Abstract
The temporal dynamics of a population are governed by variations in its demographic parameters (survival, reproduction and migration). To understand and predict the trajectory of a population, it is therefore crucial to identify the factors driving these variations, particularly in the current context of biodiversity loss and climate change. Compared with other species that are easier to study, large carnivores have rarely been the subject of detailed demographic studies, even though they are often threatened. The aim of this thesis was to investigate the intrinsic (age, size, etc.) and environmental causes of variations in reproductive rates in a large carnivore threatened by climate change, the polar bear. I adopted an evolutionary demography perspective to inform hypothesis formulation and interpretation of results.I found that age is a major determinant of reproductive success in polar bears, influencing every stage of the reproductive cycle (from entry into a maternity den, to cub survival). Most of the reproductive rates studied showed an increase in early life and a decline in late life (i.e. senescence). All in all, middle-aged females had a much higher probability than young females and a slightly higher probability than old females of successfully completing a reproductive cycle. This suggests that females make most of their contribution to the next generation during a few mid-life years.Using a capture-recapture (CR) model combined with path analysis, I showed that some females were able to invest more resources in reproduction by producing heavier litters, composed of more cubs. Although larger litters are heavier, the individual cubs are lighter than those from smaller litters, and therefore have a lower chance of survival. Nevertheless, it is adaptive for females capable of allocating a lot of resources in reproduction to have larger litters, as this leads to more cubs reaching their second year.I also built a multi-event CR model that makes use of information on the reproductive status of females obtained from bio-logging and bio-telemetry instruments (these instruments allow us to know when females are overwintering in a maternity den). Thanks to this information, I was able to estimate for the first time the probability of denning, and the probability of the litter surviving until spring. I could test the existence of a reproductive cost on future reproduction (an important theoretical expectation of evolutionary demography), as well as the impact of sea ice availability on reproductive rates. It appears that sea ice loss in Svalbard reduces the probability of denning, the probability that cubs survive until their first spring, and the probability that they survive their first year, whereas litter size has remained stable.In a final analysis, I tested whether an earlier date of emergence from the maternity den mediates the negative effect of climate change on the probability of cubs surviving to spring. To do this, I extended the previous model by combining it with a path analysis. I was able to show that following years with low sea ice availability, females leave their dens early, probably because they have depleted their energy stores, which reduces the survival prospects of their cubs.My thesis therefore identified age as a major determinant of reproductive success in polar bears, and revealed the previously cryptic impact of sea ice loss on this population. In addition, my thesis illustrates the benefits of combining CR and path analysis, and of combining datasets, to gain a detailed understanding of the mechanisms producing variations in reproductive rates.