With the loss of global biodiversity, signals of the speciation process are progressively being erased as populations dwindle and species disappear. We propose to develop and implement an integrative approach that will draw from new advances in computational biology, field ecology, and population genomics. We anticipate that our approach will be broadly applicable to other non-model biological systems, especially those that diverged recently and rapidly and for which species boundaries are cryptic. Our findings, as well as in-country collaboration, will have immediate relevance to conservation policy in the ecologically complex and poorly studied regions of Southeastern Madagascar.