Abstract
The 2022 North Sea Sandeel age reading exchange took place on SmartDots (ID 424) in May 2022 in preparation for the WKSANDEEL 2022. Seven readers took part, 3 from Norway (all advanced readers providing age data for stock assessment purposes) and 5 from Denmark (2 advanced readers providing age data for stock assessment purposes and 2 basic readers who do not provide age data for stock assessment purposes). The WKSANDEEL 2022 Working Document has been uploaded to SmartDots as the summary report.
A set of 120 otolith images were uploaded to the SmartDots platform and readers were asked to annotate, provide an age and a quality assurance score for each otolith image.
The otolith images were previously read in 2019 (https://smartdots.ices.dk/SampleImages/2019/219/2019%20North%20Sea%20Sandeel%20Age%20Reading%20Exchange%20Report.pdf) . The same 3 readers from Norway took part in both exchanges, whereas from Denmark only 1 reader took part in both exchanges. Based only on age readers who provide age data for stock assessment purposes (advanced readers) the weighted average percentage agreement (PA) was 81 % and the weighted average coefficient of variation (CV) was 24 % (based on advanced readers only). In 2022, results improved and the weighted average percentage agreement (PA) was 87 % and the weighted average coefficient of variation (CV) was 20%.
In 2019, based on all readers the weighted average percentage agreement (PA) based on modal ages for all readers was 77 % and the weighted average coefficient of variation (CV) was 26 %. An improvement was also seen here with the weighted average percentage agreement (PA) increasing to 86 % and the weighted average CV at 24 %.
Included in both the 2019 and 2022 exchanges were a subset of 40 otoliths (from SA1) with 100% agreement in the 2016 exchange (WKSAND 2016 WD). Based on only these 40 otoliths, the weighted average percentage agreement (PA) in 2019 was 85 % and the weighted average coefficient of variation (CV) was 24%. An improvement was seen in 2022 when the weighted average percentage agreement (PA) was 92 % and the weighted average coefficient of variation (CV) was 17%.
In 2019 the following age reading issues were apparent; a) incorrect interpretation of the otolith edge in Q4 where some readers were counting an extra year and b) disagreement as to whether or not a faint innermost translucent zone (present in some otoliths) should be counted as a true winter ring or not. The former issue appears to be resolved as a result of repeated calibration of readers and feedback on age reading issues. The latter is a reoccurring issue (see Image Examples) which needs attention and requires otolith microstructure examination of problematic otoliths from different areas in order to validate whether or not this is a true winter ring.
2022 results do not show any indication that a single stock or month of capture (or age) is more difficult to read even though in 2019 concerns were raised over image quality of the otoliths from san.sa.3r and san.sa.5r which were mounted in eukit and which may have contributed to the lower PA for these areas/stocks. CV at modal age 1 is highest but it should be noted that the calculation of CV is dependent on age and CV at modal age 0 is not calculated for this reason. Even though PA at modal age 0 is high (84%) it is lower than PA at modal ages 1 (90%) and 2 (93%), indicating there are some difficulties in the correct interpretation of modal age 0 with a general pattern of positive bias in relation to modal age which is interpreted as an overestimation of age.
A set of 120 otolith images were uploaded to the SmartDots platform and readers were asked to annotate, provide an age and a quality assurance score for each otolith image.
The otolith images were previously read in 2019 (https://smartdots.ices.dk/SampleImages/2019/219/2019%20North%20Sea%20Sandeel%20Age%20Reading%20Exchange%20Report.pdf) . The same 3 readers from Norway took part in both exchanges, whereas from Denmark only 1 reader took part in both exchanges. Based only on age readers who provide age data for stock assessment purposes (advanced readers) the weighted average percentage agreement (PA) was 81 % and the weighted average coefficient of variation (CV) was 24 % (based on advanced readers only). In 2022, results improved and the weighted average percentage agreement (PA) was 87 % and the weighted average coefficient of variation (CV) was 20%.
In 2019, based on all readers the weighted average percentage agreement (PA) based on modal ages for all readers was 77 % and the weighted average coefficient of variation (CV) was 26 %. An improvement was also seen here with the weighted average percentage agreement (PA) increasing to 86 % and the weighted average CV at 24 %.
Included in both the 2019 and 2022 exchanges were a subset of 40 otoliths (from SA1) with 100% agreement in the 2016 exchange (WKSAND 2016 WD). Based on only these 40 otoliths, the weighted average percentage agreement (PA) in 2019 was 85 % and the weighted average coefficient of variation (CV) was 24%. An improvement was seen in 2022 when the weighted average percentage agreement (PA) was 92 % and the weighted average coefficient of variation (CV) was 17%.
In 2019 the following age reading issues were apparent; a) incorrect interpretation of the otolith edge in Q4 where some readers were counting an extra year and b) disagreement as to whether or not a faint innermost translucent zone (present in some otoliths) should be counted as a true winter ring or not. The former issue appears to be resolved as a result of repeated calibration of readers and feedback on age reading issues. The latter is a reoccurring issue (see Image Examples) which needs attention and requires otolith microstructure examination of problematic otoliths from different areas in order to validate whether or not this is a true winter ring.
2022 results do not show any indication that a single stock or month of capture (or age) is more difficult to read even though in 2019 concerns were raised over image quality of the otoliths from san.sa.3r and san.sa.5r which were mounted in eukit and which may have contributed to the lower PA for these areas/stocks. CV at modal age 1 is highest but it should be noted that the calculation of CV is dependent on age and CV at modal age 0 is not calculated for this reason. Even though PA at modal age 0 is high (84%) it is lower than PA at modal ages 1 (90%) and 2 (93%), indicating there are some difficulties in the correct interpretation of modal age 0 with a general pattern of positive bias in relation to modal age which is interpreted as an overestimation of age.
| Original language | English |
|---|
| Place of Publication | Copenhgen, Denmark |
|---|---|
| Publisher | International Council for the Exploration of the Sea (ICES) |
| Number of pages | 28 |
| Publication status | Published - 2022 |