Some upper trophic level predators, such as marine mammals, seabirds and large predatory fish have been shown to be useful sentinels for marine ecosystems. Through their demography, diet and foraging behaviour, ecosystem changes associated with climate change can be monitored. The local marine ecosystem at the Prince Edward Archipelago in the Southern Indian Ocean is heavily influenced by the position of the dynamic sub-Antarctic front (SAF) that varies inter-annually in terms of its influence on the local environment. If the SAF migrates southwards, as climate change models have predicted, the abundance of autochthonous and allochthonous prey found within the vicinity of the archipelago will decrease and increase, respectively. We investigated the diet of an inshore forager at the archipelago, the gentoo penguin Pygoscelis papua, as a possible sentinel to the expected changes in the prey assemblages in local waters, due to the southward migration of the SAF. We collected stomach content samples from gentoo penguins, using the water off-loading technique, monthly over three years (1994–1996) and then annually during 2012, 2014 and 2015. Prey assemblages within the diet was found to coincide with the proximity of the SAF to the Prince Edward Archipelago as well as the annual life cycles of prey. Due to the plasticity these penguins exhibit in their diet and foraging behaviour, we suggest that they are important sentinel species for the local marine ecosystem of the Prince Edward Archipelago. Regular sample collection from these birds, albeit stomach content samples and/or tissues for stable isotope analysis, should be collected so that variability in the ecosystem can be easily monitored.

The Falkland Islands currently supports one of the largest Southern Rockhopper Penguin (Eudyptes c. chrysocome) populations. Archipelago-wide censuses conducted in 2000 and 2005 revealed that the number of breeding pairs had declined by 30 % during this period. To establish whether the breeding population continued to decline, an archipelago-wide census was conducted in 2010. We report a conservative estimate of 319,163 ±SD 24,820 pairs breeding at the Falkland Islands in 2010. This represents a 51 % increase when compared with the number counted in 2005. A simple stochastic population model was developed to investigate the extent to which changes in demographic parameters between 2005 and 2010 could account for the increase in breeding pairs. The population model predicted a 38 % increase in the number of breeding pairs over a 5-year period (289,431 ±SD 24,615). The increase in the number of breeding pairs was therefore probably attributed to improved vital rates in the period between the 2005 and 2010 archipelago-wide censuses in combination with other factors such as a reduction in the proportion of adult birds that abstained from breeding. Based on the 2010 Falkland Islands estimate, the global population of the subspecies E. c. chrysocome is now closer to 870,000 breeding pairs of which the Falkland Islands accounts for approximately 36 %, the second largest proportion after Chile. We conclude that despite fluctuations, the number of Southern Rockhopper Penguins breeding at the Falkland Islands has increased over the last 15 years and suggest that the ‘Vulnerable’ conservation status of the species be re-assessed.

The fourth archipelago-wide census of Gentoo Penguins Pygoscelis papua breeding at the Falkland Islands was conducted from 24 October to 8 December 2010. The number of Gentoo Penguins breeding in 2010 was estimated to be 132 321 ± 2 015, the highest number of breeding pairs recorded for this species at the Falkland Islands since the first survey in 1933. The global population of Gentoo Penguins is conservatively estimated to be about 384 000 breeding pairs, of which the Falkland Islands accounts for 34%, probably the largest component of the global population. Annually monitored study colonies accounted for 20% of the total number of Gentoo Penguin breeding pairs at the Falkland Islands in 2010 and proved to be a reliable proxy for archipelago-wide changes in the number of breeding pairs. Recent trends at annually monitored study colonies, combined with archipelago-wide trends, indicate that the number of Gentoo Penguins breeding at the Falkland Islands has increased between 2005 to 2010. However, annual monitoring data also revealed large inter-annual variability in the number of breeding pairs, which makes assessing systematic population changes challenging.

Fisheries-related mortality of globally threatened albatrosses and petrels in the Falkland Islands is largely attributed to collisions with warp cables on stern trawlers. Following the introduction Tori lines, also known as bird-scaring lines, in 2004, incidental mortality has been substantially reduced. However, there is still a requirement for further development and optimisation of mitigation methods. In this paper, we present results of sea trials to compare the efficacy of a modified Tori line design with the initial (2004) design. The modified design resulted in 28 and 33% fewer warp contacts than the 2004 design for all species, and high-risk species (black-browed albatross and giant petrels), respectively. Surprisingly, these differences were not found to be statistically significant. However, the modified design was more effective at reducing interactions with the Tori lines themselves. The modified design provided better coverage and protection of the warps, particularly in crosswinds, and was safer for the crew to use. Our study highlights the importance of continued research and development in the field of bycatch mitigation and has relevance for trawl fisheries globally.

