New model can accurately predict a forest’s future
Using only genomics and a one-time tree count, model accurately predicts tree species fluctuations
University of Illinois at Urbana-Champaign, News Bureau
image:
The Wind River Forest Dynamics plot in southern Washington state is one of dozens of study sites organized under the Smithsonian Forest Global Earth Observatory.
view moreCredit: Photo by James A. Lutz
CHAMPAIGN, Ill. — One of the great challenges of ecology is to understand the factors that maintain, or undermine, diversity in ecosystems, researchers write in a new report in the journal Science. The researchers detail their development of a new model that — using a tree census and genomic data collected from multiple species in a forest — can predict future fluctuations in the relative abundance of those species.
University of Illinois Urbana-Champaign plant biology professor James O’Dwyer led the new research with Andy Jones, a professor of botany and plant pathology at Oregon State University, and James Lutz, a professor of forest ecology at Utah State University.
O’Dwyer has spent most of his career studying the factors that drive ecological change and using that knowledge to build models that reliably predict how individual species and forest communities will change over time.
“This work is crucial because changes in abundance or loss of a species from a forest can have cascading effects on other species,” O’Dwyer said. Forests with lower tree diversity are more susceptible to attack by pathogens or pests. Being able to predict which species are at risk could help understand how forests will change in the future.
“Species diversity is lower in forests of the western United States than in other parts of the U.S., but most species have unique roles in the forest,” said Lutz, who, since 2010 has conducted an annual census of the Wind River Forest Dynamics plot in southern Washington state, the focus of the new study. “Losing one species, when there are few to begin with, could result in a less productive forest and potentially one that doesn’t support as many small plants or animals.”
But predicting future changes in species abundance is a formidable task, the researchers said.
“In a forest, there are constantly varying environmental conditions, as well as different tree neighborhoods, with species competing for resources like sunlight and water,” Lutz said. “Neighboring trees influence each other while living and after death, as snags and wood, all amidst variation in rain and soil conditions.”
Simply collecting the data needed to determine which elements are most influential requires years of painstaking work. Luckily for the researchers, many long-term studies of forests are well underway, led by teams of scientists all over the world. Some of these efforts are organized under the Smithsonian Forest Global Earth Observatory, which has amassed data from 78 sites across the world spanning several decades. The Wind River Forest Dynamics Plot is one of those sites.
The new study uses data from that plot to build on previous modeling efforts.
In a 2023 study published in the journal Nature, O’Dwyer and graduate student Kenneth Jops developed a model for predicting whether two or more species will continue to coexist in a shared environment. That approach focused on the life history traits of each species, which primarily consist of timelines of how fast each organism grows, reproduces and dies. From a matrix built from this data, along with a census of trees in the forest, the model calculated each species’ “effective population size.” Two species with the same or similar effective population sizes were more likely to continue to coexist, the researchers found.
“The upshot of that study is that we identified certain combinations of life histories across plant communities that act to maintain diversity over longer timescales, while other combinations would lead to lower diversity,” O’Dwyer said.
In a more recent study of a tropical forest in Panama, another ForestGEO site, O’Dwyer and his colleagues extended the approach to multispecies systems, finding that the effective population size was helpful in predicting short-term population fluctuations.
“We were able to infer those life history differences because we had access to data from one of the most well-studied forest plots in the world,” O’Dwyer said. “With anything less than the decades of data from that plot, our estimates would have been much less reliable.”
But collecting decades of data from a single forest is not feasible for most studies, and the team sought to find a more streamlined approach. Instead of collecting life history data for dozens of species over decades, Jones led an effort to collect genomic data from about 100 individuals of each of eight species of trees that make up roughly 90% of the stems and almost all the biomass in the Wind River plot. These were not complete genomes, but a sampling of genes that, the researchers hoped, would reflect key events in the life history of each tree species.
“Effective population size is a fundamental concept in evolutionary biology, first described almost 100 years ago. Although the true nature of the factors that ultimately determine effective population size is complex, it is perhaps easiest to think of it as the number of individuals that contribute offspring, and therefore their genes, to the next generation,” Jones said. “The effective population size is typically lower — sometimes much lower — than the number of trees of a species that we can count in a forest. This is because some individual trees leave more offspring than others, which is how populations evolve. When this occurs, we find an increase in nonrandom associations between genes.”
“That balance between random and nonrandom associations in the genome is closely related to effective population size,” O’Dwyer said. “Those life history traits are in the background, shaping that genomic data. I would say the genome is like a hidden recording device of the history of that species in that forest.”
The researchers incorporated the genomic data into their model, which also included data from a census of all the trees over 1 centimeter in diameter in the Wind River plot in 2011. The model accurately predicted fluctuations in the abundance of the eight species in 2016 and 2021, outperforming other models.
“The predictions were highly correlated with the observed fluctuations in abundance,” O’Dwyer said. “That’s very exciting.”
“My sense is that the population genomic variation that we’re looking at is an underused resource,” O’Dwyer said. “It’s carrying a lot of information about the history of that species.”
The researchers hope to continue to refine their model and determine whether and how it can be used in forests that are less well-studied than the ForestGEO plot.
“If we can further distill the relationship between genomic variation, census data and ecological dynamics, this could allow us to build predictive models, with consequences for conservation and management across a broad range of ecosystems,” O’Dwyer said.
The National Science Foundation and the Simons Foundation supported this research.
Editor’s note:
To reach James O’Dwyer, email jodwyer@illinois.edu.
The paper “Genomic demography predicts community dynamics in a temperate montane forest” is available online or from scipak@aaas.org.
Journal
Science
Method of Research
Data/statistical analysis
Subject of Research
Not applicable
Article Title
Genomic demography predicts community dynamics in a temperate montane forest
Article Publication Date
18-Sep-2025
