A Well-Established Technique
Reading the age of a fish from the rings (annuli) on its scales is a well-established technique with a history dating back over 250 years (Ricker 1975). However, like many well-established methods the general approach has failed to keep pace with modern analytic development. In particular, hierarchical Bayesian methods now make it possible to accurately predict the actual (as opposed to inferred) ages of fish by explictly modeling the biases that confound the relationship between age and annuli numbers. For example, a recent paper by Dortel et. al. (2013) details the application of hierarchical Bayesian methods to the accurate ageing of Yellowfin Tuna (Thunnus albacares) from otolith growth rings.
One Year \(\neq\) One Annulus
The process that leads to annulus formation is relatively straight-forward. As fish grow, their scales tend to get larger by depositing material on the growing edge. During the winter months, colder water temperature and restricted food limit growth which results in a darker year band or annulus. However, if winter conditions are favourable growth may continue and the annulus may be barely discernable, which results in missed annuli. Conversely, growth generally slows as fish increase in size and is impeded by processes such as the allocation of energy to reproduction (a phenomena that can result in false annuli). To top it all, during periods of stress scales can be reabsorbed resulting in the loss of annuli. As a result there is not a simple one-to-one correspondence between the number of annuli and the age of a fish.
A Black Art
Scale readers have long recognised the complexity of the relationship between the number of annuli and fish age - a complexity that has led some of them to refer to scale reading as a “black art”. To attempt to minimise biases, scale readers have tended to use all the available information. This includes the length of the fish at capture and the number of years-at-large for recaptures. For particularly challenging scales they often compare notes and if a consensus cannot be reached they assign the fish an average age. This approach makes sense in the absence of hierarchical Bayesian models if the objective is to assign each fish as accurate an age as possible.
However, if sufficient individuals have been recaught through time it is now possible to perform analyses that allow unbiased estimates of the actual ages (Dortel et al. 2013).
Count Annuli not Ages
Although modern analytic methods are now available, their use requires a fundamental shift in how scale readers approach their art. In particular, it is imperative that scale readers no longer view their task as scale ageing. Instead they should merely report the number of annuli they observe without any attempt to correct for biases. To make their task easier, readers should be presented with scales in such a way that they are blind to the fish’s length, when it was caught, or whether it has been previously aged. If they are uncertain whether one or more bands are sufficiently defined to constitute annuli then they should capture the uncertainty by assigning the fish a minimum and maximum number of annuli. All reported values should be whole numbers. Prior to commencing a study the scale readers should meet and formally define the criteria they will be using to identify annuli. But once the study starts the readers should count annuli in isolation so that they do not influence each other. Furthermore, the date and scale reader should be recorded for each count to allow multiple counts of the same scale by the same and different readers. A reader-date-stamp will also allow any changes in reader behaviour through time to be detected.
Although blind counting of annuli may seem counter-intuitive, its purpose is to ensure all scales are treated the same irrespective of available information especially capture history. As such any biases in the annuli counts for recaptured individuals of known age will be the same as for the other captures. This means that the modelled bias can be used to correct all the scales.
When Dortel et al. (2013, 10) applied such an approach to the increments on the otoliths of Yellowfin Tuna they concluded that
… the ageing error model was able to estimate the ageing biases and provide accurate age estimates, regardless of the age of the fish.
Although promising, there are two aspects of their study which may not extend to scale ageing on some systems. Firstly, otoliths deposit rings on a daily scale and, secondly, the actual fish aged were no more than five years old. Nonetheless, it seems reasonable to expect that the method will have value when applied to fish with strong seasonal growth differences that have not yet achieved their maximum length.
Dortel, Emmanuelle, F’elix Massiot-Granier, Etienne Rivot, Julien Million, Jean-Pierre Hallier, Eric Morize, Jean-Marie Munaron, Nicolas Bousquet, and Emmanuel Chassot. 2013. “Accounting for Age Uncertainty in Growth Modeling, the Case Study of Yellowfin Tuna (Thunnus Albacares) of the Indian Ocean.” Edited by Athanassios C. Tsikliras. PLoS ONE 8 (4): e60886. https://doi.org/10.1371/journal.pone.0060886.
Ricker, William Edwin. 1975. Computation and Interpretation of Biological Statistics of Fish Populations. 1975 ed. Caldwell, NJ: Blackburn Press.