The Raso lark is a Critically Endangered bird endemic to the islet of Raso, Cape Verde. This thesis investigates two phenomena that particularly put the species at risk: its extreme fluctuations in population size, and its potentially very low genetic diversity arising from small population size and severe past population contraction. More specifically, two chapters estimate year-to-year survival and explore the factors - environmental and individual - that influence it, while two other chapters examine the lark’s genetic characteristics compared to its two continental closest relatives, including phylogenetic relationships and levels of genetic diversity. The conclusion of the thesis then uses these results to make recommendations for the conservation of the Raso lark. Each of the data chapters is summarized below: Chapter 3 estimates adult survival in the Raso lark and tests whether it could be linked to two population phenomena observed in the field: a highly variable population size and a male-biased sex ratio in certain years. Using a dataset spanning 10 years, I estimated survival for both sexes to fluctuate between 0.76 and 0.94 over this period. This is much higher than the survival rate of its closest relative, the skylark. I also found strong evidence for survival fluctuating over time and differing between males and females (with males having higher survival until 2011, at which point the trend inverted), which could play a role in the aforementioned population size fluctuations and male-biased sex ratio, respectively. Chapter 4 aims at understanding which factors shape survival in the Raso lark. Two types of variables were considered: year-dependent (rainfall, population size, population mean clutch size) and individual-dependent (age, body size characters, size ratio with mate, Ase18 genotype). Amongst the year-dependent variables, only sameyear rainfall impacted survival, with a 13% decrease in survival in the wettest year compared to the driest year, making it the most likely explanation for the inter-annual fluctuations in survival found in Chapter 3. Results also hint at some of the individual factors - morphological measurements and Ase18 genotype - influencing survival. The picture that emerges is that of a species whose life history strategy is to invest heavily in maintenance and survival, but less into fecundity, which stands in sharp contrast with the mainland-dwelling skylark. This is consistent with the theory that island birds generally have slower life history strategies than their continental counterparts. Chapter 5 determines the precise relationship between members of the Alauda clade, resolving a node on the phylogenetic tree of all larks that the study by Alström et al. (2013) was unable to resolve. My RADseq results indicate that the Raso lark and the skylark are sister species, and that the Oriental lark is likely to be a subpopulation, or maybe a subspecies, of the skylark. Chapter 6 compares the population genetics of the Raso lark with those of the skylark. In particular, it estimates the genetic diversity of the Raso lark and investigates the drivers behind it. I found unexpectedly high nucleotide diversity in the Raso lark, and explain this by showing that the population contraction that the species underwent was recent enough for most of the diversity to still be present. Moreover, 16% of the Raso lark genome has levels of heterozygosity on average 6.6 times higher than elsewhere on the genome, likely due to suppressed recombination and the existence of a neo-sex chromosome in larks. Despite this, I found high levels of relatedness and of linkage disequilibrium in the Raso lark, two clear genetic signs that it underwent a severe population contraction several centuries ago.