Marine Ornithology 40: 129–133 (2012) Seabirds of the order Procellariiformes are particularly well adapted to strong winds, which they use to facilitate long-distance foraging for patchily distributed food resources (Davies et al. 2010). On land, albatross colonies are often distributed to maximise exposure to the prevailing winds, thus making it easier for birds to get airborne. At sea, most seabirds either use storm conditions to aid their movements, or avoid severe weather effects by flying away from a storm (e.g. Culik et al. 2000, Catry et al. 2004). It is when seabirds are on land, at their breeding sites, that they are most vulnerable to the effects of extreme weather events. Severe weather conditions can affect seabird EFFECTS OF A STORM ON COLONIES OF SEABIRDS BREEDING AT THE FALKLAND ISLANDS

Dassen and Robben islands are approximately 50 km apart, and currently support the largest and third largest populations of African penguins Spheniscus demersus respectively. At both islands, moult is the most synchronised and seasonal activity of the annual cycle. The main difference in moult phenology between the two islands is the degree of synchronisation. The peak moult period at Dassen Island is broader (less synchronous) than at Robben Island. At both islands, juvenile penguins moulted less synchronously than adults. Juvenile moult was less synchronous at Dassen Island than at Robben Island. Interannual variation in moult phenology was similar at Robben and Dassen islands. Island-level moult phenology was more synchronous in the year following the Treasure spill than was the case after the Apollo Sea spill. This was attributed to the larger number and proportion of birds that were affected by the Treasure spill. The temporal duration of the disruption to moult phenology was also greater following the Treasure spill, especially at Robben Island. Moult counts provide a relatively reliable method of estimating the size of the adult population at Robben Island, but not at Dassen Island where the large number of birds moulting away from the shoreline results in the population size being significantly underestimated. Although the number of active nest sites continued to increase up until 2004 at both islands, the numbers of adult moulters counted declined by 55% between 2003 and 2005 at Robben Island, and by 50% between 2002 and 2005 at Dassen Island. The decrease in the number of moulters coincided with a decreased availability of food in the region.

The breeding and moult phenology of African penguins Spheniscus demersus at Dassen Island form part of a variable annual cycle. Between 1994 and 2006, African penguins bred throughout the year. Most of this period was characterised by favourable feeding conditions. There were definite peaks in breeding activity, but these varied between years and between colonies on the island, with no obvious seasonal pattern. Synchronisation of breeding activities was greater in colonies with higher nest densities, suggesting that breeding phenology was influenced more by social interaction between individuals than by a wider-scale seasonal effect. Moult phenology of African penguins at Dassen Island was more synchronous and seasonal than breeding activities. Most penguins moulted between September and January. A large proportion (30%) of moult records was of birds moulting in nest sites rather than at landing beaches along the shore. The incidence of nest moulting was greatest from August to October, when sea conditions were generally rougher. The interval between successive moults ranged from 221 days to 546 days, with a mean of 349 days. The much greater degree of synchronisation in moult compared with breeding patterns suggests that moult, and not breeding, may represent the key driver of the annual cycle.

