When Carissa Aoki was completing her PhD at Dartmouth, she developed a model that could predict outbreaks of the southern pine beetle, a pest that is native to the southern United States, but whose range has been creeping north with climate change. A single outbreak of the beetle can destroy thousands of acres of trees, so tracking its range and predicting outbreaks is critical for people working with the forestry service.
Forest destruction caused by the Southern Pine Beetle. PC: Texas A&M Forest Service
Aoki’s model could help address that, but she was struggling to see a way to get her research into the hands of the forest professionals who needed it most.
“Turning scientific research into usable, online tools for real-world stakeholders is no small feat; most scientists don't have either the software engineering skills or the time required to stand up a project like this,” says Aoki, who is a research scientist in the Department of Biological Sciences at Dartmouth and a lecturer at Bates College.
To bridge that gap, Aoki and her PhD advisor, Biological Sciences Professor Matt Ayres, approached DALI Lab. After three years of working together to bring the research to life, the DALI team just launched Pine Beetle Outbreak Prediction, a site that will allow people working in forestry services to respond more efficiently to the threat of southern pine beetles.
Predictions are updated weekly as trapping data are entered, and can be viewed by counties within states, as well as by USFS ranger districts within National Forests.
“With this tool, the U.S. Forest Service will be able to allocate resources and prevent the worst outbreaks, much like they do for wild fires,” says Erica Lobel, program manager for DALI. “Carissa and Matt came to us with the model and we were able to build a platform for it so that it would not only see the light of day, unlike many research projects, but be put into the hands of people who are ready and waiting for it.”
The project was completed in cooperation with the U.S. Forest Service, the Georgia Forestry Commission and other state forestry agencies. But despite having so many people involved, completing the project was no small feat.
“Most of the challenges had to do with getting groups with two different skill sets — the DALI team and research scientists —to understand each other,” Aoki says. “It took us a while to figure out how to speak each others' ‘languages.’”
Sometimes that language difference was literal. Aoki’s model was written using the programming language R, explains Thomas Monfre, D ’21, the development mentor who led the DALI student team.
“We had to figure out how to efficiently run this R script on our server, written in JavaScript, to generate predictions for users,” he says. “All in all, there are a lot of moving parts, so we had to create a flexible system that can adapt to changes and stay working years into the future.”
Since the product is fairly niche, the DALI team had to work with limited user testing, Lobel says.
Trapping data collected since 2011 were used to build the prediction model.
“In all DALI projects, we constantly test designs and the product with users. Sourcing users to test is always challenging, but most of the time we can work close to home,” she says. “In this case, the foresters were our main users, and there was a limited number of them. Our designers had to make the most of every user interview or testing session to really understand them well since they weren’t available to run every design by.”
Despite that, the students were up for the challenge.
“The students' work is reaching all of these cooperators in a very real way, putting the research in their hands in a way that no journal article ever could,” Aoki says.
Although the site is targeted toward forest professionals, it can also be used by the public.
“We would encourage people to play around with the trapping data and prediction systems to see how outbreaks have changed over time,” Monfre says.
Playing with the model can help people better understand statistical modeling, and there’s even an educational component to the site.
“One of our animators made the video explaining how the statistics work,” Lobel says. “We were excited about that because that really leverages each and every one of DALI’s superpowers: design, development, and video work.”
An Insect Invasion Model site is currently in development at DALI.
Monfre, who has worked on this project since his sophomore year, is thrilled to see it live.
“It’s meaningful because it stands to really help understand and hopefully stop the spread of this beetle,” he says. “The goal is that this tool will be used among forest commissions to make decisions as to trapping and prevention efforts, and that it hopefully becomes a model for other sites and tools to understand insects like this.”
That’s already in the works — Aoki and Ayres are currently working with DALI on a similar model that will track invasive insects.
“The DALI model of training students and putting them to work on real projects is a truly innovative way for researchers to get their work out into the world,” Aoki says.
Written by Kelly Burch