Determining the factors that regulate populations through time can be complex because even the simplest nonlinear, density-dependent population models may exhibit a large range of complex dynamic behaviours [50,51]. While environmental stochasticity tends to inflate the variance in population size and demographic rates, negative density feedback at high population sizes has the opposite effect [2,4]. However, density regulation may become undetectable at lower population sizes  or when environmental conditions are favourable [12,52]. Using an extensive dataset collected from a long-lived mammal, we found that first-year survival varied as predicted with population density, but only when population size was relatively high and when models ignored indices of environmental stochasticity.
When the effects of density and environmental variation were examined together, the negative density feedback mechanism was apparently overwhelmed by the more dominant influence of stochastic environmental forcing. This observation underscores the importance of examining the competing and complex interaction between environmental control and density regulation over a large range of population sizes, especially in long-lived species susceptible to high environmental stochasticity .
These results suggest that density regulation in this system may operate when populations are at or near carrying capacity – a state where intra- and inter-specific competition for resources and intra-specific competition for mates is likely to be highest. This notion is consistent with data from studies investigating the dynamics of other long-lived mammals [12,13]. However, the magnitude of these effects is dwarfed by density-independent stochastic environmental conditions that affect food availability during a pregnant mother's foraging trip. Indeed, this forcing also overshadowed any negative influences on survival experienced during a naïve seal's first trip to sea, despite previous, albeit weak, evidence that environmental conditions during that period influence first-year survival . Although the phenomenological evidence for density dependence is pervasive across many different taxa , including the species under study, high environmental variation can sometimes mask even strong density dependence, especially if the effects are lagged [54-56]. Nonetheless, we found evidence for density regulation that would not have been detected using the low-density dataset alone, demonstrating the complex meshing of endogenous and exogenous forces in shaping animal population sizes .
Previous work has shown that many pinniped species demonstrate strong density dependence in various demographic rates and life history traits. [46,52,57-59], although these may be detectable only during poor-resource years . Density dependence in elephant seals has been shown to operate mainly during breeding where concentrated adult aggregations onshore can directly affect pup survival or the age at first reproduction [43,44,46,60] even though the exact form and strength of density dependence acting in this species is still a matter of some debate [22,47]. There is also ample evidence that pinnipeds demonstrate density-regulated somatic growth rates, with lower growth experienced at high population densities [57,61].
The incorporation of density effects into the models considered also revealed the dominant mechanisms by which environment stochasticity controls population abundance patterns over time. When environmental stochasticity (expressed as the SOI) was examined without the effects of density, the most parsimonious model predicted that the conditions during an individual's first year of life best explained variation in survival when population size was high (Table 4). However, when the effects of density were also included in the models, there was more evidence that the environmental conditions experienced by the mother when she was gaining body reserves that would eventually sustain her pup were most important (Table 6). Had we failed to consider density effects directly, we would have erroneously concluded the mechanism by which population density exerts its influence on dampening environmentally induced variation in life history traits. With density included in the model set, the pregnant mother's environmental context clearly emerged as the most dominant force in shaping her offspring's survival probability. This supports previous work suggesting that wean mass, an indirect expression of the mother's capacity to sequester sufficient resources prior to giving birth, was the most important determinant of first-year survival . However, unlike that previous study, our analyses add another piece to the puzzle by demonstrating the degree to which environmental stochasticity in the mother's foraging phase dominates intrinsic regulation.
We must also consider that the weak effects of population density on first-year survival are unlikely to capture the full mechanistic component of density regulation in this population, especially given the strong phenomenological evidence for density dependence in both relative-density periods. In addition to vital rates such as survival, density-dependent regulation may apply to other aspects of a species' biology, such as growth, behaviour, incidence of disease and distribution [57,62,63]. Eberhardt  proposed that the negative effects of increasing density on population growth are greatest in juvenile survival, followed in turn by the onset of puberty, fecundity and, finally, adult survival. In large mammals, density dependence is most commonly identified in vital rates that influence recruitment, in particular, juvenile survival, and less frequently in adult survival [4,12,19,65]. Indeed, there was evidence for density-dependent regulation in a small elephant seal population at Marion Island operating through changes to fertility , and it has also been shown that elephant seal population growth is highly sensitive to adult fertility . However, Pistorius & Bester  dismissed juvenile survival as an important driver of change in population growth at Marion Island. This may be explained by the relatively small population at Marion and direct evidence that a decline in the age of female primiparity has occurred there recently , suggesting that the dominant mechanisms driving the phenomenology of self limitation in the Marion and Macquarie Island populations may be different.
Our results have important implications for the assessment of environmental change in the Southern Ocean and Antarctic region. Given their status as upper trophic-level predators foraging over vast areas of the subantarctic and Antarctic oceanic zone, variation in abundance and life history parameters in this species may be indicative of larger changes occurring throughout the Antarctic ecosystem . The extensive demographic and population abundance data for the Macquarie Island population now span approximately five elephant seal generations , so these datasets consequently represent an invaluable source of information to determine long-term trends in this region. Elephant seal populations throughout the Southern Ocean have declined substantially over the last 50 years, although some populations are demonstrating recent stability or even recovery . Our results highlight the sensitivity of the species to long-term environmental fluctuations and argue for continued monitoring to determine the extent to which deterministic or oscillatory dynamics are affecting the region's higher predator guild.