By Jenny Neyman
The Kenai River sonar program tasked with counting king salmon is an evolving science not unlike the Kenai River itself, with twists, turns, snags and murkiness along the way to better clarity. The good news is that, with continued biological research, data analysis and the implementation of improved technology, sonar scientists with the Alaska Department of Fish and Game are confident they now have the best data ever produced on Kenai River king salmon abundance. The bad news, however, is the new-and-improved data shows that chinook abundance is and has been lower than previously thought.
The estimate of late-run kings coming into the river — 9,082 as of July 26 — is so far below the minimum escapement goal of 17,800 fish that fishery managers decided to enact drastic, unprecedented measures as of July 19. Those
measures are closing the river to all sportfishing for kings, banning retention of kings in the dip-net fishery, and closing down the commercial set-net fishery for sockeye along the east side of Cook Inlet to prevent Kenai- and Kasilof-bound kings from getting caught in the gillnets targeting the large run of sockeyes which also is heading into the rivers.
The management decisions spawned from the king-return estimate, shaping up to be the worst return on record, are having a disastrous effect locally, particularly economically — to tackle shops, outfitters and other fishing-related merchants, to sportfishing guides who would be taking clients to fish for kings in July, to other businesses that would get a boost from that tourism, and to the set-netters who have lost their chance at earning their livelihood this summer. With so many repercussions from the shutdown of both sportfishing for kings and east side set netting for sockeye, it’s little wonder people are voicing concerns about the efficacy of the management decisions and the validity of the sonar numbers on which they are based.
But sonar scientists firmly support the accuracy of the counts produced with the new, advanced technology in use at the king sonar site. Just because there has been a change in the king sonar program this year doesn’t mean the run estimate is flawed, said Steve Fleischman, a fishery scientist who analyzes the sonar data.
“I think there’s this impression out there that we’re kind of running by the seat of our pants, when in fact we know far more than we ever did before. We have far more information about what’s really going on out there. And it gets better every day because we learn more and more as we collect more data and as we make comparisons. The unfortunate part is that this is all happening during a downturn in the stock,” Fleischman said. “You could look at both sides of the coin there, it’s a good thing that we’re getting very good information at this point, because now is a very important time to have good information. We don’t want to be making the wrong decisions at this point. We don’t want to be incorrectly liberalizing or allowing the fishery to continue when the runs are very small like this.”
In 1986, a split-beam sonar was installed at mile 8.6 of the Kenai to count king salmon, and was hailed as a marked improvement over the previous, mark-recapture method of estimating run returns. A split-beam sonar transducer emits a low-frequency beam of sound waves into the river and receives echoes bouncing off of fish. That data is transmitted to a computer and appears onscreen as a series of dots representing the echoes, in patterns that biologists call fish traces. Technicians count up the fish traces to determine the number of fish passing by, and use the pattern of echoes to determine whether it’s a larger fish — a king salmon — or something smaller, such as a sockeye.
At first, biologists used a method called target-strength estimate to differentiate kings from other fish. Target strength works by measuring the amplitude, or “strength,” of the echoes bounced off of the fish the sonar sound waves hit. The higher the amplitude, the bigger the fish.
At least, that’s how target strength works when used as split-beam sonar was developed — from
a ship on the surface of the water looking down on fish below. Fish like salmon don’t twist up and down when they swim, like halibut or skates do, so a sonar receiver “looking” at the fish from above can get an accurate reading of fish size based on the amplitude of the target signal returned. But when split-beam is installed in a river the sonar waves hit fish from the side. Salmon change aspect when they swim, undulating closer to and away from the receiver as they pass by. When a fish’s tail swings away from the sonar, it returns a weaker signal, looking like a smaller fish. So a bigger king angling away from the sonar beam could return an echo pattern similar to that of a small sockeye angling toward the beam.
Biologists use other methods to assess the run return, as well. They track how many kings have been caught in commercial sockeye nets in the inlet. They survey sportfishermen in the lower river to determine their catch rate. Fish and Game also runs a test net upstream from the sonar site. They record the catch rate of kings caught in the net. They also create a net-apportioned estimate by factoring the kings caught in the net into the total number of fish detected by the sonar. When the target-strength sonar estimate deviated from the data produced in the other run-strength indices, it was clear the target-strength method wasn’t as accurate as initially thought.
