Mapping Connectivity: How UVM Researchers are Studying Safe Passages for Wildlife
Written by Kristen Munson and published by the University of Vermont, November 11, 2024. Original link here.
The Stowe Land Trust asked Jed Murdoch, a wildlife biologist at UVM, to identify species on a parcel of land the nonprofit aims to conserve to protect connectivity for wildlife movement.
Jed Murdoch’s class huddles in a parking lot in Stowe one bone-chilling Friday afternoon in March. A skunky stench wafts from the back of a University of Vermont passenger van—lure for the wildlife cameras the students will install along snow-covered woods.
“It has some staying power,” says Murdoch, a wildlife biologist, with a laugh.
The class is working with the Stowe Land Trust to model wildlife movement on a stretch of land near Route 108 that SLT is in the process of protecting. They break into teams to hike to 25 locations across the 85-acre parcel on the eastern side of the Green Mountains that could soon become a link for recreational trails between the town’s popular recreation path and the conserved lands on the side of Mt. Mansfield in an area that is also prime wildlife habitat.
Murdoch explains the goals for the class project: to conserve connectivity for wildlife by siting a new multi-use recreational connector trail on the parcel in a way that reduces impacts to wildlife movement. “So there is some real use to those maps that we are going to generate,” he says.
That is why Carolyn Loeb, MS ‘19, stewardship director for SLT, enlisted Murdoch’s help in the first place. She took his course years ago and knew that UVM students could fill information gaps about the site to help SLT adequately protect it. The parcel, a thin stretch of forest nestled along the Little River, could someday be developed unless the nonprofit can conserve it. They have until December 31 to secure funding for the Adams Camp Connector project.
“This work really matters,” Loeb says. “For a small piece of land, it has a pretty big impact.”
“Humans like to put numbers on things. Showing that electrical movement across the landscape is giving us a way to quantify connectivity." - Caitlin Drasher
She holds up a map to orient the students about where they are standing and why this particular patch of forest is worth saving. She points to the Green Mountains running north-south along the state and explains how they merge with massive swaths of land elsewhere in the Northeast and Canada.
“The Green Mountains of Vermont are an important regional linkage,” Loeb says. “So, wildlife, but also plants under climate change, are going to be trying to use intact forested pathways, especially along our mountain ridges, to move as their habitat changes. ... And what happens along the Green Mountains matters not only for Vermont but for the whole Northeast, because if we impair this linkage, it would be like constricting human movement on [Interstate] 89. So, our regional context matters a lot. This is not a big parcel, but it plays an important role in buffering that mountain range for species movement from forest fragmentation.”
Students tramp up the road to place the cameras on trees lining the steep hillsides above the river and across the forest. A band of three undergraduates slides down a snowbank to look for a suitable tree to smear with lure.
A student pulls on a pair of latex gloves and hesitantly swipes the bait along the bark of a tree with a Q-tip.
“I’d give it a little more,” Murdoch gently advises.
The student swabs another thick sticky dose of lure, like stinky jam, along the bark. In three weeks, the class will return to collect the memory cards for analysis and sorting of the imagery aided by machine learning.
“A lot of ecology is really quantitative,” Murdoch says. “It is so absurdly technical. If you want to draw inference about the messy natural world, you have to defend that and it’s not easy.”
Camera traps can confirm that an animal visited a site with a picture. But some animals, like coyotes, excel at detecting when something in an environment seems off, and they avoid it.
There is a dual problem of not knowing whether an animal is truly absent or actually there, and merely undetected, Murdoch explains. “It creates some statistical maneuvering and that’s what we focus on.”
Loeb has seen deer and fox tracks and the prints of smaller mammals like voles and mice on the site. But she doesn’t know how they move across the parcel and doesn’t have a team to study the question and create a scientific report to incorporate into trail placement decision making.
Images from the camera traps Murdoch’s students placed to help the Stowe Land Trust identify animals visiting a parcel the organization aims to protect.
“We don’t have the staff capacity to get out to do camera trapping work let alone the modeling which is why UVM is an amazing partner for this,” she tells the students on the walk back to the parking lot. “I am really excited to see what you find. … Without you, we just wouldn’t be able to do this. It would just be me alone speculating about where the wildlife are moving instead of you giving us hard data.”
Awash in data
Jed Murdoch, associate dean for academic affairs in the Rubenstein School and fellow at the Gund Institute for Environment, has spent decades studying how animals, often large mammals like wolves and snow leopards, live and move across landscapes. When he was an undergraduate, data was a problem—there wasn’t enough of it.
“We always had a problem of not enough data,” Murdoch says. “Now we have too much.”
