Statistical Assessment of the Age–Length Key

Since 1934 when Fridriksson originated the age–length key, it has been widely used by fisheries biologists to estimate age distributions of populations. In recent years, there has been a general recognition that often the key has little value, or even worse, gives biased results. The analysis presented here indicates why the age–length key is so susceptible to bias. More importantly, a criterion is presented for determining whether the age–length key should be used in a particular situation. If the key is to be used, results from examples indicate that random age subsamples (i.e. the number of specimens aged from each length category proportional to the number in each length category) are superior to fixed age subsamples (i.e. a constant number of specimens aged from each length category). Generally, small increases in the age sample will likely increase the accuracy of an age-distribution determination more effectively than relatively large increases in the length sample.