Research methods and other interests

Tools I use

Broadly, I use a combination of observational studies and manipulative experiments. Thus far, this has all been field based. Looking forward, I expect to extend my experimental work into gardens or greenhouses, and to rear bumble bee colonies in the lab to support lab and field experiments with captive colonies. Regardless of the approach or topic, though, a throughline of my work is the robust and creative use of quantitaitve tools.

A pervasive challenge in ecology is finding an analytical approach that appropriately captures our process of interest while effectively measuring or propgating uncertainty. One of my strengths is overcoming this challenge. Sometimes this means finding the right tool, and sometimes it means developing my own. For example…

• Hierarchical (generalized) linear models: the workhorse of ecology. Depending on need, I’ll run these using maximum likelihood or Bayesian MCMC¹, and have experience building complex, hierarchical models.

• Matrix and integral projection models for modeling population growth and describing life history. Where possible, I conduct these analyses in a Bayesian framework to effectively propagate uncertainty from individual vital rate models through to emergent properties like population growth or life history traits (e.g., Simpson et al 2026).

• Null models for effectively measuring significance. Many ecological analyses do not fit within the expectations or assumptions of standard statistical tests. As a result, I will develop randomization-based models to ask describe our expectation under the null hypothesis. This is common in community or network analyses (e.g., Simpson et al 2022, Aldercotte, Simpson, et al. 2026), but I have also used this approach in the context of landscape ecology to control false positive rates in a multi-species and multi-scale analysis (Simpson et al 2025), or to account for confounding process variance in population time series Aldercotte et al 2022.

• Population genetics: I am not a molecular ecologist, but I have used population genetics to estimate population parameters of interest (namely effective population density and dispersal distance; Simpson 2025). Looking forward, I also have many questions that could be most easily answered with genetics, and so it is likely I will continue dabbling in population genetics and collaborating with molecular ecologists.

• Tailored analyses and simulations: Not all questions can be answered solely by fitting a linear model to data, and instead require developing an protocol specific to the question and dataset. For example, I developed a resampling and bootstrapping procedure to measure effective to assess the effect of sampling extent on estimates of effective population density (Simpson 2025). I also use simulations to validate methods against violations of underlying assumptions (e.g., non-normally distributed group-level intercepts; Reily et al 2024).

Biodiversity-ecosystem function

Beginning my PhD, I expected this to be my main research focus. Along the way, however, I developed other interests. That said, I’ve never been able to 100% let it go or stop thinking about it.

Given the many ongoing threats to global biodiversity, it is worth considering whether, or to what extent, ecosystem functions depend on biodiversity per se. At the root of this question is whether declines in biodiversity will create negative feedbacks that accelerate ecosystem decay as global change progresses. Understanding such a process would enable us to better predict how communities and ecosystems will respond to global change. And from a practical standpoint, the relationship between biodiversity and ecosystem function could provide important motivation for biodiversity conservation… if biodiversity is indeed important for ecosystem function.

In my prior work, I examined this question in the context of bee diversity and plant pollination. I found that, due to diet complementarity among wild bee species, diverse plant communities do indeed depend on a diverse assemblage of bee species. Notably, even locally rare bees can be important pollinators for some plant species (Simpson et al 2022. This work was purely observational, however, and does not consider demographic consequences to plants of pollinator visitation rates or pollinator diversity.

If I were to continue this work, I would more explicitly consider the role of diversity in population or community response to disturbance. While the classic biodiversity-ecosystem function (BEF) approach is compelling for its simplicity, I think it relies too heavily on implicit assumptions. For example, what functions are we measuring and why are some functional outcomes preferred over others? I think the root of the BEF question is, as I stated at the top, about whether biodiversity loss will destabilize ecosystems, creating negative feedbacks that accelerate ecosystem decay. Ultimately, “does biodiversity support biodiversity?” It is about more than functional rates—it is about population and community resilience.

1. I am a Bayesian pragmatist, by which I mean I am not attached to the Bayesian statistics philosophically but, rather, I use it for its strengths. If I am running models that can be fit using a maximum likelihood algorithm, that is the easier choice. The benefit of the Bayesian approach is the flexibility of the MCMC algorithm and the availability of software (e.g., JAGS and Stan) to code your own complex models that better fit your data or biological process. ↩