An analysis of the observer data provided from the krill fishery was undertaken using a range of complementary methods of data quality control and variance investigation. A sample pooling approach analysis on the other hand revealed the distribution of variance in krill length at different scales and indicated that krill hauled in a given month/season/area were similar enough to be pooled to compute monthly length frequency distributions. Linear Mixed Models were used to remove systematic variability in krill length due to seasons, months and areas and this allowed estimates of variance components associated with different vessels and observers to be obtained; these were found to be statistically significant and a made a substantial contribution to the overall random variation. Recommendations for future sampling scheme, including the potential for an adaptive responding to the variability in krill population structure are both provided and invited.
Sarah Mackey (CCAMLR Secretariat)