By Lex Treinen

The Chilkat Valley News

Under steady wind, light rain and near-freezing temperatures on a recent weekday, biologist Stacie Evans is staring through a spotting scope at a pullout on the banks of the Chilkat River. Silently, she counts the number of bald eagles. 

244…245…246. That’s her final count of adult birds at the Council Grounds location, plus another 85 juveniles, which are recognizable by their lack of distinctive white heads and tails. 

The day’s count is more than the entire population estimate from 2020, the year Evans took over the count, and the year of the lowest eagle return since the annual bald eagle count began in the Chilkat Valley in the late 1980s. 

In all, at 12 designated sites around the Alaska Chilkat Bald Eagle Preserve, Evans counted 666 distinct bald eagles in one day. That’s pretty good, said Evans, but she stopped short of calling the census healthy. 

“I’d call it ‘reasonable’” she said. 

Despite battling a recent COVID infection, Evans has been driving the route once a week to make the count and keep the data set strong. She said it’s an important way to keep track of the health of the ecosystem and to track how climate change is affecting eagle distribution. 

“Eagles are an indicator species — all birds are,” she said. “The bald eagle preserve is a good opportunity to take a look at what’s going on. The more eagles there are the more salmon there are.”

This year, there is no shortage of fish. The Alaska Department of Fish and Game estimated the chum return was more than 10 times higher than the target escapement it sets for the run. 

But the eagle count, which normally peaks around the second week of November, is still nowhere near counts from the 1980s and 1990s, when they regularly exceeded 1,500. Evans said there aren’t necessarily fewer eagles. Climate change —  combined with this year’s El Nino weather pattern — can affect the way the eagle population is distributed in interesting ways. Evans thinks this year, the high salmon abundance was mitigated by relatively high water levels from an unusually wet fall. 

“High water makes it harder for them to get salmon. It’s the same as for anglers,” said Evans. “It’s harder for them to see the fish and there’s more places the fish can be.”

Plus, warmer than average temperatures mean few places along the river are frozen so eagles have more room in the watershed to look for food. Chilkat River chum salmon evolved to take advantage of spring-fed water that keeps the river open late into the season compared to other rivers in the region.  

An exceptionally low count in 2020 was attributed to a combination of factors, including warm temperatures around Haines that kept rivers open in a wide area. Plus, fish runs were exceptionally low around that time, which biologists suspect was linked to an earlier warm water patch in the salmon feeding ground in the Gulf of Alaska known as the Blob. 

While there still may be similar total numbers of eagles in the valley this year, they could be miles away from the main road, farther up the highway or in the Tsirku drainage. 

“The salmon, the temperature, the eagles are all interconnected,” said Evans. 

Biologists are dreaming up ways to improve population estimates to account for different population distribution in the changing climate. In the 1980s, The U.S. Fish and Wildlife Service used aerial surveys to count eagles, and Evans said the idea could be adapted with modern drone technology that could be used to photograph harder to access areas. Biologists could then count the eagles in the photos. 

Another idea is to install sound recorders at different sites and use different eagle calls to come up with a population estimate. Maia Edwards, a researcher and assistant director of the bald eagle foundation, is currently designing a study on the technique. 

Edwards said the fledgling technique of acoustic surveys relies on machine learning, which can be programmed to recognize eagle calls. Several recorders could be installed in trees in certain areas and individual eagles could be identified based on fractional variations of when the sound arrives. The technique offers the promise of more accurate surveys, since they could be installed in areas away from roadside pullouts.

Still, she said, it won’t be fully established any time soon. 

“It’s a technique that is still being tried and perfected,” said Edwards. 

In the meantime, Evans will be driving the road each week to track the eagle trends. While so far, they haven’t reached the levels from decades ago, there’s still a chance things could change. 

Plan is to try a couple different methods, it’s a technique that is still being tried and perfected by people doing bioacoustics

One method is to listen and find number of calls of an eagle in a certain time period, by knowing behavior and how often an eagle makes a sound, you know how many there are

There is a more exact way “acoustic localization”

Basically with first few recorders, I’ll be collecting data of random eagle calls from as many different eagles as I can

Software can select all of their calls, probably a few hundred eagle calls, build machine learning model that can recognize eagle calls in all our other data, it can be as specific as you can

?Can be individual variation, each has unique voice print, I could have machine learn to identify just eagle calls, or individual 

Raven Pro More accurate 

Then we can determine three or four specific sites, we’ll have three or four different recorders, they’re fairly close to each other use sound triangulation, algorithm uses data that reaches each monitor, using time difference