NOAA logo AMJ 2001 Quarterly Rpt. AMJ 2001 sidebar

Pacific Cod:
The Ageing of a Difficult Species

by Nancy Roberson

(Quarterly Report for April-May-June 2001)

Pacific cod otolith
An example of a Pacific cod otolith prepared by thin sectioning
shows a clear pattern of opaque and translucent zones.

"It is not complete exaggeration to say that the best (most reliable) specimen of Pacific cod otoliths is worse than the worst specimen which the writer has examined from the Atlantic."  K.S. Ketchen (1970)

Pacific cod (Gadus macrocephalus) is one of the most difficult fishes of all commercially important Alaska groundfish species to age. Historically, fish scales and otoliths have been the two most common structures used for determining the ages of fish species. Unfortunately, age-readers employing these structures have experienced limited success in the case of Pacific cod.

The Pacific Biological Station in Canada permanently suspended Pacific cod ageing in 1978, after age estimates derived from scale readings began yielding year classes that were inconsistent with field survey time series of length frequencies.  The Alaska Fisheries Science Center's (AFSC) Resource Ecology and Fisheries Management Division (REFM) Division is responsible for stock assessment of Pacific cod in the Gulf of Alaska and eastern Bering Sea. The REFM Division's Age and Growth Program used scales for determining the age of Pacific cod from 1976 to the early 1980s. Thereafter, the break and burn method was used on otoliths. In 1990-92 the AFSC detected a downward shift of Pacific cod length-at-age, with a dramatic decrease in the average lengths for 1-, 2-, 3-, and 4-year olds (Table 1 below). Unable to pinpoint the reason for the shift and given the inherent difficulty of ageing cod, production ageing of Pacific cod was suspended at the AFSC indefinitely.

Table 1.  The 1977 year class modal lengths-at-age (cm) compared with 1988-92 average lengths-at-age which were calculated from otoliths collected during AFSC trawl surveys.

Year Age 1 Age 2 Age 3 Age 4 Age 5
1977 22.0 33.0 41.0 51.0 61.0
1988 24.2 34.9 40.9 50.1 58.6
1989 19.2 33.8 41.7 51.9 58.9
1990 16.9 27.2 35.9 46.8 57.2
1991 19.6 28.4 36.2 47.4 55.6
1992 18.7 26.0 33.9 44.4 53.6

Pacific cod stock assessments in both the United States and Canada have depended largely on length frequency data alone to model the population age structure because of ageing difficulties. However, the use of length frequency data as proxies for age data has proven problematic. The current length-based model for Alaska Pacific cod has produced results that are difficult to explain. One explanation is that external factors such as ocean conditions affect somatic growth to such a degree that length-at-age within the population is highly variable and difficult to model. Otoliths, on the other hand, are composed of layer upon layer of daily growth accumulations and are theoretically permanent records of growth, more independent of external factors. Consequently, the Age and Growth Program initiated a new study in 1998 to reexamine the otolith ageing structure for Pacific cod. Research entailed determining the best otolith preparation method; establishing and justifying reading criteria; and investigating the 1990-92 shift in length-at-age. The following article summarizes the M. S. thesis of Nancy Roberson of the Age and Growth Program.

Otolith Preparation
At the beginning of the study a decision was made to use otoliths in conjunction with digital imaging. It was thought that the approach would yield the highest quality data for Pacific cod. The first step in this study was to find a method for preparing the otoliths that would maximize the clarity of the annulus pattern of alternating opaque and translucent zones. (Note: all further references to "zones" refer to reflected light.) The traditional break and burn method involves cutting through the transverse plane of an otolith and burning it over an alcohol flame.  For many species of fish, burning sufficiently enhances the zones, allowing the age reader to better differentiate the opaque and translucent zones.  However, burning does not consistently darken the rings of Pacific cod otoliths (Figure 1 below).

Pacific cod otolith
Figure 1.  An example of a Pacific cod otolith prepared by the break
and burn method shows the faintness of the annular pattern.

