There was a study out of OR, on the OPI coho that was really instructive to me.

They found a nice 4-variable model that explained most of the variation in marine survival for a cohort. The problem was that the 4 variables were temporally sequential, such as (I am making this all up) upwelling when they entered the ocean, PDO in late summer of first year in salt, spring upwelling, and summer ocean temps.

The described what happened but ween't good as forecasts because any one of them could overwhelm the others. The first 3 were "bad", the last "good" and you get a good run. But, that last piece of information is available just before they enter the terminal areas where updates could be done.

This may explain why the forecasts we base ocean fisheries on work so well. Conditions late in the return can overwhelm what went before, knocking numbers either way down or making them be higher than we planned for. Just another reason to manage terminally.