To address that problem, biologists switched to a different method of interpreting the split-beam data, called echo-length standard deviation, to better differentiate kings from sockeyes. ELSD is based on the width of the sonar signal returned and the variabilities between those widths. That’s a much more reliable way to assess fish swimming parallel to the transducer.
At first, ELSD data tracked well with the other indices used to estimate king abundance, and eventually Fish and Game managers switched to using ELSD as its sonar estimator method, rather than target strength.
“We were pretty excited about ELSD. Things seemed very good,” said Debby Burwen, regional Fish and Game sonar biologist.
But ELSD proved to have a weakness, too, and it’s one that couldn’t be fixed because it is inherent in the way split-beam functions. In the mid-1990s, biologists ran some tests on tethered fish and discovered that, while ELSD is better at estimating fish size than the target-strength method was, it is not good at telling close-swimming fish apart, so that, for instance, two sockeye swimming close together “look” to the sonar like one bigger king salmon. In order for ELSD to produce an accurate estimate, it would need to be looking at just kings, not a lot of sockeye mixed in with kings.
“We realized, ‘Wow, we’re really on unstable footing. We’d better hope these sockeye don’t get in the middle of the river, because we know we’ve got problems if they do,’” Burwen said.
Split-beam sonar is a long-range tool and isn’t good at “seeing” short distances. It only sampled the middle of the river, under the assumption that kings mostly swam in the middle of the channel and sockeyes mostly stuck to the sides. This understanding came from direct observations of run returns when the sonar was first installed, as well as the generally accepted wisdom of local fishermen familiar with the river. Even though biologists knew ELSD wasn’t good at distinguishing kings from sockeyes when all jumbled together, they figured the ELSD king counts were still reasonably accurate because there shouldn’t be many sockeye mixed in with the kings in the middle of the river, and there shouldn’t be many kings with the sockeyes along the banks where the sonar wasn’t looking.
But it became clear that that assumption was wrong. Toward the late 1990s, test-netting showed that, for some reason, more sockeyes were swimming in the middle of the channel with the kings. By the mid-2000s, the ELSD number stopped correlating well with the other run-strength indices, and instead was producing an estimate far above the data from the other indices.
“If you get too many sockeye out in the middle of the channel, it’s going to confuse the ELSD estimator because it relies on being able to perfectly discriminate between at least where the fish are in position to each other. It had a weakness, we’ve known it had a weakness, but, at first, it still seemed to be tracking better with the other estimates than target strength, so we felt good about that,” Burwen said. “But one of the main things we realized is we’ve got plenty of sockeye offshore where we prefer they weren’t, and we’ve got chinook salmon near shore, as well, in that zone we had previously excluded.”
Biologists don’t know why the king and sockeye behavior changed, or perhaps whether this was the norm all along and they somehow didn’t realize it. They just knew for sure that sockeye were contaminating the king sample, and that split-beam sonar simply wasn’t able to function accurately under those conditions.
“We decided, ‘We can put all of our energy into trying to figure out why, or we can move on to other, more promising measures,’” Burwen said.
That more-promising measure proved to be dual-frequency identification sonar, which emits
high-frequency beams of sound waves into the river. When it receives echoes bouncing off of fish, DIDSON transmits data to computers that can be displayed as videolike images of fish, similar to the images produced using ultrasound in hospitals. The imagery is so clear biologists can “see” fish swimming by, can more-easily tell fish apart even when jumbled together, can measure the length of a fish in the video by tracing a line over the fish’s image, and can pause, play back and save the images for later analysis.
Burwen first saw DIDSON in 2002. The sonar needs to operate in a high-frequency mode to produce the high-resolution images needed to tell kings from sockeyes in the Kenai, but at that time the best DIDSON model available was only capable of a high-resolution sample range out to 15 meters. The width of the Kenai at the king sonar site is at least 75 meters, and wider depending on the time of year and whether the tide is in. In 2007 manufacturers came out with a DIDSON model capable of high-resolution operation out to about 35 meters. That year Burwen got a grant to test a DIDSON at the king sonar site, alongside the still-operating split-beam sonar. In 2009 she got funding to lease a DIDSON unit, and in 2010 got funding to purchase a second unit, so that 66 meters of the river channel could now be covered — 33 meters with a DIDSON on one bank and 33 meters with a DIDSON on the opposite bank.
The results have been overwhelmingly positive, Burwen said.
