Impact of landscape on spatial genetic structure and diversity of Coenagrion mercuriale (Zygoptera: Coenagrionidae) in northern France

Loss and fragmentation of habitat is a current main cause of biodiversity loss in freshwater habitats. Odonates (dragonflies and damselflies) depend on these habitats to complete their development. Fragmentation may be a particular threat for odonates because it generates a network of small habitat patches within which populations could suffer from isolation and loss of genetic diversity. The southern damselfly Coenagrion mercuriale is categorized on the IUCN red list as Near Threatened, largely because of population fragmentation and demographic declines associated with changes in land use. Small populations at the margin of this species' range are of particular concern because they would be prone to detrimental effects of habitat fragmentation if this species were a poor disperser. We sampled C. mercuriale in 16 habitat patches (localities) at 4 main sites in the department of Pasde-Calais in northwestern France to quantify factors that affect dispersal and genetic diversity. Specimens were genotyped at 12 microsatellite loci to quantify genetic diversity, genetic differentiation, and the potential effect of landscape variables on genetic differentiation, and to detect any potential source-sink structure. Habitat separation had a limiting effect on dispersal by C. mercuriale, resulting in 3 main genetic clusters and weak divergence at the main site of Vallee de la Course. Genetic differentiation was low in each main site, implying that the localities within sites were connected at scales of up to similar to 2 km, albeit with some evidence for isolation at the more isolated localities. Given the degree of isolation of some areas and a lack of apparent genetic mixing in the intervening populations, any movement among the most distantly separated sites must have occurred some time ago. We identified barriers to dispersal, such as woodland, but detecting an unambiguous effect of certain variables, such as urbanization, was difficult because many landscape features were highly correlated.