CCAMLR

A survey of Patagonian toothfish, Dissostichus eleginoides,was undertaken in Division 58.5.2 in the vicinity of Heard Island in May 2004 to provide the information for an assessment of short-term annual yield in the 2004/2005 CCAMLR season. This paper provides a preliminary assessment of yield for the area of Division 58.5.2 to the west of 79 o 20’ E using standard CCAMLR methods. It presents a range of sensitivity analysis for the assessment based on deliberations at WG-FSA 2003 with respect to the surveys and ages included in the estimation of the recruitment series.

Tuck et al. (2003) describe an exact time of release and recapture stock assessment model and applied it to the stock of Patagonian Toothfish, Dissostichus eleginoides,at Macquarie Island that have been fished by Australian trawlers from November 1994. This work was previously presented to the Working Group on Fish Stock assessment in 2000 (Tuck et al. 2000; CAMLR-XIX, 2000 paragraphs 3.126, 3.127). The maximum likelihood estimation for this model has more recently been implemented in the AD Model Builder software by Dr Tuck. The AD Model Builder executable that implements Tuck et al. (2003) in order to estimate fish stocks where appropriate mark-recapture and catch data are available has kindly been made available to the Working Group by Dr Tuck. This paper describes the required inputs and the derived outputs for this implementation.

Assessments of long-term sustainable yield using the Generalised Yield Model (GYM) and the setting of total allowable catches (TAC) for Patagonian Toothfish, Dissostichus eleginoides, for the Heard Island Plateau region (Division 58.5.2) are based on data obtained from annual random stratified trawl surveys (RSTS) and from targeted commercial fishing operations via an observer program. Evaluation of both the current RSTS design for Division 58.5.2 and some options for future modifications of the design is carried out in a simulation environment using the program Fish Heaven with data analysis carried out using Splus. Fish Heaven is a simulation program that uses a spatially-explicit, age-structured fish population dynamics model. Recruitment, movement, mortality, growth, and fishing (both commercial and research) were simulated for the Patagonian Toothfish for the Heard Island Plateau region. Census of fish stocks by age class were compared to population estimates based on simulated RSTS hauls. For a series of 10 years of survey, age 4 recruitment was estimated using 111 RSTS stations, assuming the age of every fish caught in the surveys is known without error and assuming knife-edge fishing selectivity at age 4, using either (a) the age 4-alone RSTS results or (b) age 4 to 8 RSTS results using back-projections from each age to age 4 for each of 6 cohorts.

This document seeks support to allow line setting operations by autoline vessels fishing in Division 58.5.2 to occur at any time of the day/night cycle. The proposal forms part of an adaptive approach to management, which considers the risk status of the fishery, knowledge on the effectiveness of mitigation measures, mitigation performance record of the vessel, seabird mortality levels and assessment of the likely effects of individual mitigation measures to total mitigation response. Since the introduction of longline fishing in 58.5.2 in 2002 seabird by-catch mitigation requirements have exceeded those required by CCAMLR. Evidence from sub Area 48.3 (South Georgia), where both the hooking effort and number and abundance of longline-vulnerable seabird species is far greater than in 58.5.2, suggest that winter fishing with appropriate mitigation presents very low risk to seabirds. This is supported by the results of the first two years longline fishing in 58.5.2: a total of 2.2 million hooks have been set and no seabirds caught during line setting operations. Possible reasons why seabirds have not been caught are the very low abundance of longline-vulnerable seabird species on the fishing grounds between May and September, night setting, the requirement for a minimum line sink rate, the use of paired streamer lines and no offal discharge. In cases where seabird mortality is very low or non-existent it is appropriate to review mitigation requirements, particularly measures that might have adverse effects on fishing operations. Evidence from integrated weight longline experiments conducted in the New Zealand ling fishery in daylight and in summer suggest that the absence of seabird mortality in 58.5.2 is due to the low incidence of longline-vulnerable seabirds in winter, the minimum line sink rate and the use of streamer lines (the contribution to by-catch reduction of a minimum line sink rate and streamer lines, used separately, has not been determined). Removal of the night setting requirement is unlikely to result in an increased risk to seabirds in Division 58.5.2.

The difference in the number of white-chinned petrels (Procellaria aequinoctialis) and sooty shearwaters (Puffinus griseus) killed by unweighted longlines (UW; mean sink rate to 20 m depth: 0.11 m/s) and longlines containing 50 g lead/m integrated weight (IW; mean sink rate: 0.24 m/s) was examined in the New Zealand ling (Genypterus blacodes) autoline longline fishery. The experiments were conducted in the seabird breeding seasons (October/November) of 2002 and 2003 on F/V’s Janas and Avro Chieftain. White-chinned petrels and sooty shearwaters are more difficult to deter from baited hooks than the seabird species occurring in Divisions and Subareas where night setting exemption currently applies, and are considered a worse-case scenario from a risk assessment perspective. In the experiments the Janas and Avro Chieftain were followed by large number of seabirds, including 200-1,400 white-chinned petrels and up to 400 sooty shearwaters. Lines were set in day-light and at night, and a single streamer line was deployed on all sets. Differences between IW and UW lines in the catch rates of white-chinned petrels, sooty shearwaters and albatrosses, and ling and non-target fish species were determined for each pair of magazines or lines by counting dead birds during line hauling. In 2002 80 white-chinned petrels were caught on UW lines compared to one white-chinned petrel on IW; the reduction in mortality was 98.7% (CL95% = 99.8-90.6%; P