“DIDSON is a dream come true. We used to say, ‘Well, gosh, if only we could clear up the Kenai River and see what’s going on out there.’ And now that is essentially what the DIDSON has done,” she said.
With DIDSON, biologists can “see” fish swimming by. Not only is it much easier to measure fish with a DIDSON, technicians also can watch them swim and tell them apart by behavior. Jack kings, for instance, can be about the same size as sockeye, but sockeye beat their tails about twice as fast as kings do. Technicians can see that movement in the DIDSON imagery and make the correct classification.
They can see lots of other stuff, as well, such as the occasional flounder wandering up from the river mouth, and seals heading upstream to hunt salmon. They’ve got an archived video of a seal pup and its mother swimming by, and another showing a seal lying in wait, then diving out to snag salmon swimming by.
Biologists worked to fine-tune the DIDSON as it ran alongside the split-beam, to great success, Burwen said.
“The other indices are tracking very well. It’s really a nice relationship. They’re all tracking just as you’d want. And then we’re consistent with what the rest of the state is experiencing with king salmon (king rivers around the state are experiencing low run returns). So there are no indicators that what we’re measuring is anything other than as good as you can get,” she said.
Meanwhile, the ELSD estimate was getting worse and worse, farther and farther off from the rest of the indices. In 2011, the ELSD estimate got to be thousands over the DIDSON and the other indices.
Funding and resources not being infinite, Fish and Game decided to scrap split-beam for the 2012 season and run DIDSON as its sole sonar counter.
“The split-beam counts did not seem to be stable in the sense that they were providing consistent information from year to year. They were affected by fish behavior and what we found was it was easy to overestimate king salmon with the split-beam numbers, particularly when there are a lot of sockeye in the river relative to king salmon, so we didn’t see a lot of benefit in hanging on to the old split-beam scenario,” Fleischman said. “What we concluded, and I don’t think there’s any doubt about this, is that we have a lot of information with which to assess king salmon with our existing indices plus the DIDSON sonar. The old split-beam numbers would have just muddied the waters. At that point we just decided it was time to move to the future instead of trying to fix the past.”
The change has prompted concern among some in the sportfishing and commercial-fishing community. After all, DIDSON is still relatively new at the sonar site, compared, at least, to split-beam. What if something goes wrong?
Burwen acknowledges that DIDSON is not without its challenges. One problem is the sonar site, at mile 8.6, is within the tidal influence zone of the river. Water floods in and out twice a day, sometimes sweeping along trees and other debris that can knock over the sonar equipment, causing temporary lapses in sampling. And when the tide is in water floods behind the DIDSON transducers, high enough to where fish could be swimming along the bank behind the transducers.
DIDSON also doesn’t sample as frequently as split-beam did, because it doesn’t measure as large a range at a time. In order to get high-resolution images, DIDSON can only sample in 5- to 10-meter increments. So the DIDSONs on each bank cycle through four ranges of the river in 10-minute increments. Both sample a near-shore range, then 8 to 12 meters out, then 12 to 23 meters out, then 22 to 33 meters out, then repeat the 22- to 33-meter range, then cycle back nearer to shore. Split-beam used to sample in 20-minute increments, but with DIDSON units costing $100,000 a pop, plus $20,000 for the equipment to move and aim the units, Burwen said it would be difficult to come up with more units to increase the sampling time. But as it is, the 10-minute samples of 10-meter ranges are considered to be perfectly acceptable.
“In the world of funding not being unlimited you have to look at that. Everybody subsamples. If you talk to the statisticians, they will tell you that that is very adequate,” she said.
And the evolution continues. The sonar team has been looking for a sonar site upriver to get out
of the intertidal zone. They’ve found a promising spot at about river mile 14, just downstream from the Kenai Riverbend Resort off Ciechanski Road and have installed new, even-better DIDSON units for testing. The new DIDSON can sample up to 35 meters with high resolution, and the river at mile 14 is about 75 meters wide. There is a side channel on one bank that’s currently outside the DIDSON’s range, but it’s unlikely kings are passing through it, and Burwen said they might look at getting a third DIDSON to keep an eye on the channel anyway.
The new site is providing promising results. Burwen went out Friday to check the data, and immediately saw two kings on the screen. She also saw a blank patch on one of the screens, indicating a depression in the river bottom, with a king hanging still just above it.
“It’s a big redd area,” Burwen said, pointing to the depression.
