News

UChicago ecologist tracks the effects of climate change on the spread of viruses

Greg Dwyer has spent decades studying invasive species to uncover the relationship between defoliation, insect populations, and climate change.

For ecologist Greg Dwyer, PhD, studying natural systems has been part of his life since childhood.

“I grew up way out in the country. So, I was sort of a country kid— learning all about the birds and the bugs and plants and the flowers.”

Dwyer attended Cornell University as an undergraduate where he worked in the laboratory of Simon Levin, PhD. Dwyer’s research project focused on myxomatosis, a viral disease that is intentionally used to control wild European rabbit populations in the farmlands of Australia, France, and other countries.

At that point, myxomatosis had been successfully used for over thirty years, although its effectiveness waned. This was thought to arise from a cycle of natural selection: as the rabbits with specific immune defenses would survive and reproduce, the virus would evolve, driving the rabbits to also adapt, and so on. Dwyer set out to predict if— and when— a balance between these cycles could be achieved. 

In his 1990 publication, Dwyer presented a set of computer simulations that modeled the viruses’ infectiousness as a function of time and considered the efficiency of viral transmission through other hosts, such as fleas. These models were much more intuitive than had been previously attempted and provided new tools to predict outcomes of a “co-evolutionary arms race,” a phenomenon he interrogated less than a decade later. These and similar models were part of the foundation for the mathematical theory of pathogen evolution established in the 1980’s and 90’s. Myxomatosis remains the textbook example of a host-pathogen arms race to this day.

Dwyer later pursued his PhD at the University of Washington, working with Peter Kareiva, PhD. Turning to insect pathogens, Dwyer designed models to predict how a viral disease spreads through populations of the Douglas fir tussock moth. He used features such as population density, age, and spatial structure – how the insects were positioned in relation to each other – to forecast virus transmission. If these sound familiar, it’s because they are the same parameters used to calculate the spread of human diseases such as COVID-19.

These formative experiments, remarked Dwyer, would motivate him for the rest of his career. 

“And so all that was inspired by the computer model with the rabbit disease,” he said. 

Forest defoliation is caused by the invasive spongy moth

Forests are crucial for carbon cycling and storage, regulation of temperature, and providing habitat for a diversity of wildlife and plants. Short-term events such as wildfires and logging cause flux in these essential functions. However, long-term events, including defoliation— the removal of leaves from trees— can have devastating consequences on the maintenance of this critical resource. 

Defoliation of forests can be driven by insects. When mass outbreaks of infestations occur, trees are unable to perform carbon cycling, often leading to the death of thousands of trees. Not only does this remove a recreational resource and halt timber production, it also transfers large amounts of carbon to the atmosphere via decomposition, which is a contributor to climate change. In fact, warming temperatures can create ideal conditions for insects, exacerbating the cyclical nature of the problem even more.

The spongy moth—an endearing name for the insect Dwyer studied as a postdoc, and formerly known as the “gypsy moth”— is one of these organisms. It’s an invasive species that kills trees after feeding on its leaves. Dwyer and his students have spent years developing ways to model the delicate relationship between forests, the spongy moth, and its natural pathogens, including nucleo polyhedrovirus (NPV). Similar to myxomatosis, NPV is intentionally used to control spongy moth epidemics.

Outbreaks of the spongy moth may occur due to a combination of climate and biological influences. For instance, moth populations boom when environmental factors cause their predators (mice, squirrels) to decrease. In turn, high moth densities allow the NPV to wreak havoc, leading to a population drop. A 2004 publication, one of Dwyer’s first at the University of Chicago, was the first to fold all these factors (the spongy moth, its predators, and its pathogen) into one elegant model. These results marked a breakthrough in understanding the cyclical patterns of spongy moth densities.

While predators and parasites hold a large sway over spongy moth outbreaks, another member of the forest also has a stake: the tree itself.

