The distribution of the icefish Champsocephalus gunnari on the Kerguelen Plateau is confined to the inner shelves of Heard and Kerguelen islands and outlying banks where the water depth is less than 350 m. For fisheries management purposes, it is necessary to know whether fish from the various shelves and banks belong to one single or several separate populations. Analysis of the mitochondrial DNA of fish from various localities using restriction endonclease enzymes revealed no significant genetic heterogeneity between any of the localities. This does not preclude the existence of isolated stocks at the present time but indicates that the populations have been separated too recently for genetic isolation to be manifested, or that low-level interchange of fish between the populations may be maintaining genetic homogeneity.

In 1991 Chamsocephalus gunnari <1 year old were sampled at Shag Rocks on 18 January and off five bays on the north-east coast of South Georgia between 19 January and 12 February. Two separate size-groups of 31–46 mm and 55–87mm standard length (SL) were found at Shag Rocks. However, samples taken off Possession Bay, South Georgia, within two days of those at Shag Rocks, contained only one size-group of 25–57 mm SL. Samples taken at other locations at South Georgia also contained only single size-groups between 24–63 mm SL. The mean and median length of the large size-group at Shag Rocks are significantly larger (P<0.001) than those at South Georgia. For those at South Georgia there was an apparent growth of around 0.76% SL d−1, which is comparable with previous reports of their growth rate during the first year. Based on this growth rate, the differences between the average size of the larger size-group of C. gunnari at Shag Rocks and those at South Georgia indicates that they hatched at periods separated by about three months. This and other published evidence suggests there may be separate populations of C. gunnari at Shag Rocks and South Georgia.

Enzymatic polymorphism was used to detect variability within Notothenia rossii from two sites on the Kerguelen Plateau and Champsocephalus gunnari from the same sites and the South Orkney Islands. No polymorphism was found in the second species and it was low but not statistically significant in the first. This apparent homogeneity does not substantiate suggestions from other results that the populations can be separated, especially in the case of C. gunnari. Other approaches will be necessary to solve definitively the question of population separation.