South Africa is a global hotspot for oil pollution. The regional oiled seabird cleaning centre, the South African Foundation for the Conservation of Coastal Birds (SANCCOB), has handled over 50 000 seabirds from its inception in 1968 until 2005. The majority of seabirds oiled in South Africa are African penguins Spheniscus demersus, followed by Cape gannets Morus capensis, both of which are classified as Vulnerable to extinction. On the basis of the proportion of the population that has been affected, the African penguin is considered to have suffered more from oiling than any other seabird species globally. The rehabilitation success (proportion of birds known to have survived for at least one month in the wild) and restoration success (proportion of rehabilitated birds attempting to breed) of de-oiling penguins and gannets are higher than has been reported for any other species. The financial costs of de-oiling African penguins are substantially lower than the costs of de-oiling seabirds in the Northern Hemisphere. De-oiling contaminated birds is thus a valuable conservation intervention for these species, both of which are relatively localised in areas within or close to major shipping routes and ports, where a single spill can threaten a large proportion of the global population. There are, however, long-term effects of oiling on penguins and gannets. De-oiled gannets survive slightly less well than un-oiled birds, but the difference is similar to inter-colony differences in survival. Approximately 27% of rehabilitated African penguins are unable to breed following their release. In addition, oiling has a long-term negative impact on the breeding productivity and cost of reproduction in de-oiled birds. The primary objective should therefore be to prevent or reduce oil spills in the first place. However, future oil spills are inevitable and the authorities need to ensure that they have plans in place and the required capacity to respond rapidly to spills when they do occur. One of the ways to reduce the number of penguins becoming oiled during a spill is to evacuate birds from the affected area. The continued capture and cleaning of penguins and gannets that do become oiled is justified on conservation grounds. Thus, de-oiling should be a twin objective to prevention in South Africa's oil spill management strategy, and every effort should be made to further improve both of these aspects. Keywords: AFRICAN PENGUINCAPE GANNETDASSEN ISLANDDE-OILEDMORUS CAPENSISOIL SPILLSREHABILITATIONRESCUERESTORATIONSANCCOBSPHENISCUS DEMERSUS

The bulk ore carrier Apollo Sea sank south-west of Dassen Island off western South Africa in June 1994, oiling approximately 10 000 African penguins Spheniscus demersus, most of which were collected from Dassen Island. A total of 4 076 de-oiled penguins was released with flipper bands. From 1994 to 2005, follow-up research using re-sighting and capture-mark-recapture methods indicated that about 73% of the de-oiled penguins observed back at Dassen Island attempted to breed, and were thus successfully restored into the breeding population. For de-oiled breeders, the median interval between their first recorded sighting and first recorded breeding attempt was 11 months, indicating a short-term delay in restoration. At least 45% of the de-oiled breeders were still being re-sighted five years after their release, and a minimum of 4% survived into their ninth year. These results represent the most successful restoration estimates anywhere in the world. The proportion of de-oiled juvenile penguins re-sighted back at Dassen Island and recorded breeding was lower than that of birds in adult plumage. De-oiled non-breeders spent significantly more time along the shore and less time within breeding colonies than de-oiled breeders. The mean proportion of de-oiled breeders that abstained from breeding each year during the study period was greater than expected. There was a negative relationship between breeding and subsequent survival and breeding, suggesting a cost of reproduction for de-oiled birds. Keywords: AFRICAN PENGUINAPOLLO SEABREEDINGCAPTURE-MARK-RECAPTUREDASSEN ISLANDDE-OILINGOILINGREHABILITATIONRESTORATIONSOUTH AFRICASPHENISCUS DEMERSUS

The reproductive success of African penguins Spheniscus demersus at Dassen Island from 1994 to 2000 was variable, but much higher than previously reported figures for the species. Breeding success was positively related to the abundance of anchovy Engraulis encrasicolus and sardine Sardinops sagax, and the high reproductive output during the study was attributed to the large biomass and high availability of these two species for much of the study period. De-oiled penguins from the Apollo Sea spill had a slightly lower hatching success but a significantly lower overall breeding success than that of un-oiled birds, driven mainly by the reduced number of chicks which fledged. Nests with two de-oiled Apollo Sea parents were less successful than nests with one de-oiled bird. There was increased mortality of chicks 40 days and older in nests with de-oiled birds. Chicks from nests with one de-oiled Apollo Sea parent grew at a similar rate to chicks from nests with no de-oiled parents. However, chicks from nests that comprised two de-oiled Apollo Sea birds had significantly slower growth rates than these other two groups. Breeding success and chick growth at nests with de-oiled birds were more negatively impacted when feeding conditions were less favourable. These results suggest that one of the main reasons for lower breeding success in de-oiled birds was their reduced ability to provision chicks, especially during the period in which the energy demands of the chicks is greatest. The rate of mate fidelity was lower in de-oiled birds than un-oiled birds, and there was a positive association between mate fidelity and breeding success.

