As any of our regular readers will know, 2016 Coho season in Puget Sound was anything but typical. A dire early-season Coho forecast, collapses during season setting negotiations, and eventually a surprisingly strong Coho run after all. We wanted to analyze the resulting Tribal and Sport fisheries, and on October 21st, 2016 we made a public records request for copies of the Puget Sound Tribes fish tickets. Four months later (Feb 24), after a number of twists and turns, we received the data and have begun our analysis.
One of the first fisheries we wanted to study was the Lake Washington Coho run. This run offers a unique baseline in Puget Sound — annual fish counts conducted at the Ballard Locks. Combined with other data sources like the sportfishing creel counts, Issaquah Hatchery counts, and even buoy water temperature data, we set out to reconstruct what happened in the 2016 fishery.
What we found simply doesn’t add up… In the end, after much analysis, we find a preponderance of the evidence indicates that the Tribal catch reports, specifically the Muckleshoot Tribe reports, are questionable. Not only did an unexpectedly large number of hatchery fish disappear en-route to the Issaquah Hatchery, but the patterns exhibited in the submitted catch reports are inconsistent with the known run size and shape.
Let’s dive into the details…
Ship Canal Fisheries
Due to closures in the Straits and Sound last year, the first significant fisheries that Coho returning to the Lake Washington system encountered was when they rounded Meadow Point and headed into Salmon Bay. The first fishery — the Suquamish gillnet fishery — takes place in the ½ mile stretch between Ray’s Boathouse and the Ballard Locks. The second fishery — the Muckleshoot gillnet fishery — takes place in the 4 ½ mile stretch between the locks and Ivar’s Salmon House on Lake Union.
Our first step was to compare the reported catches between these neighboring fisheries to see if they correlated. These fisheries occur sequentially in the Ship Canal, within a golf shot of one another, using the same gear, so we expected similarities. What we found for September is summarized in this chart:
The Suquamish catch reports (in Red) resemble the overall shape of the counts taken at the locks, and begin to soften as September ends. Aside from data missing on Saturdays (17th and 24th of Sept) we find the Suquamish data to be credible. On the other hand, the Muckleshoot fish ticket data (in Blue) falls short of credibility by comparison to the Suquamish catch, or when contrasted with the significant numbers of fish counted at the locks in mid-late September. The Suquamish reported 2601 fish caught for September, the Muckleshoots just 306. Again, the Muckleshoot fishery is just upstream from the Suquamish and had 10 times the waterway length to deploy nets.
Additionally, the Muckleshoots reported catch on only 7 days of the 15 days they had nets in the water in September, and 9 of 24 total days the Ship Canal fishery was open overall. That means over 62.5% of the days they were netting the Ship Canal, there are zero reported fish caught for their entire fleet of boats. This doesn’t pass any reasonable test.
But the Ship Canal is just part #1 of the fishery, let’s move upstream…
Lake Washington Fisheries
RECREATIONAL: The recreational fishery on Lake Washington has a well-deserved reputation for being unproductive for anglers. Creel count data from this year (Sand Point and Kenmore ramps) backs up that reputation, averaging only a single fish per 7 angler trips. Combined creel counts at Sand Point and Kenmore launch ramps show a total of 152 fish total caught for the season. These ramp fish checks provide an accurate count of only a subset of fishermen, as not all boats will exit through the monitored ramps and monitoring wasn’t round the clock. Based historical exploitation rates, we estimate the creel counts captured only 40% of the actual sport catch, but to be sure we’re not undercounting, we’re setting the creel counts at just 25% of the sport catch — yielding a total of 608 sport-caught coho.
TRIBAL CATCH: In addition to their Ship Canal fishery, the Muckleshoots took the unusual step of setting nets in Lake Washington this year. We characterize this as an unusual fishery because we’ve scoured the records back to 2008, and have not found a similar lake opening for the Tribe. Of particular concern is that this was one of the very few areas open for salmon sportfishing in 2016, and the nets presented a significant gear-conflict for recreational fishermen. In any case, the Tribe gillnetted the Lake for 7 days, but fish ticket results show catch reports for only the first 5 of those days, which is troubling. The reports we do have show a catch of 2658 fish taken in the Lake.