The sonar had been re-aimed just that morning, and the new angle included a king salmon spawning bed. Which is interesting, but Burwen wants to sonar to see fish passage, not stationary fish spawning, so she adjusted the angle from the computer onshore. It’s a process of finding the best angles to use at the new site.
“We’ve been moving it around — how’s it look here, how’s it look there?” she said.
Most promising of all is the new sonar is more powerful, so it could be set to sample 20-meter ranges at a time, rather than 10.
“It’s got more power and flexibility of settings, and it can cover more than one range at once. That will allow us to sample more of the time. That’s huge,” she said.
Depending on funding and available personnel, the new site could be up and running next year. Ideally the new site would run concurrently with the current sonar site at mile 8.6 in order to compare numbers to each other and to the other run-strength indices, she said, but she doesn’t know yet if or when that might happen.
In the meantime, the sonar team will continue tweaking the new DIDSON site and monitoring the current one, which still is producing the best, most-accurate sonar fish counts the department has had in years, she said.
“In 2010 we felt like we had pretty good side-by-side estimates, in 2011 they were very good and now in 2012 we’re still doing well,” she said.
“It’s just self-evident from looking at the data that we’re getting good information, that DIDSON can tell large fish from small fish very well, and from doing that we’re able to get a good estimate of the total number of king salmon,” Fleischman added.
Estimates to escapement
Even though split-beam ELSD sonar was found to be overestimating the number of kings, due to the sample being contaminated with sockeyes, and DIDSON has proved to be a more “pure” estimate, more accurately counting just kings, there has been debate about the decision to shut down two fisheries based on sonar numbers that don’t relate to the king salmon escapement goals the way the old system of measurement did.
The current escapement goals in the king salmon management plan were set when split-beam sonar was in use. At the time the goals were set, the split-beam sonar estimates were likely biased high, due to sockeye contamination of the king samples, but managers didn’t know it at the time.
“The escapement goal was based on these old split-beam counts, which for a number of years seemed to work out fairly well but as we learned more and more about them we found that they were not doing what we hoped they would do,” Fleischman said. “We have in place a management plan that is based on this old system of numbers. It contains information about sockeye salmon within it because it’s contaminated by sockeye counts to some extent in any given year.”
The DIDSON numbers do not appear to be contaminated by the presence of sockeye. That’s good, in the sense that Fish and Game is getting a more accurate and true count of kings. But it’s challenging in the sense that DIDSON numbers don’t currently directly relate to the escapement goals in the management plan, because they were set taking into consideration split-beam data that was estimating too high. So managers aren’t currently using DIDSON data to make their decisions, Fleischman said. Rather, they’re looking at all the run-strength indices and using DIDSON data to corroborate how accurate the other indices are.
Going forward, the department will need to set new escapement goals, but that’s going to take looking back at past years’ split-beam numbers, comparing that with the other run-strength indices and trying to come up with a more-accurate estimate of how many kings were returning to the river in the past.
“We’re going to use all the information that we have and start from scratch. To set these goals we need to reconstruct the historical abundance through time and that’s what gives us information that allows us to determine what a sustainable goal would be. There will be a new goal, but it won’t be a conversion of the old, it will be an analysis of the data as best we can reconstruct them to provide that escapement goal that provides for sustainable fisheries,” Fleischman said. “We’ve always looked at all the sources of information we’ve had available to us, that’s just the nature of having to manage a fishery is that you don’t want to ignore any information that’s available. We will continue to do that even after we have a DIDSON-based goal. We will still be looking at these other sources of information, and we will be looking for whether or not things are consistent with one another.
“By having DIDSON available to us, we are now able to confirm that these other sources that we’ve looked at for years and years and years, really are good. They really do what we thought they were doing. It was the old sonar stuff that was leading us astray. So, in fact, this year and last where we’ve had the DIDSON counts available to us plus these index data, we know far more about the true abundance of king salmon than we ever have in the past,” he said.
To the sonar scientists, like Burwen and Fleischman, the Kenai king run strength estimate data is more accurate than it has even been in the past. But with two fisheries shut down, and particularly with the Board of Fish in an emergency session last week voting to not take up the emergency petitions submitted by set-netters hoping to be allowed to fish, the debate over the levels at which escapements are set and the policies set and enacted in the name of protecting those escapements, are still matters of contentious debate.