An alternative method for preparing otoliths is known as “thin sectioning.” This method, used by many laboratories worldwide and particularly for daily growth studies, involves removing a thin cross section of an otolith and encasing it in resin before analysis under a microscope. After trial and error with various resin types and section thickness, a preparation method was chosen that uses a black resin (Technovit 3040) and sections the otolith to a thickness of 0.2mm with a Hilquist thin section machine.  This method provides better contrast between opaque and translucent zones and greater uniformity of color within the respective zones, thus making it easier to discern patterns (Figure 2 below). Furthermore, sectioning an otolith creates a polished, two-dimensional surface that facilitates digital measurements and imaging, which the coarse surfaces of break and burn preparations do not.

Pacific cod otolith
Figure 2.  An example of a Pacific cod otolith prepared by thin sectioning
shows a clear pattern of opaque and translucent zones.

Reading Criteria
Reading criteria are guidelines age readers use to interpret otolith patterns. Reading criteria are particularly important in the scientific study of ageing fish because they are the qualitative information on which age estimates are based. The criteria used in this study were designed to instruct the age reader on how to classify the translucent zones of an otolith pattern.  A translucent zone is a compilation of many closely spaced daily growth rings, which in theory occur when there is decelerating or slow somatic growth. There are two categories of translucent zones: annuli (formed once a year and normally associated with the age of a fish) and checks (any subannular mark).

Pacific cod have a strong tendency to develop otolith checks at young ages (under 6 years), making it difficult to age the species to an exact age.  Evidence from tagging studies and surveys indicates that the species is relatively short-lived, with a maximum age of approximately 13 years.  Differentiating between checks and annuli, therefore, is crucial in the age determination of Pacific cod. Checks that are counted as annuli result in overestimation of age and underestimation of size-at-age. For this reason, working with validated reading criteria is critical.

This study followed traditional qualitative ageing criteria to distinguish annuli from checks: completeness of a zone around an entire section, zone darkness, and spacing between zones.  Pacific cod otoliths are confusing to interpret because their checks frequently have the same characteristics as their annuli: the checks are just as dark, and their ring-like pattern just as complete. This in turn makes it difficult to differentiate the two, other than by spacing between zones (Figure 3 below). Fortunately, as Pacific cod get older it is easier to distinguish spacing patterns between zones.

ring pattern of 1-year old Pacific cod otolith

Figure 3.  A ring pattern indicative of a 1-year old fish illustrates the difficulty of ageing young Pacific cod. The checks, marked on the photograph, share the same characteristics as the annulus: darkness and completeness around the otolith. The subtlety of the differences can lead to the misclassification of checks as annuli. The result is overageing of young fish.

In this study, rings that could be interpreted as both annuli and subannular marks  were classified as checks. In addition to counting annuli, the annuli and strong checks were hand-traced with Optimas 6.5 imaging software, resulting in measurements for ring area and ring axis. This made qualitative data more quantitative and enabled the reading criteria to be statistically evaluated.

A total of 174 otoliths collected during the NMFS 1989 Bering Sea Groundfish Survey served as the initial study sample to develop the reading criteria.  The 1989 study sample enabled calibration of age estimates with ages generated prior to the 1990-92 problem years.

Justifying Reading Criteria
Reading criteria should reflect the actual age of the fish.  To evaluate the accuracy of the   reading criteria developed from the initial study sample, a second sample of approximately 106 otoliths from a previous tag and recovery study was used. The advantage of using otoliths of tagged fish was that length at release was available.  This information was helpful in two applications:  1) to determine whether back-calculated fish lengths based on ring areas are superior to those based on ring lengths and 2) to provide an indirect validation of ageing criteria by comparing predicted growth increment for a given time at liberty versus the growth increments actually observed.  (This second computation utilized von Bertalanffy growth parameters previously estimated from Pacific cod tag recovery data and the reader age of the otolith at the time of recovery.)

The study found that fish lengths back-calculated using ring areas were much closer to observed lengths than those calculated with ring lengths (Figures 4 below). Although back-calculations are typically performed using radial or diametral measurements, these findings are not surprising in that ring area is a more comprehensive measure of otolith three dimensional growth.

