13Large-scale Transportation Infrastructures (LTIs; roads, railways, etc.) are among the main de-14 terminants of landscape fragmentation, with strong impacts on animal dispersal movements and 15 functioning of metapopulations. Although the detection of the impacts of LTIs is now facilitated 16 by landscape genetic tools, studies are often conducted on a single specie, while it is acknowledged 17 that different species might react differently to the same obstacle. Multi-specific approaches are 18 thus required to get a better overview of the impacts of human-induced fragmentation, especially in 19 landscapes crossed by multiple LTIs whose impacts can accumulate. We surveyed two vertebrates 20 species (the grass snake Natrix helvetica and the midwife toad Alytes obstetricans) and two insect 21 species (the butterfly Maniola jurtina and the ground-beetle Abax parallelepipedus) in a landscape 22 fragmented by six LTIs: a motorway, a railway, a country road, a gas pipeline, a power line and a 23 secondary road network. Using multiple linear regressions and commonality analyses on both clas-24 sical and hierarchical genetic distances computed over reduced spatial scales, we showed that 38% 25 of the overall explained genetic variability across all species was due to LTIs. While the butterfly 26 was seemingly not impacted by any LTI, the genetic structure of the three ground-dwelling species 27 was mostly influenced by roads, motorway and railway. LTIs, and especially roads, mostly acted as 28 barriers to gene flow, barrier effects accounting for 85% of the overall variance in genetic distances 29 explained by LTIs across species. Although the power line did not affect any studied species and the 30 1 gas pipeline only impacted gene flow in the ground-beetle through forest fragmentation, other LTIs 31 systematically affected at least two species. Importantly, we showed that some LTIs could some-32 how promote gene flow, embankments probably providing favourable habitats for vertebrate species.
33Considering the high variability in species response to LTIs, we argue that drawing general conclu-34 sions on landscape connectivity from the study of a single species may lead to counterproductive 35 mitigation measures and that multi-species approaches are to be more systematically considered. 36