In-season Fraser sockeye forecasting… making it up as we go

On a post earlier this week I linked to the August 20, 2010 Watershed Talk — a publication of the Fraser River Aboriginal Fisheries Secretariat (FRAFS).

Watershed Talk August 20 2010

FRAFS staff (biologists) have written a decent summary of some of the shenanigans of this year’s sockeye forecasting: “Fraser Sockeye In-season Update – A good year for sockeye!

Apparently some of my comments from earlier this year on effectiveness of some of the hydroacoustics and DIDSON-related counting methods may not be that far off. Some folks have tried to suggest that these methods are rather “exact”; others have suggested they are voodoo-science at best. I think my perspective is somewhere in the middle… decent tools for getting a rough idea of fish migration in deep, large, murky rivers.

The article outlines how Pacific Salmon Commission staff have discovered that the hydroacoustic counting methods utilized at Mission (in the lower Fraser Valley — Vancouver suburbs) has count discrepancies when compared to similar counting methods utilized a little ways further upstream at Qualark. The Qualark counts have been much higher than the Mission counts — they should be quite similarly related as it is the same river, with a few tributaries that sockeye may migrate up between the two counting stations.

While the estimates have been correlating fairly well for part of the run, Qualark has been generating slightly higher numbers than has Mission. However, in recent days that difference has grown significantly. Salmon Commission staff have analyzed the differences and are quite confident that the recent Qualark estimates are indeed correctly reflecting a stronger return (higher number of fish) than has Mission. They have adjusted all their in-season run size estimates and are currently assessing the reasons for the discrepancy. The width of the river at Mission, water depth, and fish behaviour are thought to be confounding the Mission hydroacoustic estimates.

“Adjusting in-season run size estimates” due to “confounding” factors (e.g. fish behavior… those darn fish, what are they thinking…) — I call that “making it up”. And, I don’t mean that as a shot at the professionalism or scientific acumen of folks involved… it’s simply that using electronic machinery to count biological critters generally results in estimates at-best.

And really, how can we be so sure that the “higher” returns being seen at Qualark are more accurate?


And what are the numbers informing in-season estimates… well… scale samples from test fisheries.

How many scale samples?

Well… here’s the charts from the Pacific Salmon Commission info sent out prior to conference calls:

PSC scale samples from test fishing

Along the far left “Area/Gear” refers to fishing area — for example first line is Area 20 (just west of Victoria in Juan de Fuca) purse seine. The second column with the date and (n) with “n” referring to the number of scale samples (as far as I understand it… anyways). Then it’s % of Fraser sockeye. Then it’s the stock percentages broke out into the four groups made up of approximately 19 stocks, of which there are actually about 200 distinct Fraser sockeye stocks. Each acronym for those stocks is at the bottom broken down into the four groups.

How accurate is this info?

Well, on the Area 12 purse seine on Aug. 21 — approx. middle line of chart — over 26,000 sockeye were caught in that particular segment of the test fishery that day. And n=99?

Is one to surmise that 99 scale samples out of 26,000 fish caught gives a good representation of the stocks… hmmm??

On Aug. 19th there was an all time record catch in the Area 12 purse seine test fishery: 84,040.

Scale samples: n=99.

99 scale samples from 84,000 fish sounds like a very accurate prediction tool…

_ _ _ _

My point here isn’t necessarily to take shots at the methods, as I’m not a pre-eminent fisheries scientists and would not be able to make better suggestions. My point is that the methods of “confirming” all these estimates (e.g. computer models, simulations, test fishing, scale samples, hydroacoustics, etc.) is by confirming Sockeye spawners on the spawning grounds.

Spawner estimates is a wonderful example of more — exactly as it says — “estimates”. There’s mark-recapture, counting fences, stream walks, helicopter overflights, and other wonderful estimating tools.

As much as many folks suggest this whole salmon thing is a very precise practice… it is far from it.

It is simply fancy tools that “kick-out” fancy estimates.

There are lots of good folks working hard at these estimates — however they’re still fancy estimates with absolutely no method to truly “confirm” that the models, scale sampling, test fishing, and so on are “accurate”.

Then throw in terms like “ecosystem-based management” — ever present in the Wild Salmon Policy — and I tend to call ‘bullshit’.

We simply don’t know… we’re trying hard to estimate; but really… we’re making it up as we go along… and that’s OK; it’s the defensiveness and insistence by those involved that we do know what’s going on…

well… i think this year is a fine example that we definitely do not know.

2 thoughts on “In-season Fraser sockeye forecasting… making it up as we go

  1. Bob

    As someone who made run size estimates for years, got a better idea? Perhaps we should just leave them alone, in the river, and wait until they have all spawned. Then we can concentrate all the effort on a carcass census and get (a much better estimate of) the real number of fish in the run. Of course that would mean no fishing.