These in-lake recreational and Tribal fisheries are the last the Coho would encounter en-route to the hatchery, so we now have a picture of the run. Let’s see how these numbers all add up…
Checking the Numbers
Combining the sport (608) and both Muckleshoot fisheries (Ship Canal at 450, Lake Washington at 2658) we can now account for 3716 fish caught upstream of the locks. That number is a problem — and to understand why, let’s drill a bit deeper into the data gathered at the Ballard Locks.
It turns out that annual locks counts do more than just count the fish, they also record the percentage of marked fish which are bound for the Issaquah Hatchery. Last year, the total count of 18,779 fish was a mix of about 88% Issaquah hatchery fish mixed with 12% wild fish born naturally in feeder streams within the system. So at the moment the returning fish passed the locks, 16,500 (88%) of them were destined for the Issaquah hatchery. The problem is that only 3905 of those 16,500 fish arrived there. Somewhere between the locks and hatchery 13,000 fish disappeared, yet we can only account for 3716 caught based on the reported fishing activities.
Worse, of those reported caught, 12% of those fish (supported both by locks count and re-verified in recreational creel counts) were wild fish, so at this point we’ve actually only managed to account for 3270 of the 13,000 hatchery-bound fish that disappeared. That’s roughly ten thousand Coho who should have made it home, but didn’t. Which brings us to the next part of our story…
The Case of the Missing 10,000 Fish?
How could 10,000 fish have disappeared? We’ve explored the possibility of “strays” and freshwater mortality due to high water temperature. Could these have played an unusual role in the 2016 run? Our short answer is we don’t see any evidence that 2016 was unusual.
- Water temperatures (Lake Washington Buoy) in Mid-September 2016 were lower than in 2013 and 2014, and both of those years saw dramatically higher percentages of fish returning to the hatchery. In fact 2013 and 2014 were two of the highest temperature years recently and both saw very high rates of return to the hatchery. We don’t see a correlation between lake water temperature and fish returning to the hatchery.
- Strays? While some percentage of fish surely stray somewhere between the locks and the hatchery the percentage would seem to be relatively low historically. In recent years we’ve seen runs where impacts from Tribal & Sport fishing, freshwater mortality and strays combine for less than 30% mortality (2009-19%, 2011-28%).
- We believe a historically reasonable rate of up to 15% of the run could have been strays + freshwater mortalities. This could only account for 2,500 of the missing 10,000 fish.
So what happened?
We still are left with thousands of fish that we can’t account for. We think there are only two plausible explanations for what happened last year. Either the fish counts at the locks were inaccurate or the Muckleshoot Tribe reports understate their actual catch by a significant amount.
We believe a preponderance of the evidence indicates that that the Muckleshoot reports are inaccurate/understate their catch significantly. The following indicators all support this assertion:
- Divergence of those reports from the locks counts patterns-both 2016 and historical.
- Divergence of those reports from the Suquamish catch report patterns just downriver.
- Lack of reporting on more than half the days the ship canal fishery was open.
- Average catch reporting of 18.75 fish/open day for the entire Ship Canal fleet (often 10+ fishers) fails any basic economic analysis. Captains would never fish these openers for less than 2 fish per boat per day.
- Extraordinary 2016 effort: There were 24 days open for fishing in the Ship Canal, and 7 days open (overlapping) in Lake Washington — for 31 total area/days. That is the largest number of area/days fished in the 9 years of records we’ve reviewed. In fact the Tribe had more open area/days this year than in 2008, 2009, 2012, and 2014; all of which had 20%-35% larger runs counted at the locks.
At this point, there are many questions to ask and avenues to follow — which we’ll do in future stories…
- When a fish run we have such excellent data for is experiencing questionable Tribal reporting — it certainly undermines credibility of fisheries with less available data.
- Trust-but-Verify: How can we begin to rebuild trust between co-managers by improving this process? Would agreeing to mutual surprise audits and verification improve both relationships and reporting accuracy?
- The numbers, including our estimated 15% hatchery strays, indicate the two Tribes netted more like 12,000 coho on this run last year–well in excess of their 50% of the harvestable fish. Will the State attempt an internal reconstruction of this run, and pursue efforts to reclaim any of these fish? Under the repayment provisions of the Federally Approved Puget Sound Salmon Management plan, the Tribes would owe catch allocation back to the State for 2017.