Most of my work since 2013 has involved developing simulation models, exploring mathematical models numerically, and writing analysis scripts and workflows for statistical ecology projects. I strive to make sure that my code is replicable and accessible. If someone else can piggy-back off of my code to carry out their own analysis or construct their own model, that’s a huge win for me. In recent years the Tidyverse team has put together a really efficient workflow for developing packages. I’ve started to incorporate package development into my workflow, but many of my past projects involved one-off models or analyses that didn’t seem to warrant learning and implementing R package development. Below is the running list of these projects; feel free to reach out if you need help or clarification in implementing or building upon any of these.

Plant defense synergies and antagonisms affect performance of specialist herbivores of common milkweed. Edwards, Agrawal and Ellner. Accepted in Ecology. Data + code on GitHub.
This project involved substantial analyses of structured plant using mixed effects models and random forests, as well as some supporting simulations and mathematical explroations. A highlight in this repository is the code to implement a novel method for using Random Forests to evaluate possible synergies and antagonisms between traits.

Using structured decision making to guide habitat restoration for butterflies: a case study of Oregon Silverspots. Doll, Converse, Edwards and Schultz. 2022. Journal of Insect Conservation. Data + code on GitHub.
In this project I was the quantitative mentor for masters student Cassie Doll. I provided supervision and guidance on the analyses and simulations, but ultimately all credit for the repository belongs to her.

Estimating abundance and phenology from transect count data with
GLMs. Edwards and Crone. 2021. Oikos. Data + code on Dryad, data + code on GitHub.
In this project I developed a new method for fitting structured butterfly transect data to gaussian curves with regression models. Our main goal was to provide this tool to other ecologists, and the code has a thoroughly commented tutorial with that in mind.

Evolved phenological cueing strategies show variable responses to
climate change. Edwards and Yang. 2021. American Naturalist. Code and data on Dryad.
Here I developed a mathematical model of organismal responses to climate variables, and simulated the evolution of populations across 78 sites in North America and Hawaii. My goal in with the code was to provide a flexible simulation framework to allow others to re-use the code to evaluate the effects of other climate variables, or use the climate data from other sites. The scale of the data we used (thousands of years worth of daily measures of climate variables) and the scale of the resulting simulation results preclude storing this on GitHub.

Changes in phenology and abundance of an at-risk butterfly. Bonoan, Crone, Edwards, Schultz. 2021. Journal of Insect Conservation. Data + code on GitHub.
In this project we implemented the GLM approach of Edwards and Crone 2021 to transect data for an endangered butterfly species. This code can serve as a more involved example of our methods, but otherwise is tailored to this system. My role was as the data science collaborator; the code and analyses in the repository was generally mine (excluding 3_scripts/Online Resource 2.rmd).

Aggregating fields of annual crops to form larger-scale monocultures can suppress dispersal-limited herbivores. Edwards, Rosenheim, Segoli. 2018. Matlab code provided as supplementary files.
In this project I developed and implemented a cellular automaton model of herbivores moving through agricultural fields. This was the culmination of work I did in my undergraduate, and reflects my earlier, more limited approaches to organizing code. That said, I was particularly proud of developing (probably re-inventing) an approach to represent complex dispersal in a cellular automaton model using a series of elementwise multiplications. For those interested in this kind of work, I would encourage you to look at more modern tools for spatially explicit simulations, like NetLogo and its R extension. Coding these simulations from scratch was a useful exercise, but we now have more efficient approaches to use (if they fit our needs).

Linking demography with drivers: climate and competition. Teller, Adler, Edwards, Hooker, Ellner. 2016. Methods in Ecology and Evolution. R code provided in Appendix S1, S2, and Data S1.
In this project we developed a novel approach to use functional smoothing methods to estimate the effects of climate and spatial competition from demographic data. My role was to implemented the spatial competition methods; I wrote the majority of the code relating to that. Note that aspects of this approach can now be implemented using the dlnm R package.