Corridors with good-quality habitats maintain the spatial dynamics of metapopulations by promoting dispersal between habitat patches, potentially buffering populations and communities against continued global change. However, this function is threatened by habitats becoming increasingly fragmented, and habitat matrices becoming increasingly inhospitable, potentially reducing the resilience and persistence of populations. Yet, we lack a clear understanding of how reduced corridor quality interacts with rates of environmental change to destabilise populations. Using laboratory microcosms containing metapopulations of the Collembola Folsomia candida, we investigate the impact of corridor quality on metapopulation persistence under a range of simulated droughts, a key stressor for this species. We manipulated both drought severity and the number of patches affected by drought across landscapes connected by either good or poor-quality corridors. We measured the time of metapopulation extinction, the maximum rate of metapopulation decline, and the variability of abundance among patches as criteria to evaluate the persistence ability of metapopulations. We show that whilst drought severity negatively influenced the time of metapopulation extinction and the increase in drought patches caused metapopulation decline, these results were mitigated by good quality corridors, which increased metapopulation persistence time and decreased both how fast metapopulations declined and the inter-patch variability in abundances. Our results suggest that enhancing corridor quality can increase the persistence of metapopulations, increasing the time available for conservation actions to take effect, and/or for species to adapt or move in the face of continued stress. Given that fragmentation increases the isolation of habitats, improving the quality of habitat corridors may provide a useful strategy to enhance the resistance of spatially structured populations.

Microcosms hosting the soil Collembola metapopulations were set up at week 1 and then sampled weekly by image analysis processes. The count data only includes the total number of individuals in each landscape and the standard deviation between patches. 

R studio/ R/ Excel