In the past, equipment to collect data on wildlife was limited, making it challenging to gather records on where species occur, especially species that often live in low densities and have large home ranges. Wildlife biologists now use GPS collars to track animals that can gather tens of thousands of data points and also infrared camera traps—like those used in the class—to reliably collect animal locations 24/7. Murdoch’s students learn traditional camera trapping methods and modern data analysis techniques that incorporate AI to help identify animals in the thousands of photos collected and test questions about animal movement patterns across complex landscapes. Murdoch wants budding wildlife biologists to learn how to find a pattern in the data.
We are in this important zone, where there is this need for more data about how animals move across the landscape but gaps in the technology make it hard to do so, Murdoch explains. “So we have to look for other ways of doing that.”
One idea—electrical circuit theory—used in conservation ecology emerged from the field of electrical engineering. It models how electricity moves to connect nodes—essentially mapping how it moves through a circuit of varying resistances—and applies the same principles to animal movement. For instance, as moose navigate across an area, they select habitats to move through based on how conducive they are to movement and the quality of that habitat, potentially moving away from housing developments to avoid detection or follow streambeds. Modeling animal movement involves integrating data on species occurrence and habitat resistance to generate maps showing the flow of animals across the landscape.
Flow maps like this can help identify places where movement is expected to occur and pinch points of movement in the landscape that represent a potential conservation concern, Murdoch explains.
They can also identify opportunities for interventions that can help.
Prioritizing safe passages
For instance, drive most stretches of I-89, and you will probably spot a pulpy mess of fur and bone—a casualty of the more than 25,000 kilometers of Vermont roadway bisecting wildlife habitats and the human desire to move at speed. While we can’t easily unbuild the American highway system, we can unwind some of the damage through the targeted placement of infrastructure that protects wildlife and safely facilitates their movement. Big interventions such as wildlife overpasses can be enormously expensive, and projects specifically aimed at protecting wildlife don’t often get bumped to the top of transportation spending lists. That is, unless you can build wildlife protection into your evaluation of transportation needs.
In 2022, Murdoch and Ph.D. student Caitlin Drasher ’17 used circuit theory to map landscape connectivity for eight terrestrial species, including bear, bobcat, moose, deer, fox, coyote, skunk, and raccoon, and create the Terrestrial Passage Screening Tool for the Vermont Agency of Transportation (VTrans) in partnership with the Vermont Fish and Wildlife Department. The tool allows VTrans staff to assess the connectivity value of transportation infrastructure such as culverts, bridges, and underpasses that support targeted efforts to facilitate wildlife movement across the road network.
“If you think about it, [Highway 89] is cutting right across the spine of the Green Mountains so that's where a lot of my models were showing all of the electricity just moving up and down the Greens,” explains Drasher, the black bear biologist for the Maine Department of Inland Fisheries & Wildlife.
Her analysis used a software program called Omniscape which combines topographic layers with occupancy data of where these eight species live to reflect how animals move across Vermont. When Drasher ran the model, she found it confirmed a major hotspot VTrans had flagged as a site of interest for protecting wildlife movement.
Last December, the agency received a $1.6 million grant from the U.S. Department of Transportation’s Federal Highway Administration to design an underpass beneath I-89 and Route 2 in Waterbury that connects two important sections of forest and riparian habitat.
“All of my species models identified that location and it just lights up with electricity,” Drasher says. “It's one of the best locations in the state to be doing that so my work kind of confirmed the suspicions that they had.”
Drasher’s doctoral work builds on this effort. She and Murdoch had a grant from the Northeastern States Research Cooperative to develop a regional map for Maine, New Hampshire, and Vermont that adds another dimension to connectivity: population genetics. By partnering with state and federal agencies in each state to collect samples from hunted and trapped species, the researchers can see how 10 terrestrial species (the same eight from the VTrans project plus marten and fisher) in different areas relate to each other and how genes move across a landscape.
These maps could serve as a tool for land managers and town planners when trying to understand how potential development projects like housing construction may affect wildlife. This can remove some of the uncertainty people face when trying to balance the need for human housing with wildlife concerns, Murdoch explains.
Vermont, like many states, is facing severe housing shortages. By identifying critical wildlife corridors, it might help move stalled projects forward and protect land that has the most conservation value for wildlife, including important species like deer, moose, and bear. Drasher is running connectivity models on UVM’s supercomputer this fall.
“Humans like to put numbers on things,” she says. “Showing that electrical movement across the landscape is giving us a way to quantify connectivity … so it helps us put a number on things which can be useful for land use planning, species management or forest management.”