Cyclical outbreaks of the spongy moth vary from forest to forest, but seem to be more severe when an area contains a large amount of oak trees. In a 2013 article, first author Bret Elderd, PhD and other Dwyer lab members investigated how an innate defense mechanism of oak trees influences outbreaks of the spongy moth. Using a combination of field work and computational techniques, Elderd found that hydrolyzable tannins, a mildly toxic molecule secreted by the tree when it is defoliated, reduce variability among spongy moth larvae in their susceptibility to a baculovirus infection. The author’s success in reproducing this complex interaction addressed a long-overlooked factor in spongy moth epidemiological models.  

Nevertheless, as global temperatures rise, even the most sensitive models may become inadequate. This was made evident in a 2020 publication in which former Dwyer lab member, Colin Kyle, PhD, designed models that explored how host density and weather conditions influence the outbreaks of a spongy moth pathogen, Entomophaga maimaiga. The best-fitting model revealed that a combination of weather and moth density drives the host-pathogen dynamics, predicting epidemics better than those that didn’t account for fluctuating temperatures.

While these publications are only highlights of the lab’s extensive portfolio, they underscore the critical need for methods to curb defoliation, especially in wake of climate change.

“The thing we learned with the study of the spongy moth is that modest changes in temperature and humidity can have very big changes on interactions between species. I didn't see that coming. And that's a very important conclusion.”

Forecasting defoliation is no walk in the park

Uncovering the relationship between defoliation, insect populations, and climate change is not without obstacles. One of the biggest challenges, says Dwyer, is defining the variables, such as temperature, insect age, and virus lethality, that shape a model’s predictive power. 

“These models have many parameters,” said Dwyer. “For each parameter value, you have to run the model maybe 100 times because the model is stochastic. Once you run it 100 times, you take some kind of an average of the goodness of fit. You need to do this for maybe 10^6 and 10^7 parameter sets.”

After refining the list of inputs, one must find a way to mathematically simulate real-world scenarios. This poses challenges when the data are longitudinal and ever-changing. Researchers working with this type of data often use statistical methods such as Markov Chain Monte Carlo, but even then, members of the Dwyer lab often find it more useful to develop their own algorithms. One example is line-search MCMC, a tool used to make inferences about pathogen growth inside the spongy moth.

However, even the most reliable models are of little use without good quality input data. To come up with the data, the Dwyer lab chooses to take matters into their own hands. 

“We often collect our own data. In the case of the tussock moth, it means driving out to Washington State to do experiments,” said Dwyer. We collect insects from naturally occurring populations and bring them back to the lab and see who's infected, and what strain they're infected with. We put them together on a branch, and we see how and what fraction gets the virus. So it's making little epidemics on branches.”

The unique interests and skillset of the Dwyer laboratory have poised them for several collaborations, ranging from climate scientists at Argonne National Laboratory to researchers investigating the transmission of viruses in humans, such as HPV. Dwyer has even worked with researchers at the University of Chicago’s Harris School of Public Policy studying how spongy moth outbreaks affect housing values. Most notably, the group has an ongoing collaboration with the US Forest Service, providing guidance for how and where to spray the tussock moth NPV. This type of work, said Dwyer, spreads their research mission even further.

“Some private landowners in Nevada would like us to project whether their trees are going to be protected by the virus. And now we're branching out to the southwest. That's a way that the Forest Service got us very involved in the day to day management of the forest,” he said.

Looking forward

The repercussions of climate change continue to motivate Dwyer. Armed with hearty amounts of time-series data collected across Washington state, Idaho, and Oregon— all provided by the Forest Service— the lab is next tackling the question of how to forecast tussock moth epidemics across different areas and populations. This meticulous conservation research, said Dwyer, is necessary to keep up with a changing planet. 

“Climate change will likely reduce the ability to disease control the insect,” he said. “And that's, of course, a huge problem.”

For now, the lab will continue doing what they do best: rotating between the field, the desk, and the lab, fitting epidemic models to real-world data. After nearly 30 years spent designing ecological models, tracking viral transmission, and predicting spongy moth epidemics, Dwyer reflected that some of the most important lessons aren’t learned at the bench. 

“The first rule of the lab is to never leave anything on the roof of a vehicle.” 

Research reported in this profile article was supported by the National Science Foundation, the National Institutes of Health, and the US Forest Service.

Explore the Biological Sciences Division