Past landscape explains forest periphery-to-core gradient of understory plant communities in a reforestation context | Le paysage passé explique le gradient de périphérie à coeur de massif des communautés floristiques de sous-bois dans un contexte de reforestation

Aim. To disentangle whether long-range periphery-to-core gradient of forest understory plants could be attributed to past forest landscape and/or to current environmental filtering processes. We investigated (i) whether species response to past land use (ancient vs. recent forest) was consistent with species response to distance to present forest edge (core vs. periphery) (ii) what life-history traits explained plant response to distance to present edge and past land use (iii) whether distance to past forest edge explained current species distribution better than distance to present forest edge.Location. Temperate forests in the northern half of France (80,000 km²).Methods. Local climate, soil and stand characteristics, past land use, and present and past landscape metrics were collected on 11,936 plots using National Forest Inventory data and historical maps from 1831. Logistic regressions were applied to determine the response patterns of 181 species to present and past landscape, while controlling for local habitat quality (soil, climate and stand).Results. Species response to distance to present edge very well matched response to past land use. Plant traits related to colonisation capacity explained species response to present edge and past land use. The spatial distribution of 42 species was better explained by distance to forest edge in 1831, 37 species were better explained by distance to present edge and 24 species were better explained by distance to present edge and past land use.Main conclusions. Two mechanisms were responsible for the long-range periphery-to-core gradient: (i) past landscape and colonisation processes and (ii) present edge-related mechanisms. This suggests that plant community differences between periphery and core zones are thus related to dispersal limitation and not only to environmental filters. Our results underline the need to combine landscape ecology and history and have important implications for forest plant dynamics and conservation in the context of climate change.