
Cheers for the debate to and fro guys,
I'd like to sum up my position that, this technique is absolutely suitable for spate rivers (and not just something for benign chalkstreams)
In areas where goosanders are likely to make a significant impact (i.e. where depth, flow and dimensions of channel are low enough for them to hunt with their famous very high efficiency) - you are also likely to be able to successfully install cabled tree "kicker" cover to at least always prevent a total wipeout of local grayling populations for example.
Finally, even on rivers that sometimes rip banks out (and angular concrete is incredibly susceptible to the river working slowly behind/under it and prising it away from the "true" bank) - there will be reaches where it will be possible to anchor them well enough to give you several years' benefit. Co-incidentally, these slightly more sheltered areas would tend to be the most severely predated (because the pace of the current would be more favourable to the birds in those areas).
Now - I'm not saying that these kind of things can be put indiscriminately anywhere on a raging wild Scottish/Alaskan river; but that on balance, there will be spots in every river system where tipping the balance of odds between survival and mortality due to predation by birds (and everything else) will have massive benefits to wild trout, salmon and grayling populations.
In fact, I would go so far as to say that there is not a single river SYSTEM where (if there was a deficiency in the amount of marginal brashy cover - if it ain't broke don't fix it) tree kickers could not be installed at strategic points.
The danger for anyone reading an initial response along the lines of "all very well but wouldn't work on a big, steep river" could go away thinking that this approach is totally irrelevant to their local fishing. Whereas, in fact, it will
almost always be possible in the appropriate site, using the appropriate anchoring.
N.B. re willow photo - that is just one that I happen to have where there is a bit of water going down. Coincidentally, the river Wharfe (site of another much larger tree kicker) at Huby is probably 3 x the width of the Goyt at this point (see photos on sheet 11 and sheet 13 out of 24 in this section of the upland rivers habitat manual:
http://www.wildtrout.org/images/PDFs...s_section5.pdf
-photos taken during absolute low flow condition and sheet 13 photo shows the width more accurately because it avoids the foreshortening evident on sheet 11 (PS the guy with the white hair is professional yorkshireman Ollie Edwards

)
Although, speaking of width actually, once the channel width exceeds the width of the tree kicker sufficiently to prevent "edge effects" of the opposite bank (i.e. friction causing turbulence from the opposite bank that affects the near bank structure) - it doesn't matter at all whether the river is a mile wide; it just depends on the reach mean velocity of water (a function of discharge, cross sectional area, bed slope and hydrological "roughness"). In the classic case, the flow velocity will, of course, be much greater in narrow "pinch points" of the channel.
I take the point that, generally, the larger the width of river, the larger the discharge; but there are a lot of other things that determine the shear velocity experienced by the structure hanging in the water.
Basically, I could have shortened the above essay to "If you have a predation problem, give the WTT a call and we can assess where/if cover can be installed. There will be very few instances where this will prove impossible"

N.B. Edit - I do, of course, accept that in specific reaches where 5-ton boulders are moved on an annual/sub-annual timescale, it will be difficult to find suitable spots!