Conflicts between humans and large marine predators are increasing around the world, often resulting in excess mortality of animals without satisfactory results in terms of effectiveness (to reduce risk) and respect for conservation issues (regarding threatened species). The risk management response in aquatic environments often consists of non-selective removal of animals, such as for sharks after human fatalities, based on the assumption that the risk is correlated with their density. In our study, we propose to test an alternative hypothesis based on the existence of problem individuals (PIs) -demonstrated in terrestrial environments- in two species (a toothed cetacean and a shark) in which parental guidance diverges significantly. This hypothesis of PIs is based on the existence of divergent temperaments and personality traits such as boldness-shyness (management of novelty) and risk-taking-avoidance (management of danger) which would be transmitted from one generation to another. With an ethological approach, the existence of these behavioural traits will be tested on killer whales in Antarctica and on two behaviours: i) the hunting technique of voluntary stranding on beaches to predate elephant seal juveniles and ii) the depredation of hooked fish in the context of commercial longline fisheries. In parallel, two populations of bull sharks will be studied in Fiji in the framework of observation dives based on artificial feeding; the behaviours studied will be i) the attitude towards artificial food (novelty), ii) towards divers (danger) and between conspecifics (sociability and agressiveness). The observations will take place over two years in order to be able to i) discriminate individual divergent behaviours (Work package - WP1) and show ii) that they are consistant in time and iii) according to various situations (WP2). This fieldwork will be accompanied by DNA sampling of monitored animals. WP3 will consist in demonstrating the heritability of personalities between generations in the bull shark (RAdseq analyses on n=60 individuals) and in killer whales (pedigree with n=30 individuals); we will also try to explore the molecular mechanisms on which this heritability could rely. WP4 will synthesize the results concerning the genesis of PIs (influence of genetics vs learning processes) within the two taxa and will conduct a socio-anthropological study on the perception by the various stakeholders of the human-large marine predator conflicts. These approaches should allow us to better understand the evolutional and survival issues for both taxa (fundamental science), but also to address societal issues (applied science), in terms of proposing improved risk and conservation management. The research consortium will be composed of four institutional partners, namely CRIOBE (EPHE-CNRS), CEBC (CNRS-ULR), MARBEC (IRD) and SENS (IRD), providing three research directors, three research fellows and two PhD students (not funded by ANR - sex-ratio F 33%) directly involved in the WPs. The project will be co-coordinated by two of the senior researchers with solid experience in managing international scientific projects. They will be supported by an external monitoring committee composed of four senior researchers with international expertise in shark ecology (Univ. of Hawaii), vertebrate ethology (Univ. Quebec in Montreal), genetics (CRIOBE) and human-wildlife conflict management (Norwegian Institute for Nature Research). The project plans to recruit two PhD students to work respectively on the collection of ethological data on killer whales in Antarctica and sharks in Fiji, an 18-month post-doc to support the genetic component and five Master2 students. A grant of 750 k€ is requested, 46% of which is allocated to the recruitment of students, 33% to the collection of data in Antarctica and Fiji and the rest (9% excluding management fees) essentially for genetic analyses, anthropological study, communication and publication costs.