View more stories by categories: Research View more stories by categories: Synthesis


We set out to answer the general question: How do metacommunity dynamics mediate community responses to disturbance across the ecosystems represented in the LTER network? Metacommunity theory provides a framework to predict when different types of community assembly processes should control the composition of the species pool both at local and regional scales. Thus, metacommunity theory has the potential to explain why different trends in long-term biodiversity data occur across different taxonomic groups in ecosystems that experience different disturbance regimes. Initially, our specific objectives for the synthesis group were to:

  1. Characterize metacommunity stability and sensitivity to environmental fluctuations (disturbance)
  2. Characterize predictors of stability (metacommunity parameters)
  3. Assess how well metacommunity parameters predict stability across LTER datasets
  4. Use metacommunity simulations to conduct a sensitivity analysis of links between predictors and measures of stability

Evolving Goals and Challenges

In hindsight, our first objective was more ambitious than we realized. At our first working group meeting, to tackle the first objective above, we intended to spend our time using existing tools (i.e., R packages) to

quantify variability for as many LTER biodiversity data sets as we could. However, during that first meeting, participants identified a gap in how we assess variability in metacommunities—there was not a good metric to describe scaling of temporal variation in community composition in a metacommunity. Consequently, a large effort from our group has focused on developing a compositional variability metric and an R package to easily apply the analysis to long-term site-by-species-by-time data sets.

In addition to the tangent of developing a new metric and R package, we found it was much more difficult than we had anticipated to find long-term data sets that were intercomparable and appropriate for our analysis. Luckily, as part of a separate project, Tom Miller and Aldo Compagnoni were in the process of building the Popler R package to access and query a database of LTER population and community data sets. We were able to beta-test this tool and it provided an invaluable resource for finding candidate data sets for our project. However, even after identifying good candidate data sets, it took considerable effort to develop reproducible workflows to access, clean, and harmonize the data.

A third challenge that we quickly identified during the initial stages of data discovery was the lack of consistency and discoverability of ancillary data (e.g., spatial and environmental companion data sets to the biodiversity data that served as our response variable). In some cases, spatial and environmental data were found in the same data set as the biodiversity data. In some cases, environmental and spatial data were published elsewhere, but links among data sets were spelled out in the documentation. In other cases, we were not able to identify any supplemental data beyond the biodiversity data.

These challenges led us to re-evaluate our goals for subsequent meetings. Given the interests and skill set of the working group members, we have prioritized:

  1. the development and assessment of our compositional variability metric and its R package,
  2. the documentation and curation of a collection of ~30 harmonized (i.e., comparable) LTER metacommunity data sets along with documented reproducible workflows that pull from the original source data,
  3. an assessment of environmental variability based off remote imagery corresponding with the location and timespan of a subset of the curated biodiversity data sets, and
  4. a synthesis of these products that should ultimately provide insight for our original research question—preliminary results of our final synthesis were presented at this past ESA2019 meeting in Louisville, KY.

Meeting structure

We had four in-person meetings. They included graduate students, postdocs, and mid-career scientists who self-identified as researchers who were interested applying the metacommunity framework to their system (i.e., their taxonomic group of interest at their LTER site). Each meeting kicked off on Monday morning with a set of presentations to provide a foundation for the rest of the week, and an afternoon of discussions to determine goals for the week. The remainder of the week involved break-out groups to focus on coding, data-munging, and writing, with periodic report-outs to the group as a whole to gauge progress. In the interim, between meetings, we have kept up semi-regular zoom calls to maintain progress on the project.

Looking forward

Our last objective was to assess empirical links between predictors and measures of metacommunity stability/variability using modeling. Our group has spent some time discussing this topic, but it has largely been descoped because we have just run out of time. That said, a spin-off from our group is exploring how our synthesis results might inform how mechanism-based metacommunity models might be applied to predict how biodiversity will respond in non-stationary ecosystems.

Advice for proposers

I would recommend future synthesis working groups find funding to support a dedicated postdoc or graduate student. Our group was comprised largely of early career scientists (postdocs) who got jobs — which is a great problem for the group to have — but progress on working group goals has been slowed as participants’ careers have matured.

by Eric Sokol

—on behalf of the Lead PIs of A synthesis to identify how metacommunity dynamics
community responses to disturbance across the ecosystems represented in the LTER network

Publications in review or to be submitted

Voelker, N. M., P. L. Zarnetske, N. I. Wisnoski, J. D. Tonkin, C. Swan, S. Record,  L Marazzi, N Lany, T Lamy, A Compagnoni,  M. C. N. Castorani, R. Andrade, and E. R. Sokol. In revision. Novel insights to be gained from applying metacommunity theory to long-term biodiversity data. Ecosphere

Lamy, T., N. I. Wisnoski, R. Andrade, M.C.N. Castorani, A. Compagnoni, N. Lany, L. Marazzi, S. Record, C. M. Swan, J. D. Tonkin, N. Voelker, S. Wang, P. L. Zarnetske, and E. R. Sokol. In prep. The dual nature of metacommunity variability

Sokol, E. R., R. Andrade, MCN Castorani, C. Catano, A Compagnoni, T. Lamy, N. Lany, L. Marazzi, S. Record, A. Smith, C. Swan, J. D. Tonkin, N. Voelker, N. Wisnoski, and P. Zarnetske. In prep. Using long term data to understand links between environmental variability and metacommunity stability.

R Package

The ltmc (Long-term metacommunities) package for R

Data products

The curated LTER metacommunity data set and derived metacommunity variability metrics, along with the code to reproduce the data set from original source data are currently available on our github page, but will be submitted to the EDI data portal when Sokol et al. is submitted for peer review.