This document outlines the concept and rationale for movement to an instrument-based (moorings and gliders) program of oceanographic and ecological observations and research to support the U.S. commitment to CCAMLR and ecosystem science in the Southern Ocean. The U.S. AMLR Program's historical ship-based approach is no longer economically feasible. An instrument-based research program, in lieu of a ship-based research program, can reduce costs while still achieving NOAA’s mission in the Antarctic.

Unmanned Aerial Vehicles specially adapted for operation in polar regions were used to study distribution and population size of three Pygoscelis penguin species on King George Island, Chabrier Rock and Penguin Island (Subarea 48.1) in three austral summer seasons (2014/15, 2015/16, 2016/17). During these three expeditions 32 successful photogrammetry missions (total distance 3963,42 km) were performed. The total flight time of the missions was 40 hours and 56 minutes. Flights covered the area of 34.51 km². Images were taken with digital SLR Canon 700D, Nikon D5100, Canon T5i 700D with a 35mm objective lens. Flights altitude at 350 – 550 m AGL (Above Ground Level) allowed images to be taken with a resolution GSD (ground sample distance) of 3-6 cm. Obtained images were used for the estimation of population size. Precision estimation of individuals’ count was based on the photo interpretation of high-resolution Red-Green-Blue and one Blue-Green-NIR imagery, as well as automatic interpretation using ImageJ software. In selected colonies a comparison with the results of measurements taken at the same time from the ground level were done. Obtained images enabled us to locate and estimate the number of nests of: Adélie, gentoo and chinstrap penguins. In total approx. 30 000 nests in twelve breeding colonies were identified in 2016. The use of UAV for monitoring of indicator species enabled data acquisition from areas inaccessible by ground methods. Results will support the basic knowledge about the population size and distribution of krill-depended indicator species in Subarea 48.1.

In the 2013-14 reproductive season, the 26,000 pairs of Adélie penguins from Pétrel Island, Terre Adélie, failed to fledge their chicks. This catastrophic breeding failure resulted from two factors. First, the season saw an extreme sea-ice extent that forced Adélie parents to cover great distances to reach the open water during the incubation and early chick-rearing phases. The extended absence of parents meant that the partner and/or the chick fasted longer at the nest, increasing the risk of desertion by the partner and the risk of dying of prolonged fast for the chick(s). Second, the season was characterized by unusual precipitations of rain around the turn of the year that killed massively the already weakened young chicks. While we haven’t seen yet the consequences of this “zero year” on the demography of the population there, a second complete breeding failure took place in the season 2016-17, leading to the deaths of all the chicks of the 28,000 Adélie pairs that bred in this season. Once more sea-ice extent was the major cause of failure. Sea-ice conditions in 2016-17 were even more drastic than in 2013-14, as sea ice extended as far as 80 km from the colony but over the entire breeding period. Snow precipitations occurred only over 2-3 days in mid-November 2016 and several nests were buried under more than a meter of snow, but the snow falls were immediately followed by almost a week with temperatures continuously above 4°C that sent flows of melting snow through the colony, wiping out nests and eggs. Two “zero years” occurring within three breeding seasons will definitely affect the demographics of the population at Pétrels Island. Furthermore, the situation is not going to improve as long as the B9B giant berg stays anchored in Commonwealth Bay, on the east of the island. The B9B blocks icebergs from the Mertz glacier in front of Pétrel Island. This is a serious threat to the populations of Pétrel Island and a strong incentive for the creation of the D’Urville Sea/Mertz Marine Protected Area in the EARSMPA.

To facilitate access to spatial data for use in analyses and generating graphics (i.e. maps) in R the CCAMLRGIS R package has been upgraded to provide direct access to polygon data that can be viewed on the online GIS. This package can also be used to create polygon data that can either be used directly in R or be output as shapefiles that can be loaded into other programs and/or back onto the online GIS.  In this paper we provide information on the data format requirements for generating polygon data and explain how polygon data on the online GIS can be loaded into R. 

This study used an operating model to simulate a Dissostichus eleginoides population in Research Block 5843a_1 with a fixed exploitation rate and various tagging rate scenarios to assess the impact on the numbers of tagged fish recaptures and accuracy of tag-based estimates using the Chapman estimator. The results showed that tag-based biomass estimates obtained from a low numbers of recaptures will have a high risk of over-estimating biomass and that for a given number of tagged fish available a minimum numbers of recaptures are required to ensure a low likelihood of biomass overestimation. For Chapman-estimates to achieve a 25% level of accuracy at least 50% of the time, required a minimum of 6 recaptures when there were approximately 270 tagged fish available for recapture. To increase the chance of achieving the same level of accuracy to 80%, a minimum of 20 tagged fish recaptures were required when approximately 815 tagged fish were available.

The minimum number of recaptures required to achieve a specified level of accuracy will vary depending on the initial population size, the exploitation and tagging rate and the consequent number of tagged fish available. Simulations of toothfish populations could be performed for all Research Blocks to assess the data requirements to obtain Chapman estimates with a high likelihood of being within a desired level of accuracy. Based on these simulations, the Working Group consider developing criteria such as a minimum number of annual tag-recaptures required in each Research Block to determine when it is appropriate to use a tag-based estimate of biomass.

This study has used bootstrap methodology to quantify the uncertainty in toothfish biomass estimates from Chapman mark-recapture and the CPUE-by-seabed area methods in Research Blocks in Subareas 58.4 and 48.6. The Chapman mark-recapture point estimates and their confidence intervals were generally much larger than the point estimates and confidence intervals from the CPUE-by-seabed area biomass estimates.

For Patagonian toothfish in Subarea 58.4 the confidence intervals of both methods overlapped in most years, providing some confidence in the estimates from the two different methods. The large differences in Antarctic toothfish biomass estimates using the two different methods suggests that the underlying assumptions currently made in one or both of the methods are likely to be invalid.

We recommend that Working Group consider reviewing assumptions in the CPUE-by-seabed area and the Chapman biomass estimation methods at the Research Block scale and prioritise resolving issues in Research Blocks that show the largest discrepancies. 

This paper reports the preliminary results of by-catch analysis as a part of progress report for the research fishery of Dissostichus spp. in Division 58.4.3a and 58.4.4b by Japan and France during 2012/13–2016/17. We have used extracted data provided by CCAMLR secretariat on 2^nd June, 2017.

Although the by-catch occurred in most longline research operations, the number of by-catch individuals was less than 50 individuals in most of the operations. In most of longline operation, by-catch weight was less than 100 kg and less than 20 % of total catch. GRV and ANT are common by-catch species, which are caught at every research block and every year. RFA is also common by-catch species, but mainly fished at the research block 58.4.3a.

The results of the GLM and its model selection with AIC criteria indicates that the number of by-catch individuals is highest at the research block 58.4.4b_1 for GRV and ANT. As temporal change of by-catch, common three by-catch species represents obvious decrease in number of by-catch individuals. The results of GLMs also imply the influence of fishing method and gears on the by-catch pattern.