Surveys of all known breeding sites of wandering, black-browed and grey-headed albatrosses were carried out at South Georgia in the 2003/04 breeding season. Wandering albatrosses were censused by ground counts, and black-browed and grey-headed albatrosses mainly by yacht-based digital photography and subsequent counting on computer screen using Adobe Photoshop software. In total, an estimated 1,553 pairs of wandering albatrosses, 75,500 pairs of black-browed albatrosses and 47,800 pairs of grey-headed albatrosses were breeding at South Georgia in the 2003/04 season. Compared to results from a predominantly yacht-based survey of black-browed and grey-headed albatrosses over the whole of South Georgia conducted in the mid 1980s, numbers of these two species appear to have decreased by 26% and 14%, respectively. However, comparison of annual totals for Bird Island, which was censused on both occasions mainly by ground counts, indicate more rapid declines: 4.0% p.a. from 1989/90-2003/04 for black-browed albatrosses, and 2.9% p.a. from 1990/91-2003/04 for grey-headed albatrosses. Due to the lower accuracy of the methods used in the 1980s, it is likely that the Bird Island figures are more indicative of the population trends at South Georgia. The decline in wandering albatrosses is even more pronounced: 30% (1.8% p.a.) since the previous survey in 1984. The magnitude of these population decreases is alarming, given the long time span involved and consistent downward pattern. Of particular concern is the acceleration since 1997 in the rate of decline of wandering albatrosses at Bird Island, which now stands at 4.5% per year. Unless these long-term declines can be halted or reversed, there must be some doubt over the long-term viability of the breeding populations of these species of albatrosses at South Georgia.

Age determinations estimated by means of otoliths and age classes derived from CMIX analysis have been compared in Champsocephalus gunnari using material collected around South Georgia during four Russian surveys in 1988, 1990, 2000 and 2002. Considerable differences in age estimates between the two methods were revealed. A number of suggestions including an Age Determination Worksop to be held in Kaliningrad in 2005 and the exploration of methods in addition to CMIX are made to reconcile some of these differences in the future.

The paper presents the application of the bootstrap method to estimate accuracy of mixture distributions parameters. The method allows to estimate statistical characteristics not only component densities but another parameters: mean length of components, their standard deviations or equation parameters to determine these standard deviations. It is possible to estimate parameters correlation and bias.
The method has been applied to data from UK survey in 2002 that WG-FSA-03 used for toothfish recruitment calculations in S. Georgia area. The accuracy of mean length of component is high (CV~0.04), but total densities have CV~0.3 – 0.5. CV of parameters of linear equation is more then 1.0. It was marked high correlation between some parameters. Standard errors of densities exceed the values calculated by CCAMLR program.
The method can be used to determine the number of trawl stations in a survey to get the recruitment estimates with given accuracy.

This paper presents bycatch information for the Australian fisheries in Divisions 58.5.2, 58.4.2 and 58.4.3, with estimates of total removals by fishing ground for the 2002/2003 and 2003/2004 seasons. Bycatch in the trawl fisheries was low, approximately 1 to 2% of the total catch (target plus bycatch). Bycatch in the longline fisheries was higher, ranging from 8% to 14% of the total catch. Much higher percentage bycatch rates occurred in longlining grounds where the fishing effort and therefore target catch was low. The main bycatch species were skates and grenadiers in the D. eleginoides fishery and skates and C. rhinoceratus in the C. gunnari fishery. The total skate catch in Division 58.5.2 (including those cut from longlines) was 43 tonnes in 2002/2003 and 55 tonnes in 2003/2004. Results from the skate tagging program in Division 58.5.2 indicate that skates move very little between release and recapture, even after extended periods at liberty. The distances travelled ranged between 0.2 to 7 nm with periods at liberty ranging from 208 to 822 days. Growth rates estimated from recaptured skates were 20mm per/year in total length, 21mm per year in disk width and 0.14 kg per year in weight. Recapture rates were 2.5% for B. eatonii and 0.8% for B. murrayi for skates tagged in the trawl fishery and 0.05% for skates tagged in the longline fishery.

During tagging programmes for Dissostichus eleginoides at Heard and Macquarie Islands, fish have been treated with strontium to mark their otoliths for validation of subsequent age estimation. Otoliths from 69 fish that were at liberty for at least 350 days were eventually examined. Results indicate that current methods for ageing D. eleginoides are appropriate: the number of observed annuli on the otoliths after the strontium mark was consistent with the time at liberty of the fish after marking.
Current and future work on age validation is designed to address outstanding problems with age determination in this species: investigating the accuracy of ageing small fish ( 1000 days; and developing techniques to analyse the daily marks at the centre of the otolith in order to pinpoint the position of the first annulus.