"An evaluation of appropriate levels of sampling needed to develop age composition estimates was made by appealing to multinomial sampling theory. Random sample sizes of at least 600-800 read scales are recommended for each stratum where age composition is of interest. This will suffice to assure confidence intervals on proportions in each age class no wider than 0.05, provide acceptable relative precision for reasonably small proportions, and have a high probability of detecting rare age classes when present. Sample sizes have tended to be larger than the recommended levels at Iron Gate Hatchery, Trinity River Hatchery, the Yurok tribal fishery, and in some cases, the total river recreational fishery and the Hoopa Valley tribal fishery. Spawners in some sectors (e.g., Scott River, Bogus Creek,Willow CreekWeir) have generally been sampled adequately for scales, while others (e.g., Salmon River, Shasta River, lower Trinity mainstem) have frequently been undersampled"--Executive summary.

National Marine Fisheries Service (NMFS)