L1 predicted with otolith areas
Figures 4 (above & below.)  Observed versus predicted back-calculated length-at-age estimates using measurements of ring area and measurements of ring length of otoliths from tagged Pacific cod.
L1 predicted with major axis on otolith

The strong relationship between ring area and fish length was important to this study because it provided a quantitative means of evaluating the qualitative reading criteria through back- calculations. Back- calculation allows the examination of fish length-at-age less than the age at capture. However, back- calculations are valid only when they are based on true annuli, not on checks. Length-at-age based on back-calculation can be a test of whether ideas of true annuli and checks are correct. In order to perform this test, use was made of the original 1989 training sample.  Using the previously measured areas, fish length-at-age was back-calculated twice: once using only what were considered annuli and then again including checks (Table 2 below).

Table 2.  Two separate calculations of the 1989 fish length-at age (cm), estimated from back-calculations counting checks and from not counting checks compared to the 1977 year class modal length-at-age.

Fish Age 1989 back-calculated fish length (cm) not counting checks 1989 back-calculated fish length (cm) counting checks 1977 year-class modal length-
at-age (cm)
1 20.7 16.6 19
2 32.5 26.6 33
3 41.9 35.2 41

The back-calculated fish length-at-age estimates for the 1989 sample using annuli alone were closer to previously validated length-at-age measurements (1977 year class) than the back- calculated lengths using checks.  These calculations supported the criteria used to differentiate checks from annuli. Fortunately, these reading criteria could be somewhat validated using the otoliths from the tagged fish.  Using von Bertalanffy growth parameters estimated from the tagged fish and their age at recovery (estimated with current ageing criteria), the growth increment for each tagged fish was estimated and compared to the observed growth increment.  Estimates calculated with the current reading  criteria were roughly the correct ages (Figure 5 below).

otolith growth comparisons
Figure 5.  A comparison of estimated growth increments and observed growth increments based on otoliths from tagged Pacific cod. This provides an indirect validation of ageing criteria because estimated growth increments are a function of von Bertalanffy growth parameters estimated from tagged fish and the otolith age assigned to a fish at the time of recovery.

A final test of reading criteria was performed through a more direct comparison: simply ageing the tagged fish from length-at-release plus the time at liberty.  Seventy-five percent of these fish were within 1 year of age based on otolith readings, and 94% were within 2 years.

Validation of reading criteria is most convincing when achieved through daily growth counts and marked otoliths in tag and recapture studies.  However, the indirect validation techniques employed in this project using back-calculations on ring area measurements and otoliths from tagged fish provided evidence that criteria used for this study  were accurate.

1990-92 Shift in Age at Length
The final phase of this investigation examined the decrease in length-at-age of Pacific cod in 1990-92. One hundred and sixty four otoliths, subsampled from the 1992 Bering Sea trawl survey, were used  in conjunction with the 1989 sample to compare current ageing criteria to the criteria used in 1992.  The sample was composed of specimens less than 50 cm, representing young fish, because the shift was seen in ages 1-4 years old.

The 1992 sample was problematic from the standpoint of otolith preparation. First, it was impossible to thin section the otoliths because over the years the survey sample had been examined so many times that no whole otoliths were available to section.  As a result, it was necessary to use previously broken and burned otoliths.  Also, the annuli on the burns were too faint to examine clearly, and only the outer edge of the otolith could be measured. Fortunately, edge measurements represent otolith size at fish age and could be used in the comparisons as proxies for the ageing criteria used in 1992.

For each sample, the outside edge of the transverse plane of the broken and burned otoliths was traced and the areas calculated. The original 1992 ages were used.  For the 1989 sample, ages and edge measurements collected earlier in this study were used.  Since both samples had similar season sampling dates (early summer), it was concluded that differences in edge growth were negligible.

When 1989 and 1992 otolith areas were compared, the average otolith size-at-age was smaller in the 1992 fish than in the 1989 fish.  The smaller otolith size alone can largely explain the change in size-at-age of Pacific cod between 1989 and 1992 (Table 3 below). However, these results cannot explain whether the smaller ring sizes in 1992 were due to misread ages (counting checks inadvertently) of the 1992 sample or real differences in otolith size-at-age.

