Few studies were conducted on fish genetic structure in the North-Western Indian Ocean, an area under strong climatic influence, characterized by the alternating monsoon seasons. The particular geography and climate of Soqotra Island is responsible of the presence of wadi and "Khor", which are often separated from the sea by gravelly, sandy berms and roads. Thus, in Soqotra the connectivity between brackish and marine water could be temporally limited. The aim of the present study is to better understand 1) the genetic structure of estuarine fishes on Soqotra and 2) the connectivity between estuarine and marine habitats. Such information could contribute to the first stage for future estuarine conservation plans around Soqotra Island. This study focuses on the grunter Terapon jarbua locally known "Habraham" on Soqotra Island. This species inhabits marine and brackish waters as adults and juveniles respectively, khors playing the role of nurseries. Although this fish is not of particular economical interest on Soqotra, its omnipresence in almost all khors of the region make it a suitable model species to study genetic structure and habitat connectivity. The population structure of T. jarbua is described by analysing sequence variations in a fragment of the mitochondrial gene Cytochrome c Oxidase subunit I (COI) and the genotyping of nine microsatellites in 288 individuals from 10 locations around Soqotra Island and the south coast of Yemen mainland. COI analyses stressed some genetic heterogeneity within this region, with three locations significantly different from the rest. Considering the variability of the microsatellites, this study detected a significant genetic differentiation between estuaries, where geographical distance was not a structuring factor for the populations. Indeed some close genetically related populations were separated from each other by several hundred kilometres while significantly different populations were separated by less than a few kilometres. This spatial structure might be the result of stochastic environmental variables (i.e.: currents, khor opening, mortality) on estuary recruitment, postlarvae settlement and juvenile survival. Thus some nurseries might be randomly dominated by particular subpopulations while others might be colonized by a possible larval pool stemming from different subpopulations. Those hypotheses were later confirmed by micro-chemical studies of the otoliths (earbones). The results demonstrate that it will be difficult to select just a few estuarine sites for conservation as their value in terms of its nursery function will vary in time and space according to the climatic conditions. It is rather recommended that basically all estuaries should enjoy at least a basic level of conservation management.