My research aims to build quantitative, predictive understanding of ecological communities. It falls into three broad categories: mathematical theory, field-based natural history, and predictive models of communities. I try to integrate these fields in my work, with the goal of developing tools that are quantitatively and logically rigorous, but are sufficiently informed by biology to remain realistic, and simple enough to be parameterized for real-world systems.
Ecological communities include a lot of moving parts, with the direction and magnitude of interactions between individual species or species and environments varying greatly across space and time. This makes it difficult to analyze them using traditional tools. My interest in theoretical ecology is driven by a need for mathematical, statistical, and algorithmic tools that can be used to analyze and understand these complex systems. I also spend a lot of time working with and redesigning existing tools that I find especially useful or interesting. Examples of projects that I have worked on that fall in this category include:
1. Quantifying ecological stability in real world systems
Although there is ample theory relating to stability in ecological systems, actually measuring stability (and in particular the stability of coexistence) is often very difficult in practice. As part of the FIGS Workshop at UFZ and iDiv, we have identified several statistical rules that often govern how measurements of stability vary across different spatial and temporal scales, and across different numbers of observed species. Accounting for this scale dependence is especially important when comparing results across different studies, or when trying to extrapolate results from the scales at which systems are measured out to the scales at which important ecological processes operate. Results are available in a 2021 paper in Ecology Letters. Additionally, to make it easier to apply these scaling functions to real-world data, I have developed a toolset that can be applied to quantify different sources of variability (e.g. observation error, process noise, and deterministic dynamics) in complex, non-stationary systems. A description of the method is published in a 2022 article in Ecosphere, and is available in the pttstability R package.
2. Analyzing the effects of space and time
In many ecological systems, community dynamics are strongly driven by three major spatial processes: dispersal limitation, negative density dependence, and environmental filtering. However, these processes can interact at many different spatial scales, making it difficult to determine the relative influences of each. Working with Matteo Detto, Helene Muller-Landau, and other researchers at STRI, we published a set of tools for analyzing spatial patterns that help separate the relative influences of these processes at different spatial scales. Our results are available in a 2018 paper in the Journal of Ecology.
3. Approximating mechanistic models of ecological communities
Especially in diverse ecological communities, it can be difficult to accurately estimate the strength of interactions among species. Working with Claudia Neuhauser, I developed a method for using variability in empirical estimates of pairwise interaction strengths (e.g. based on observed between-plot variability) to approximate underlying mechanistic processes (e.g. resource competition). We published our results in a 2018 issue of Theoretical Population Biology. In another project, I worked with Lindsay Turnbull and others to demonstrate how trade-offs between model complexity and generality can lead simple models to outperform more complex models when trying to extrapolate results to new conditions. These results are available in a 2020 issue of Journal of Ecology.
4. Accounting for incomplete sampling in biodiversity-ecosystem-functioning experiments
One of the most commonly applied methods for analyzing the effects of diversity on ecosystem functioning in experiments was published in a 2001 paper Michel Loreau and Andy Hector in Nature. One limitation of this approach has been that it requires monoculture data from all species in a community. Working with Kathryn Barry, Michel Loreau, and others at iDiv and UFZ, I developed an augmentation of the original partition that can be applied even when samples are missing for a subset of monocultures. The method is potentially also useful for analyzing the effects of biodiversity in natural systems, where monocultures are rarely available for all species. Our results are published in a 2019 paper in Methods in Ecology and Evolution.
5. Testing for causal interactions in complex systems
To build predictive models, it's important to first have a good understanding of how components of a system influence on another. In 2012, the Sugihara lab at the Scripts Institute at University of California, San Diego, published a numerical method for testing for causal relationships in complex systems where experimental data is lacking. In collaboration with them, I built an R package (multispatialCCM) that extends their method to work for short time series that are replicated spatially (i.e. plot replication). This is useful for many ecological systems, where plot-based replication is more common than long time series, and experiments can be difficult to perform. Our results are published in a 2015 paper in the journal Ecology. An extension of these methods, meant to forecast dynamics in complex, stochastic timeseries is published in a 2022 article in Ecosphere, and is available in the pttstability R package.
Field-based natural history
An enormous advantage of natural science is that the real world provides a definitively "correct" set of outcomes against which to test the predictions of the models that we dream up. Experiments and field observations help inspire simple abstractions of complex mechanisms, and help me make sure that the theory I work on remains relevant to ecological problems. In the past, the majority of my taxonomic and field experience has been with ant communities in North America and the Caribbean. Now, I work mostly in grasslands, especially prairies in the US Midwest. Field projects that I have worked on include:
1. Prairie plant community change in old field succession
Studies of old ecological succession - and especially of old field succession - were one of the first research programs that sought to explain how diverse communities of species influenced one another and their environments. Old field succession is particularly relevant today, as it describes the recovery of natural communities following disturbances such as farming or intensive grazing. Working at the Cedar Creek LTER (predominantly in experiment 14 and experiment 54) in 2016 and 2022, I led the two most recent re-surveys of a long-term "chronoseries" characterizing almost a century of succession across more than 20 old fields (the 2016 resurvey campaign was immortalized in a painting by resident artist Glenn Terry, shown above). Based on results from this, and past, surveys, we found that successional trajectories could be broadly explained as a function of field-level soil fertility, fire history, and weather. Results from the first of these re-surveys are published in a 2019 issue of Journal of Ecology.