From 1997 to 2005, the distribution of sardine Sardinops sagax, an important prey item for four seabirds off South Africa, shifted 400 km to the south and east, influencing its availability to breeding birds. It became progressively less available to seabirds in the Western Cape Province, where the number of African penguins Spheniscus demersus breeding decreased by 45% between 2004 and 2006, survival of adult penguins decreased and penguins established a new eastern colony in 2003. In that province, the number of Cape gannets Morus capensis breeding decreased by 38% between 2001/2002 and 2005/2006 and the contribution of sardine to the diet of gannets fell from an average of 40% during the period 1987-2003 to 5-7% in 2005 and 2006. The proportions of Cape cormorants Phalacrocorax capensis and swift terns Sterna bergii breeding in the south of the province increased as sardine moved south and east. In the Eastern Cape Province, the number of penguins breeding halved between 2001 and 2003, whereas after 2002 there was an increase in the number of Cape gannets that bred and in the contribution of sardine to their diet. It is likely that in that province sardine became increasingly available to gannets but remained beyond the shorter feeding range of penguins. Management measures that may mitigate the impacts on seabirds of an unfavourable, long-term change in the distribution of their prey include the provision of breeding habitat where prey is abundant, spatial management of fisheries competing for prey, and interventions aimed at limiting mortality.

The reproductive success of African penguins Spheniscus demersus at Dassen Island from 1994 to 2000 was variable, but much higher than previously reported figures for the species. Breeding success was positively related to the abundance of anchovy Engraulis encrasicolus and sardine Sardinops sagax, and the high reproductive output during the study was attributed to the large biomass and high availability of these two species for much of the study period. De-oiled penguins from the Apollo Sea spill had a slightly lower hatching success but a significantly lower overall breeding success than that of un-oiled birds, driven mainly by the reduced number of chicks which fledged. Nests with two de-oiled Apollo Sea parents were less successful than nests with one de-oiled bird. There was increased mortality of chicks 40 days and older in nests with de-oiled birds. Chicks from nests with one de-oiled Apollo Sea parent grew at a similar rate to chicks from nests with no de-oiled parents. However, chicks from nests that comprised two de-oiled Apollo Sea birds had significantly slower growth rates than these other two groups. Breeding success and chick growth at nests with de-oiled birds were more negatively impacted when feeding conditions were less favourable. These results suggest that one of the main reasons for lower breeding success in de-oiled birds was their reduced ability to provision chicks, especially during the period in which the energy demands of the chicks is greatest. The rate of mate fidelity was lower in de-oiled birds than un-oiled birds, and there was a positive association between mate fidelity and breeding success. Keywords: AFRICAN PENGUINAPOLLO SEABREEDING SUCCESSCHICK GROWTHDASSEN ISLANDDE-OILEDMATE FIDELITYSPHENISCUS DEMERSUS

The bulk ore carrier Apollo Sea sank south-west of Dassen Island off western South Africa in June 1994, oiling approximately 10 000 African penguins Spheniscus demersus, most of which were collected from Dassen Island. A total of 4 076 de-oiled penguins was released with flipper bands. From 1994 to 2005, follow-up research using re-sighting and capture-mark-recapture methods indicated that about 73% of the de-oiled penguins observed back at Dassen Island attempted to breed, and were thus successfully restored into the breeding population. For de-oiled breeders, the median interval between their first recorded sighting and first recorded breeding attempt was 11 months, indicating a short-term delay in restoration. At least 45% of the de-oiled breeders were still being re-sighted five years after their release, and a minimum of 4% survived into their ninth year. These results represent the most successful restoration estimates anywhere in the world. The proportion of de-oiled juvenile penguins re-sighted back at Dassen Island and recorded breeding was lower than that of birds in adult plumage. De-oiled non-breeders spent significantly more time along the shore and less time within breeding colonies than de-oiled breeders. The mean proportion of de-oiled breeders that abstained from breeding each year during the study period was greater than expected. There was a negative relationship between breeding and subsequent survival and breeding, suggesting a cost of reproduction for de-oiled birds.

Off South Africa, anchovy Engraulis encrasicolus and sardine Sardinops sagax are the main prey of African penguins Spheniscus demersus. The combined spawner biomass of these fish increased from less than one million t in 1996 to more than nine million t in 2001 and then decreased to four million t in 2005. The combined biomass of young-of-the-year of these species increased from 0.2 million t in 1996 to 3.2 million t in 2001 before failing to 0.4 million t in 2005. There was a large eastward shift in the distribution of sardine between 1999 and 2005. The number of African penguins breeding in the Western Cape Province increased from 18 000 pairs in 1996 to more than 30 000 pairs from 2001 to 2005 before failing to 21 000 pairs in 2006, as the availability of fish decreased near breeding localities. Numbers of penguins breeding and numbers of birds in adult plumage moulting were significantly correlated with the young-of-the-year biomass of anchovy and sardine and with the available biomass of spawning sardine. The increase in the number of penguins breeding was mainly attributable to a greater proportion of birds breeding and improved breeding success. The decrease probably resulted from high mortality. Delayed first breeding and abstinence from breeding during periods of food shortage may both increase survivorship when food is scarce and enable seabirds rapidly to take advantage of improved feeding conditions. Although long-lived seabirds are buffered against short-term variability in food supplies, environmental change that influences the abundance and availability of prey can have severe consequences for central-place foragers, such as penguins, if there is long-term displacement of prey to regions where no suitable breeding localities occur.

