Reducing bias in the estimate of fishing mortality in fisheries stock assessments by testing the unrealistic and critical assumption that natural mortality is known and constant

A very widespread but unscientific practice in stock assessment work is to posit a value for natural mortality, conduct assessmetn mathematical modeling based on this hypothesis, and then fail completely to test the hypothesis in any way. Usually the estimate of natural mortality is on the instantaneous scale and is denoted by M. If, in fact, the posited value for M is incorrect, the estimates of fishing mortality and stock size will be biased, possibly severely. The most common method for estimation of F involves subtraction of the estimate of M from total instantaneous total mortality, Z. Even if the estiamte of Z is unbiased, if the input estimateM is biased, then the estimate of F will be biased. I give examples from three assessments in which the constant value for M was directly or indirectly tested and rejected. The cases are a catch-survey assessment of Delaware bay blue crabs, tag-recature assessment of F in Chesapeake Bay striped bass and the assessment of weakfish.