Abstract
Systemic conservation planning attempts to identify priority conservation areas. Bringing together tools and theories of several fields of research in ecology, the outcomes of the systemic planning of the conservation reflect the uncertainties inherent to the different stages of its process. This thesis proposes new approaches to address several methodological challenges related to various sources of uncertainty in systemic conservation planning.First, we propose a new conceptual framework for integrating uncertainties related to species distribution data into the systemic conservation planning process. The optimal conservation solutions, relative to the distribution scenarios, are identified through integers linear programming in, and take into account, through a post-selection approach, the variability of the species distribution models predictions throughout the process. This approach avoids the tacit trade-off between flexibility and efficiency of conservation solutions.Secondly, we provided a methodological background for optimizing three measures of functional richness in a set of reserves, thanks to integer linear programming. The differences between the corresponding conservation solutions highlight a source of uncertainty related to the definition and operationalization of the functional richness. At the origin of this uncertainty, a functional space built on different assumptions as to the calculations of functional distances between species.Third, we identified a source of uncertainty inherent to the species-area model fitting, and its impact along the systemic conservation planning process. We have shown that the application of one particular model may not provide reliable predictions for all habitats, which affects the estimation of conservation targets. Depending on the model used, the set of reserves selected is either ineffective or overestimated for habitat protection, resulting in a waste of conservation resources or inefficiencies in protecting biological resources. We then suggest performing a multi-model inference to provide robust habitat-specific conservation targets.