The bulk ore carrier MV Treasure sank between Robben and Dassen islands off western South Africa in June 2000. Although releasing less oil than the Apollo Sea spill six years earlier, the number of African penguins Spheniscus demersus oiled (approximately 19 000) in the Treasure spill was almost double the number oiled in the Apollo Sea incident. The majority of oiled penguins was collected from Robben Island. The capture, stabilisation and transport of oiled penguins following the Treasure spill were substantially more successful than during the Apollo Sea rescue operation and resulted in an increase in the percentage of oiled penguins that were released back to the wild. In addition to the 19 000 oiled penguins that were caught, 19 500 un-oiled birds (12 345 from Dassen Island) were evacuated to Cape Recife, 800 km to the east, to prevent them from becoming contaminated. The evacuated birds returned rapidly to Dassen Island and many resumed breeding soon after their return. The restoration success (proportion of birds observed at Dassen Island that were recorded breeding) was substantially greater for evacuated birds than for de-oiled birds. Lightly oiled birds that were released without being cleaned had lower survival rates and were less successfully restored into the breeding population than de-oiled birds, especially if more than 5% of their body was covered in oil. Reproduction exerted a greater cost to de-oiled birds than to evacuated birds. A total of 1 787 orphaned chicks was hand-reared at two separate facilities on the mainland and were released at Robben and Dassen islands. The chicks did not necessarily return to the colony at which they were released. The survival rates and restoration success of chicks reared at the two facilities differed, probably as a result of one group being overfed. Despite the death of about 2 000 African penguins in the Treasure spill, and the substantial disturbance that resulted from the rescue operation, the penguin colonies at Robben and Dassen islands increased by 18% and 26% respectively in the year following the spill. This increase was attributed to a significant increase in the biomass of anchovy Engraulis encrasicolus and sardine Sardinops sagax in 2001. These findings highlight the importance of preventing oil spills in the first place, and in the event of a large oil spill, the conservation value of evacuating un-oiled penguins to minimise their risk of becoming contaminated.

From 1987 to 2005, numbers of African penguins Spheniscus demersus breeding in South Africa's Western Cape Province increased by about 50%. Numbers decreased at the four northernmost colonies in the region: Lambert's Bay and the three colonies in Saldanha Bay, although at Jutten Island the decrease is inferred from an estimate for 1987, derived from interpolation. Numbers also decreased at Geyser Rock and Dyer Island on the South Coast. At five colonies between Saldanha Bay and Dyer Island there were large increases. At a sixth colony in that region, Seal Island, where Cape fur seals Arctocephalus pusillus pusillus limit breeding space, numbers remained stable. At two colonies that were initiated in the early 1980s, Robben Island and Boulders, increases were initially rapid (>20% per annum) and matched growth of the South African stock of sardine Sardinops sagax. Strong growth at Dassen and Vondeling islands, between Robben Island and Saldanha Bay, was observed from about 1996–2002, when there was a large increase in the biomass of pelagic fish off South Africa. Increases at colonies between Saldanha Bay and Boulders slowed after 2002, whereas the colony at Dyer Island stabilised at that time. In 2003, a new colony was initiated east of Dyer Island at De Hoop Nature Reserve. These latter trends followed an eastward shift in the distribution of sardine. Small penguin colonies may act as foci for growth in a period when the distribution of prey is changing. Hence, it is important that they be maintained, especially those that, if lost, would increase the isolation of regional populations. Some of the small colonies are less susceptible to oil spills than colonies in the proximity of harbours, and for that reason also are important.

