Age Estimation Error Research Articles

Mortality Estimation: Biased Results from Unbiased Ages

Estimates of mortality rates from age distributions are biased by imprecision in age estimation, even if age estimates are unbiased. I have derived a method for predicting the magnitude of this bias from information on the precision of age determination. Monte Carlo simulations show that bias can be accurately predicted. The commonly used Chapman–Robson mortality estimator is shown to be robust to imprecision in age determination if all age-classes are included. Errors are likely, however, if one or more age-classes are excluded or if other mortality estimators are used. Biases can be corrected if the distribution of age-estimation errors is known.

The Forgotten Requirement for Age Validation in Fisheries Biology

A survey of 500 studies published between 1907 and 1980 that included estimates of fish age indicated that only 65% mentioned age validation or attempted to validate the ageing technique. In less than 3% was the technique validated for all age classes. Among the 35% that did not consider age validation, many did not consider the possibility that ages may be incorrect. Among 75 additional publications published in primary journals between 1965 and 1980 that assessed stock dynamics and used fish ages, only 40% mentioned or attempted age validation, and none successfully validated all age groups used in the analysis. Many investigators continue to neglect the critical study of age validation despite the clear direction of the early work on age determination. We show that estimated ages greater than the maximum validated age must not be considered accurate. Use of inaccurate ages has caused serious errors in the management and understanding of fish populations. Only by mark-recapture studies or use of known-age fish can all age classes in a population be validated. If such studies are not possible, fish should be aged by several methods, and the possibility of errors in age estimates must be considered. Received April 5, 1983 Accepted August 9, 1983

The effects of ageing biases on stock assessment and management advice: a case study on Namibian horse mackerel

We explore the influence of age-estimation errors on the results of the age-structured production model (ASPM) used for horse mackerel stock assessment in Namibia for the period 1961–2003. The analysis considered age data from eight readers collected during an otolith-reading workshop. Four scenarios of age-estimation errors were assumed: Case 1 — a reference age computed as the modal age of estimates obtained by the four most experienced readers; Case 2 — age readings from a precise and experienced (Namibian) reader of horse mackerel otoliths; Case 3 — age estimates from a reader that displayed positive bias compared with the reference ages; and Case 4 — age estimates from a reader that displayed negative bias compared with the reference ages. The age–length key of each case was applied to length distributions of survey, pelagic fleet and midwater fleet landings (1991–2003) to obtain catch-at-age data. These data were then used in the ASPM. Results obtained from Case 3 differed most significantly from the others and appeared to be unrealistic in terms of the state of the stock and negative log-likelihood estimates. The conclusion is that more resources need to be directed towards age determination, because management recommendations are highly sensitive to errors in ageing. Most effort should be placed into age estimation of age groups 3–5 (20–30 cm total length), but significant effort needs to be devoted to age estimation of midwater commercial samples. Finally, the extent of sampling and the raising strategy of length frequencies should be improved.