A central aim in population biology is to discern the relative contribution of intrinsic (density-regulated) and extrinsic (environmental) factors to fluctuations in population size and demographic composition, with increasing emphasis placed on quantifying the complex interplay between the two [1-4]. The mounting number of long-term ecological studies available for the measurement of population dynamical parameters, although still relatively rare, is providing a more refined understanding of the combined effects of these mechanisms [4-8]. For instance, investigating the relationships between population density, environmental conditions and survival probability using mark-recapture techniques has provided important advances in this regard [e.g., [9-11]].
Given that populations of large, long-lived mammals tend to have a relatively low capacity for growth due to their long generation times and low reproductive output , it is hypothesized that intrinsic factors should regulate growth only near carrying capacity. Indeed, there is good evidence that this is the case in many large mammal species [8,12,13], with many studies concluding that extrinsic factors are the predominate drivers of change when populations are below carrying capacity [12,14,15]. However, the complex relationships that exist between extrinsic and intrinsic control mean that there is no species for which there is a complete understanding of how abundance is regulated over the complete range of population densities . Another bugbear is that many populations with a high degree of age-dependent fecundity and mortality may not reveal density dependence if the time series used in the investigation is short relative to generation time [4,12,17]. In practice, it is usually only possible to examine the combined effects of density and environmental conditions through measurements made over long periods spanning an array of population densities or levels of food availability. As such, there are only a few case studies where this has been done for long-lived mammals, and most of those have focussed on island populations of ungulates [2,18,19].
Changes in the population size of large marine predators is potentially indicative of larger ecosystem changes given that their predominate regulator appears to be environmental stochasticity influencing food availability over vast oceanic foraging regions (e.g., [20-22]). Upper trophic-level marine predators such as seabirds and seals are particularly amenable to the examination of such mechanistic hypotheses because they are easily monitored during their obligatory onshore breeding phase . Access to such rare datasets is particularly important given the predictions of climate change over the next few decades [22,23], and recent evidence for broad-scale changes in population trends in birds and mammals throughout, for example, the Southern Ocean [24-29].
The well-documented population decline and possible recent stabilization of one of the most wide-ranging Southern Ocean predators, the southern elephant seal (Mirounga leonina) at Macquarie Island, has been the focus of intensive demographic studies for over fifty years [22,26,30]. Population censuses from the 1940s to the present and capture-mark-recapture studies from the 1950s and 1990s have provided extensive demographic data for this population at both low and high population densities [22,26,30-32]. There is strong evidence that this population responds to environmental stochasticity via modifications to individual survival given that this parameter is highly sensitive to the at-sea foraging conditions experienced by an individual over its predominately aquatic life cycle [22,32]. Foraging elephant seals breeding at Macquarie Island range widely over millions of square kilometres of the Southern Ocean [21,33,34], and it has been established that their feeding areas are associated to some extent with the pack ice zone and colder sea surface temperatures – these environmental conditions are known to fluctuate with El Niño-Southern Oscillation (ENSO) patterns [22,32,33]. In this region, ENSO follows an approximate seven- to eight-year cycle during which time ocean productivity can fluctuate substantially [35,36] (Fig. 1B).
It has been suggested that changes in ocean conditions affect southern elephant seals either directly by modifying the availability of food resources, or indirectly by affecting sea ice dynamics and hence, ocean productivity . While some studies have shown that survival , weaning mass  and weanling sex ratio  are reduced or modified during El Niño conditions, the Macquarie Island population of elephant seals has shown a consistent positive relationship between El Niño and pup survival . These contrasting relationships may arise from the different climatic conditions associated with ENSO events in different regions of the Southern Ocean .
There is also some evidence for density regulation in southern elephant seal populations, mainly via space limitation on land while breeding [42-44]. However, competition for food at high population densities may also occur during the at-sea foraging phase [22,45-47]. In this paper we expand greatly on previous work by amalgamating capture-mark-recapture data collected over two extended periods of differing population density at Macquarie Island: (1) the years between 1951 and 1960 when the population was relatively abundant, and (2) between 1993 and 1999 when it was approximately 50 % smaller. Our main aim was to identify whether there is evidence for density and environmental effects on survival rates and how these mechanisms combine to explain the observed phenomenological trends of population size over the last 50 years. We achieve this by (1) assessing the concurrent age-, and sex- specific survival at the two different density levels, (2) testing for density dependence in adult and first-year survival between and at both densities, and (3) testing for the effects of environmental variation as represented by ENSO on adult and first-year survival.