Of over 20 000 African penguins Spheniscus demersus that had been flipper-banded as chicks between 1978 and 1999, 2% of those re-sighted after fledging settled to breed at non-natal colonies. This represented 14% of the banded birds that were subsequently recorded breeding. Only one of these immigrants had previously been recorded breeding at its natal colony, the rest presumably being first-time breeders. The largest proportions of banded chicks that emigrated came from Dyer Island on the south coast of South Africa, all of which settled at colonies to the west or north. Penguins emigrating from Namibian breeding colonies either relocated to the Western Cape of South Africa or settled at colonies farther to the north in Namibia. Emigration and immigration of African penguins are thought to be driven by changes in the distribution and availability of their prey. Eight penguins that were banded in adult plumage were found to have attempted breeding at more than one locality. All were survivors of the Apollo Sea oil spill of 1994 and had been cleaned and released by the Southern African Foundation for the Conservation of Coastal Birds. This is thought to be the first documented evidence of attempted breeding by African penguins at more than one locality.

The direction, distance and seasonal patterns of movements of African penguins Speniscus demersus between breeding colonies were investigated for birds that had been flipper-banded between 1970 and 1998. These were comprised of 3 986 penguins banded as chicks at nests, 1 672 banded as adult birds at breeding colonies and 4 691 adult penguins that had been oiled, sick or injured and subsequently rehabilitated. An analysis of observations of flipper-banded penguins in KwaZulu-Natal, Eastern Cape, Western Cape and Northern Cape provinces of South Africa and in Namibia, showed that 35% of birds banded as chicks and 9% of those banded as adults had visited mainland sites or breeding colonies, other than the bird's own natal or breeding colony. Penguins banded as chicks at Eastern Cape colonies and sighted within two years of banding exhibited a marked clockwise movement around the coast. Penguins from all regions that had been banded as chicks and subsequently found dead also showed a clockwise pattern of movements, whereas live re-sightings of birds banded as chicks in the Western Cape or at Namibian colonies did not reveal a significant directional trend in their movements. Penguins banded as adults at Eastern Cape colonies moved in a clockwise direction around the coast but there was no obvious pattern shown by adults banded at colonies in the other regions or of rehabilitated birds. Most re-sightings of penguins banded as chicks were made in spring or summer, but there was no seasonal pattern associated with the re-sightings of penguins banded as adults. Distances travelled by penguins banded as chicks averaged greater than for those banded as adults. Juvenile birds usually return to their natal colony, but it is evident that some young birds have the capacity to settle and breed at non-natal colonies.

The number of eastern rockhopper penguins Eudyptes chrysocome filholi breeding at subantarctic Marion Island decreased from about 173 000 pairs in 1994/95 to about 67 000 pairs in 2001/02. During 1994/95-2002/03 pairs fledged on average 0.40 chicks per year, an amount thought insufficient to balance mortality of breeding adults, and there was a decrease in the mass at arrival at breeding, colonies of both males and females. Except in 1997/98, the mass of chicks at fledging was less than that recorded at two other localities. These factors suggest an inadequate supply of food for rockhopper penguins at Marion Island. Decreases of rockhopper penguins at several other localities also have been attributed to inadequate food. Rockhopper penguins at Marion Island continued to feed mainly on crustaceans during chick rearing. There was a marked increase in the contribution of fish to the diet in 1999/00 that coincided with an increase in mass at arrival Lit colonies of both males and females. Trends in numbers of pairs breeding in different sections of Marion Island were not always consistent, indicating the need for island-wide monitoring to establish the overall trend.

The numbers of gentoo penguins Pygoscelis papua breeding at subantarctic Marion Island fell by 40% from 1994/95 to 2002/03, from 1 352 pairs to 806 pairs. Apart from a slight increase in 1998/99, there was a steady decrease in numbers breeding between 1995/96 and 2000/01, when the population stabilized. There is indication that in some years not all breeders nested and that some birds relocated to another colony after disturbance. From first clutches, pairs on average fledged between 0.01 chicks in 1997/98 and 0.58 chicks in 2002/03 (mean 0.38 ± 0.21). In 1994/95, replacement clutches increased the overall production of fledged chicks by 11%. Based on demographic parameters measured at other localities, the production of chicks at Marion Island was inadequate to maintain the population during the period 1995/96 – 2000/01. Consistency in trends in breeding success at five colonies suggests that factors operating at a mesoscale, rather than those specific to particular colonies, often influenced breeding success. Laying was later than normal in 1997/98, when there was almost total breeding failure with large losses of eggs and small chicks to returning Subantarctic skuas Catharacta antarctica. Future research on this Near Threatened species at Marion Island must take full account of its susceptibility to human disturbance.