Species from all major taxonomic groups are experiencing major population declines due to human-driven pressures, including habitat destruction, climate change, pollution, overexploitation, and invasive species. Islands are especially vulnerable, as they host many endemic species with small populations and limited genetic diversity, making them highly sensitive to environmental disturbance. To protect endangered species, conservation actions often focus on the most vulnerable populations through protected areas, population reinforcement, translocations, and invasive species control. Among these strategies, the eradication or control of invasive alien species (IAS) is widespread on islands. Success rates have improved in recent years, even on remote or technically challenging islands. However, removing invasive small mammalian predators remains difficult, as eradication outcomes depend on many interacting factors: technical feasibility, cost, ecological impact, social acceptability, risk of reinvasion, and species behaviour. Trapping efficiency depends heavily on ecological knowledge such as animal density, home range, and movement patterns. Behavioural traits such as trap avoidance can reduce effectiveness, especially in mustelids and other carnivores. In inhabited or poorly isolated areas, reinvasion from surrounding regions can also compromise success, particularly where short dispersal corridors exist, such as land connections.

This context is especially relevant for the White-breasted Thrasher (Ramphocynclus brachyurus), a bird strictly endemic to the Caravelle peninsula in Martinique, a French Caribbean island. The species occupies about 1,000 hectares and is listed as endangered. One of its main threats is predation by the small Indian mongoose (Urva auropunctata), an introduced predator known to prey on eggs, chicks, and probably fledglings and adults. To assess whether mongoose extirpation could be considered a realistic management option, we evaluated different scenarii of control design on simulated mongoose populations by combining field data on the movement of female mongooses and litterature on their population dynamics. We captured and equipped 19 individuals from dry forest patches of the peninsula with GPS devices during the dry season, when control measures are most efficient for reducing population growth. Of these, 16 produced sufficient data to estimate their movement parameters and home ranges. These results were then mobilised to simulate the movements of a hypothetical population of females across the peninsula, to evaluate how trap spacing, trapping duration, and individual behavior regarding traps could influence eradication success. The costs associated with the different successful trapping designs were calculated to determine optimal strategies regarding cost-benefit ratio.