    Then, after all is said and done, we still wouldn’t have much of an idea as to what resulted in those numbers; spawning escapement of the parents, juvenile river/lake survival, outmigrant survival, ocean survival, poor/good fishing conditions for returning adults, etc.

    Right back were we started……
    Any good ideas?

  2. salmon guy Post author

    I hear you Bob and thanks for the comments.
    I don’t know if i have good ideas, I’ll leave that for others; however, my point on a few posts has been that I see very little – if any – effort in research on traditional salmon-human relationships… I hesitate to call it enumeration.. or counting methods. In my travels through salmon territory I’ve heard various stories and methods. Most of these combined selectivity and analysis of run-size and health. One of the most common was the use of fish weirs (this being limited to river sizes where this can work), as well as fish traps. There are some fantastic pictures of these in the BC Archives and sketches in Hilary Stewart’s book on salmon, as well as “Cedar”.

    For example, in the Yukon near the town of the Dawson City is the “Klondike” River. It, as I have had it explained to me, was once one of the greatest producers of Chinook salmon on the entire 3000 km long Yukon River. The Tr’ondëk Hwëch’in people for eons caught Chinook on the river. And as I understand it Tr’ondek is the proper pronunciation — “Klondike” is the anglicized version of the version. The meaning of Tr’ondëk in the Gwichin language of the area is something to the effect of “the sound of stakes being pounded into the river bed.”

    If you have, or do visit Dawson City, just downstream of the junction of the Klondike and Yukon is a cultural centre built in the 1990s. The centre was built in such a way as to represent a salmon trap or weir, with elements of fish drying racks. It’s a beautiful building.

    The point of this — and I’ve heard this from many different First Nations folks — is that weirs and fish traps were a method of capturing salmon alive and being able to selectively harvest by species, size, and sex – as well as get a good idea of the size and health of runs. There were also the political aspects to these weirs, in that downstream folks could have a big impact on upstream folks by not allowing fish through.

    In a sense… counting fences provide some of this same effect — however, there are the excuses of (wo)man-power to operate the fences.

    Dip netting is also a non-lethal method of counting. And then there’s the work of LGL and their fish wheels. Much more use of fish wheels could allow non-lethal methods of enumeration — as well as harvest.

    And then there’s just good old stream walks.

    My issues isn’t necessarily with some of the enumeration methods — just the lethal ones, like test fishing, it’s not required. My issue is with the way in which enumeration methods are held up as gospel, and their effect on harvest strategies, and the connection back to computer modeling and simulations. For example, this year, DFO set out to harvest only 25% of the Early Summer group of Fraser sockeye. This is in exception to their goals of 60% on other groups (Summer and Late Summer). The reason for the reduced harvest was to try and limit impacts on endangered stocks such as the Bowron and others in the upper Nechako that really are on a death spiral. There were many concerns brought forward during pre-season planning by upper Fraser First Nations. And some credit needs to be given to DFO on setting that goal with conservation concerns in mind.

    Looking at in-season info it looks like the harvest rate on Early Summers will be about 23-25% or so… However, this won’t be known for sure until there is confirmation of in-river spawner counts which will be bounced back over test fishing, commercial catch, and the various computer models. All of those methods are generally regarded as akin to gospel. And thus, much frustration when in meetings with fisheries officials and they start tossing around these numbers as if we actually “know”.

    We don’t… as is being made very clear by the fish this year.

    If anything, I think – and hope – that most scientists looking at these issues are being greatly humbled by how much we don’t know (and some are, from what I hear and have heard). As the saying goes, the more we know, the more we learn we don’t know. I use the analogy of the old Greek (or Roman) monster Hydra — cut off one head and two pop up in it’s place… it’s the same with questions about salmon.

    Lastly, one of the more common suggestion I’ve heard from many folks this year (First Nations and non)… leave the Fraser sockeye alone for a life cycle or two (4 to 8 years). Then there wouldn’t need to be all of this tension between folks looking to harvest them. As well as, there wouldn’t be all of this finger pointing and searching for one “smoking gun” (as referenced by folks at the Fraser Sockeye Forum in late March of this year). As well as pretty unproductive comments about black market sales, un-reported catches and so on.

    just a few ideas… for what they are worth.

    I appreciate the questions, comments and discussion — that was my whole purpose for setting up and maintaining this weblog… There has to be a different way; the history of many of the methods we use in the present day; and this insistence by scientists that the ‘answers’ lie in science — I find somewhat naive. There are hundreds of thousands of people throughout salmon country that have intimate knowledge… call it community knowledge, local knowledge, traditional knowledge, or maybe just… knowledge.

    Other ways of knowing. I just don’t think the issue is going to be solved by intricate equations and computer models. That’s the same method that brought us financial derivatives, and we know where that got us… (e.g. sub-prime mortgage).

Leave a Reply

Your email address will not be published. Required fields are marked *