2. Role of seed dispersal in shaping grassland communities
At Cedar Creek, dispersal limitation has been hypothesized to be a major driver of local plant community composition. To quantify differences in species dispersal abilities, I worked with Lauren Sullivan to generate estimates of seed dispersal kernels for the 50 most common herbaceous species in the region. To do so, we collected seed and plant traits, and designed a custom-made light meter that measured seed terminal velocities (picture above). We summarized our results, and design plans for the seed meter, in a 2018 article in Ecology.
3. Species-environment interactions at the microclimate level
Microclimates have major impacts on the distribution of many species, but are also impacted by ecological communities in turn. These interactions can be highly complex, and are thought to be especially influential in "extreme" environments, such as deserts or arctic systems. However, microclimates are also exceedingly difficult to measure - espeically in these kinds of extreme systems. To this end, I worked with two graduate students (L. Mühlbauer and G. Zavattoni, who led the paper), as well as my colleagues Risto Virtanen, Martin Grube, and Bettina Weber, to help design and test a distributed sensor network based on cheap, easily programmable microcontrollers. Over a two month period, we used this system to collect a total of ~3.5 million temperature and moisture measurements across 400 unique sampling points at the . Results, along with detailed building plans and all related code, are published in a 2023 article in Ecological Solutions and Evidence.
3. Ant community responses to plant diversity and experimental warming
While I was an undergraduate, I did my thesis work on ant communities working under Brian Farrell, Jessica Rykken, and Stefan Cover. With their help, I surveyed communities on the Boston Harbor Islands and Dominican Republic as part of an All Taxa Biodiversity Inventory (ATBI). We found that ant diversity on the harbor islands was associated with both island isolation and size, as expected from the theory of island biogeography, and published these results in a 2011 article in PLOS ONE. As an undergraduate, I also worked as a technician on an experiment testing ant community responses to global warming at the Harvest Forest. I also carried out a similar ant-based study in the BAC biodiversity and warming experiment at Cedar Creek. This is the only existing experiment that tests insect responses to both changes in plant biodiversity and warming (that I know of - if I'm wrong, please email me!), and includes one of the first checklists for ants in central Minnesota. The study is published in a 2022 article in Soil Organisms.
Predictive models of communities
For me, the most exciting piece of ecological research is to discover simple mechanisms that explain why the world works the way it does. The ultimate test of a theoretical model, therefore, is whether it can make quantitative predictions that match empirical observations in the real world. While this doesn't necessarily mean that the model is "right", it is at least a good indication that the model can be a useful predictive tool. Projects I have worked on that are related to prediction of community dynamics include:
1. Community-wide tests of resource competition
In order to test theoretical mechanisms that influence community dynamics, we first need simple, tractable models that are able to make accurate predictions of community dynamics based on simple, widely reproducible field measurements. Working with David Tilman and Clarence Lehman, I developed a resource competition model that uses measurements in experimental monocultures to predict the relative abundance of prairie plants in multi-species plots in the Big Biodiversity experiment at Cedar Creek. The model is thus meant both as an empirical test of resource competition theories, as well as a predictive model that can be used in ecosystem engineering. The results are published in a 2018 issue of Ecology Letters.
2. Eco-evolutionary dynamics of salamanders
Although it is clear that both environmental and biological interactions shape show ecological communities assemble and evolve over time, it can be difficult to quantify these effects because of the large spatial and temporal scales across which these processes play out. Using data on the current distributions and phylogenetic history of salamanders in the Appalachian Mountains - a particularly well-studied lineage - Richard Barnes and used a individual-based model to simulate 65 million years of species interactions, historical climate change, and orogeny. We found that jointly considering all three processes provided the best reconstruction of species evolutionary histories. Results are published in a 2017 issue of American Naturalist.
3. Synthesis of multiple predictive models of community composition and coexistence (sMultiMod)
What are the major mechanisms determining coexistence and community composition in ecological communities, and how do these differ across systems and sites? Though many mechanisms are hypothesized to explain coexistence in ecological communities, there have been comparatively few tests to determine which mechanisms matter the most within any given site or system. In my current position, in collaboration with Stan Harpole and Helmut Hillebrand at iDiv, we are working to synthesize understanding from observational studies and experiments in grasslands and aquatic systems to quantify the relative importance of potential coexistence mechanisms. We focus on these systems because of the wide availability of existing theory and data. For more details, see here.
We are working to address the following questions: (1) What are the primary mechanisms that are currently hypothesized to be important for determining community composition in grassland and aquatic systems?; (2) How much support is there for each of these hypothesized mechanisms across several case studies from well-studied experimental systems?; (3) How can we develop predictive models that identify when and why mechanisms’ influences change across systems and sites?
One of the main lessons that we have learned is that empirical systems pose a particularly thorny problem for studying coexistence, as we can usually only observe a fraction of the full range of spatial and temporal scales that are necessary for maintaining coexistence. We summarize this problem, and discuss how cross-scale analysis might partially address it, in a 2019 article in American Naturalist.
Page last updated: 1 September 2023