Wildlife conservation: Devising statistical tools anyone can use

At the time, her dream job was a consulting statistician position in the College of Forestry. This role worked with students and faculty on a wide range of statistical questions from experimental design to analysis and even teaching.

When the position became available, she applied but the job was offered to Lisa Ganio, current Department of Statistics head. A few months later Huso found out Ganio was taking on a new role, and the consultant position was open again. She applied and remained in the role for 15 years.

Ganio and Huso ended up working together on a number of projects, and even co-taught a graduate statistics course for non-statistics majors.

During that time, a fortuitous encounter with a graduate student in Forest Science propelled Huso down the path of wind power and its effects on wildlife.

“I realized there was a big opening there for somebody with a statistics background to help them understand how to best estimate what the mortality was at wind power facilities using rigorous statistical tools because what they were using before was pretty ad hoc,” she said.

In 2011, Huso published a paper suggesting a statistical answer to mortality questions. Simultaneously, the Corvallis USGS office opened a research statistician position focused on wind power and wildlife and she was offered the job.

The idea for Evidence of Absence came during conversations with U.S. Fish and Wildlife. Huso worked with coauthor Jessica Kenyon (M.S. ’12), another Oregon State graduate, and David Dail, a former statistics Ph.D. student.

“I pointed out that when they go out and search for bats or birds, particularly rare ones, they didn’t have any tools to help them interpret what it meant when they found none,” Huso said. “If you have a high probability of finding something, like 90%, and you don’t find anything, you can say maybe you missed one or two, but you didn’t miss 100.”

The answer gets more difficult with a lower detection probability. If the probability of finding something is 1/10, then it would be easy to miss 10, or even 20. And when you are talking about a rare species those losses matter.

“It’s just so statistically fascinating. There are so many pieces to it and it’s so easy to think about it all the time, to think about where we can improve and what components we have missed.”

The program Huso helped design can tell government agencies and industry the probability of mortality rates based on the probability of detection. This helps them decide if mortality rates are too high for the successful management of any given species.

Her next goal was to develop statistical tools for people trying to estimate impacts, whether that was how many birds are killed at one facility, comparing across different facilities to find patterns in regions or habitat types, or testing methods for reducing mortalities. The result was GenEst.

Huso acknowledges GenEst was a team effort from the start. “I don’t want to take credit for the actual statistics and the coding of this. That was all done by Dan Dalthorp, a statistician with the USGS.” Jeff Mintz, an Oregon State statistics graduate, was also a coauthor on the project, and Lisa Madsen, professor of statistics, collaborated as well.

Beyond software creation, her work with USGS also involves experimental design. Her team takes methods that companies have proposed to reduce mortality rates and sets up experiments to test them.

After spending more than a decade working on wind power and avian species, Huso is still captivated.

“It’s just so statistically fascinating,” she said. “There are so many pieces to it and it’s so easy to think about it all the time, to think about where we can improve and what components we have missed.”