• Information was taken from CCAMLR observer reports and interviews with fishing masters;
• The variation in reported conversion factors for similar reported products can be attributed to a number of reasons:.; o Differences in the method of processing the same products o Seasonal variation o Spatial variation o Krill water content o Reporting errors (recording percentage recovered instead of conversion factor)

• Uncertainty in reported Conversion Factors does not necessarily lead to uncertainty in total catch estimates as the majority of vessels interviewed made direct estimates of green weight catches; and,
• A need for closer monitoring of processing methods, calculation of green weight catches, and conversion factors where appropriate, this could be enhanced by greater observer coverage.

Catch uncertainty is a component of uncertainty that is not routinely considered by CCAMLR. However, given the currently reported variability in conversion factors for krill, a nominal reported catch of 600 000t could actually represent a catch in ‘green weight’ of 2.5 million tonnes. Quantifying the level of uncertainty in reported catches of krill would require information on product specific conversion factors (including the time-scale over which those conversion factors were produced) as well as the product composition of catches.

This paper describes French fishery targeting Patagonian Toothfish inside CCAMLR area, what are the data collected and how they are checked, a short description of prioritisation of observer’s task is given.

With the intent of stimulating discussion, we present four options for subdividing the precautionary catch limit of krill for Area 48 into the small-scale management units (SSMU) adopted at CCAMLR-XXI. The first three are static allocations: the first is proportional to estimated predator demand in each SSMU; the second is proportional to the estimated standing stock of krill in each SSMU; and the third is proportional to the standing stock less predator demand in each SSMU. The fourth option is a dynamic allocation based on land-based predator monitoring conducted just prior to or early in the fishing season.

The standard procedure used to extract catch-weighted length frequencies can only make use of CCAMLR data for D. eleginoides in Subarea 48.3 collected in the seasons 1992/93 and from 1994/95 onwards. The procedure cannot use the earlier data collected between 1984/85 and 1991/93, and data from 1993/94 because of a mismatch in Flag State which presently occurs between the length and catch data. A new procedure was developed which can use the mismatched data. In addition, historic data from 1986/87 to 1988/89 were entered in the CCAMLR database. The new procedure and recent data entry have extended the existing time series of catch-weighted length frequencies for D. eleginoides in Subarea 48.3 by nine (9) seasons. In all, data are now available for 21 seasons from 1984/85 to 2004/05. New data from the 2005/06 season will be added as soon as submitted. A comparison between the standard procedure and the new procedure indicates that the catchweighted length frequencies are similar where there is an overlap of seasons. However there is a systematic difference in the overall number of fish estimated each season. The basic procedure generally estimates fewer fish than the standard procedure and the difference between the two procedures is typically 5-10% (range: 1 to 34%).