Table 3.  The difference in whole otoliths areas (mm2), at different ages, obtained from 1989 sections in the current study and break and burns aged in 1992.  This difference is converted into differences in fish length by using the otolith area versus fish length relationship: for every mm2 increase in otolith area, fish size increases 3.4 cm (regression of fish length on otolith size using 1989 and 1992 sample.) This gives a predicted difference in fish lengths (cm) between 1989 and 1992. A comparison is made with the observed difference in fish length from the same sample of fish.
Fish age Difference in otolith area for 1992 and 1989 sample results The predicted difference in 1992 and 1989 fish lengths using a  multiplier of 3.4 cm Observed sample difference in 1992 and 1989 fish lengths (cm) 
1 +0.52   1.77   1.08
2 -1.03 -3.50 -4.89
3 -1.28 -4.35 -3.94
4 -0.47 -1.60 -3.01


Table 4.  Fish length-at-age (cm) estimated from back-calculations, the original 1992 ages, and a re-ageing of some of the 1992 otoliths using the current ageing criteria.
Fish age 1989 back-
calculations not counting checks
1989 back-
calculations counting checks
1992 original ages 1992 re-ages using ageing criteria from this study
1 20.7 16.6 18.7 20.8
2 32.5 26.6 26.0 31.8
3 41.9 35.2 33.9 40.2
4 51.1 43.2 44.5 43.9

 In an attempt to resolve this issue, the 1992 samples were read according to current criteria.  This enabled the comparison of 1992 fish size-at-age using historic and current ageing criteria (Table 4 above). It seemed clear that the 1989 back-calculated lengths at age, counting checks, were similar to those associated with the 1992 original reading.  It also appeared that the 1992 re-ageing using current ageing criteria had resulted in size-at-age closer to the 1989 back-calculation without checks.  Therefore, it appears that the original 1992 criteria may have counted checks. However, there is still some concern that counting checks might not account for all the differences in length for 1-, 2-, and 3-year-olds shown in Table 1 above.  For example, consideration must also be given to the observation that halibut length-at-age declined sharply at the same time that Pacific cod size-at-age was seen to decline.  It could be that both a true decline in length-at-age and the inadvertent counting of checks both played roles in the observed decline in the length-at-age of Pacific cod in 1990-92. This issue will be examined in greater detail as the investigation of  production ageing of Pacific cod continues.

Thin-sectioning Pacific cod otoliths was investigated as an alternative preparation method for ageing Pacific cod to the break and burn method. The break and burn method is inadequate for ageing Pacific cod because it fails to provide sufficient contrast between the translucent and opaque zones on a consistent basis. Also, it is difficult to create an “even” burn over the entire surface without overburning or incinerating parts of the otolith.

Thin sectioning otoliths encased in black resin seems to maximize the clarity of the viewing surface by adding contrast to the faint annulus pattern. Furthermore, the resin preserves the physical integrity of the otolith so that all axes of an otolith can be examined. Precision testing will be conducted in the future to determine whether this truly is a better preparation method.

This investigation generated indirectly validated age reading criteria for Pacific cod otoliths through the use of a thin section preparation method and otoliths from tagged and recaptured fish. Back-calculations, based on digital measurements of what the reading criteria classified as annuli, were used to estimate size at initial capture of the tagged fish to show that the reading criteria were roughly correct. Analysis based on back-calculation showed that checks should not be counted. Back-calculated fish lengths based on otolith areas are superior to those based on otolith lengths.  These results support the use of otoliths to production age Pacific cod.

The decline of size-at-age observed in Alaska Pacific cod between 1989 and 1992 also was investigated. It could be that both an inadvertent counting of checks and a true decline in length-at-age played roles in the observed decline of the length-at-age of Pacific cod in 1990-92.  This issue will be examined in greater detail as the Age and Growth Program resumes production ageing of Pacific cod this fall.