Intraguild predation (IGP) simulation
In 2021, I worked with Dr. Akira Terui's lab on simulations of intraguild predation in branching habitat networks, with the goal of assessing how ecosystem complexity influences food chain length. We summarized some of this work in a pre-print, available here.
​
In addition, we are in the process of building an R package which allows for the simulation of branching networks and the simulation of community dynamics within patches. We are hoping to release this package in late 2021 or early 2022.
​
You can see the developmental version on Dr. Terui's GitHub account here.
Selected projects and publications are described here. For a full list, visit my Google Scholar page by clicking on the Publications link above.
Individual Size Distributions
Individual size distributions, also known as body mass-abundance, or size spectra, are one of the few "universal" patterns in biological communities. Simply stated, there are lots more small things than large things in a community. This is true in soil communities, which have untold numbers of bacterial and fungal cells, some herbivorous arthropods, and a handful of (relatively) large predatory nematodes and mites, and this is also true in the ocean, from seemingly infinite numbers of phytoplankton, all the way to a blue whale, one of the largest creatures to ever exist on earth (See figure below from Blanchard et al. 2017). Increasing body size is also highly correlated with trophic position, i.e., prey is small, predators are big, with some notable exceptions. Therefore, characterizing the ISD relationship also tells us a lot about the food web structure, without having to conduct those pesky diet surveys.
​
A remarkable feature of the ISD relationship is that the decline in large-bodied individuals relative to small-bodied individuals (i.e., the slope) is very similar across undisturbed communities and habitats. The ISD slopes are so similar that deviations from this have been hypothesized as a "universal" indicator of ecological status. Indeed, changes in these slopes have been detected in response to a number of human impacts, including overfishing, land use change, exposure to contaminants, and legacy effects of mining (See AMD work below).
​
However, the slope of the ISD relationship also varies in response to some natural gradients, including resource availability, productivity, and temperature.
Understanding both the natural and anthropogenic sources of variation in ISD relationships has motivated much of my research. Knowing when and where ISD relationships "should" naturally deviate from what we expect will aid in our ability to assess the ecological status of communities which are affected by human activities.
Environmental drivers of ISD relationships across the NEON stream sites
With Drs. Jeff Wesner and Jim Junker
Increasing temperatures and resource availability are likely to influence the pattern of biomass distribution within communities (i.e., ISD described above). Increasing temperature is thought to cause a decline (steepening) of the ISD slope, whereas increasing resource supply is thought to increase (shallowing) of the slope (See conceptual figure above). However, resource availability is also likely to change in response to increasing temperatures. Therefore, understanding the main effects of these two drivers, and their possible interactive effects, remains an open question to ecologists. As the effects of human-induced climate change continue to manifest, it will become increasingly important to understand how stream communities respond to changing temperature regimes.
​
We recently found that increasing temperatures was related to steepening ISD slope in macroinvertebrate communities across the NEON streams (Pomeranz et al. 2021). In addition, we received an NSF Macrosystems Biology and NEON-Enabled Science grant to which will allow us to expand the analysis by quantifying resource availability and including fish data into our ISD estimation. Read the funded abstract here.
Community responses to legacy impacts of acid mine drainage (AMD)
AMD effects to ISD relationsip
For my PhD research, I studied the effects of AMD on stream communities across the south island of New Zealand. My first chapter looked at using ISDs (described above) to assess the ecological status of communities across a gradient of AMD inputs. We found that ISD relationships changed consistently in response to increasing AMD, with impacted communities characterized by reduced community abundance, reduced range of body sizes, and an shallower decline (i.e., slope approaching 0) of large bodied individuals. We expected the slope to be steeper because of a decline in larger-bodied individuals. Instead, we found a complete loss of large-bodied individuals (horizontal shift in panel A below), and hypothesize that their removal released mid-sized individuals from predation pressure, thereby supporting a higher abundance than expected (shallower slope in panel B).
In addition to finding consistent effects of increasing AMD stress, all of our un-impacted sites had nearly identical slopes. This is an important finding, because the sites that were free from AMD stress were situated across a natural gradient of pH conditions (~4-7 pH). The remarkable consistency of ISD relationships across this natural gradient supports the hypothesis that biomass distributions within undisturbed communities are organized according to some higher-level processes (i.e., metabolic ecology, energy use efficiency), and that human activities can alter these patterns.
​
For more on this, click on the image below to access a pdf of the published manuscript.
​
Data and code are available for this analysis at Data Dryad and GitHub, respectively.
Food web structure and stability in AMD streams
ISD relationships are strongly correlated with food web structure, particularly in aquatic communities where body size is a strong driver of trophic position. In addition, a large body of theoretical and empirical evidence suggests that food web structure is related to the stability of communities. Based on the results of my first PhD chapter, I wanted to explore this idea further. Specifically, I inferred the structure of stream food webs across the AMD gradient, and performed a community stability analysis. I found that impacted stream communities actually increased in stability compared with reference communities.
​
Increased community stability in impacted streams has management and restoration implications. For example, restoration activities in mine-impacted streams generally have a goal of the return of desirable species, usually fish. However, a healthy, functioning macroinvertebrate community is necessary to support higher trophic levels like fish. Communities with a high stability may be more resistant to species recolonization, thereby slowing the return to a pre-impacted state. We hypothesize that disturbance events (i.e., flooding), may be necessary after remediation efforts to facilitate the colonization of desirable species and the re-assembly of a reference condition food web.
​
For more on this, click on the image below to access a pdf of the published manuscript.
​
Data and Code for the full analysis are available. Additionally, a simplified example of the method using Bernoulli trials (see section below manuscript title) is available on GitHub
In addition to the main findings of this study, I was particularly excited about the methodology that I developed during the analysis phase of this project. Specifically, I wanted to incorporate uncertainty in the food web structure, particularly since I did not assess this empirically with gut content or stable isotope analyses.
Briefly, I estimated the probability of all pairwise species interactions within a community using niche and neutral processes (see Inferring predator-prey interactions below). Estimating species interaction probabilities led to the creation of square matrices, where each matrix element was the interaction probability between species i and j. The probability matrices were converted to binary adjacency matrices (i.e., 1 = interaction, 0 = no interaction) by conducting numerous Bernoulli trials (See conceptual figure below). The adjacency matrices were then available to estimate food web parameters (i.e., number of links, connectance, generality, etc.), and perform stability analyses on. Because many Bernoulli trials were conducted (N = 250), this allowed us to estimate a range of possible food web topologies, and estimate the general response.
​
I have always been struck in the past that a large number of published food web studies consider the structure to be a fixed entity, whereas in nature the interactions are constantly changing in response to variations in environmental conditions, prey resources, seasonal conditions, etc. This method is far from addressing all of those shortcomings, however, I think it is a step in the right direction.
Inferring predator-prey interactions
Click on the manuscript title to open a pdf.
More details coming soon!
