodiversity Research (iDiv) Halle-Jena-Leipzig, discusses his article: Weak reciprocal relationships between productivity and plant biodiversity in managed grasslands
Setting The Scene
Global change impacts both plant diversity and ecosystem functioning with detrimental consequences for nature’s contributions to people. Predicting these consequences has emerged as a focal topic in ecology, but making such predictions is difficult. A long-standing challenge is to tease apart complex feedback effects between diversity and productivity, particularly when diversity and productivity are sampled at only one point in time. Another challenge is to account for their complex interaction with other biotic or abiotic factors because both diversity and productivity are jointly controlled by numerous factors including management activities, and this likely confounds their observed relationships in natural systems. To overcome these challenges, we need new analytical tools and frameworks to ‘untangle’ their feedbacks!
Entangled in feedback loops
While drafting my PhD proposal, my supervisor Fons van der Plas (@fonsvanderplas) suggested solving this problem using complex structural equation models. However, once I started my PhD and after endless discussions, trials, and literature searches, we quickly felt entangled in the feedback loops. Then, after presenting my work at Utrecht University, we got a promising tip from Laura Dee (@LauraEllenDee; CU Boulder, US) and her team from the other side of the globe. They were using an innovative approach to study causal relationships in observational settings. After reaching out to her, she was eager to help, and through our discussions, an idea began to take shape.
The Study
In our study, we proposed a solution to resolve feedback loops between diversity and productivity by using methods from the field of ‘causal inference’ which are rarely utilized in ecology, and combining these with high-resolution temporal data on plant diversity and productivity measured at multiple times across the growing season.
Our novel approach has two major advantages:
(1) We can quantify the causal effect of biodiversity on productivity, and vice versa, by simultaneously accounting for unobserved confounding factors (i.e., irrespective of whether they are measured or not!).
(2) By using temporal data we can tease apart feedback effects by constructing temporal sequences of cause-and-effect events (e.g., spring species richness as a cause of summer productivity). By doing so, we can detect effects of biodiversity on productivity, or vice versa, even if they are temporally delayed (e.g., through delayed effects of plant-soil feedbacks).
Importantly, we also compared our novel statistical approach (two-way fixed effects models) with methods more commonly used in ecology (in this case, mixed effect models) to assess if this would change our ecological conclusions.
Together, Fons van der Plas, Laura Dee, and I, along with support from Alexandra Weigelt (co-supervisor) and many other collaborators, we tested this framework, by collecting seasonal observational data over two years, across 150 managed grassland sites in Germany as part of the large-scale Biodiversity Exploratories project (@BExplo_research). This meant hands-on work: crawling through grasslands, sweating on sun-exposed dry meadows, wading through swampy sedges, and occasionally making a run from overcurious cows.
Key Findings
We found only weak feedback effects between plant diversity and productivity in our managed grasslands, likely because of intensive management practices. Surprisingly, we also found weak effects of productivity on biodiversity, contrasting both the often strongly negative relationships observed in managed grasslands and the positive relationships seen in biodiversity-ecosystem functioning experiments. We argue that the weak effects were likely because of regular biomass removal by grazing/mowing which can prevent negative effects of productivity on biodiversity by alleviating light competition, indicating that management is a key factor modifying diversity-productivity relationships.
However, our ecological conclusion was highly sensitive to the models used: estimated effects from conventional models were more susceptible to bias from unobserved confounding variables (e.g., climatic shocks), while our modelling approach remained robust. Yet, to improve our predictions, longer time-series are needed!
Synthesis And Future Work
Disentangling feedback effects in natural systems is a daunting task – but there is hope! Alternative approaches can help us overcome the limitations of many commonly-used approaches and in turn open up new avenues for investigation.
Our findings highlight that temporal data, in combination with methods from ‘causal inference’ form a promising tool to better understand feedback loops between plant diversity and productivity in natural systems. This also has important implications for elucidating how both biodiversity and ecosystem functioning will respond to climate change, which will be the focus of my future work as part of a Marie Curie postdoc fellowship (at Utrecht University, Netherlands and CU Boulder, US). Even beyond diversity-productivity relationships, our framework could provide novel insights into various types of feedback loops in natural systems.
