Context Land use and land cover (LULC) change is a major part of environmental change. Understanding its long-term causes is a major issue in landscape ecology. Objectives Our aim was to characterise LULC transitions since 1860 and assess the respective and changing effects of biophysical and socioeconomic drivers on forest, arable land and pasture in 1860, 1958 and 2010, and of biophysical, socioeconomic and distance from pre-existing forest on forest recovery for the two time intervals. Methods We assessed LULC transitions by superimposing 1860, 1958 and 2010 LULCs using a regular grid of 1 9 1 km points, in a French Mediterranean landscape (195,413 ha). We tested the effects of drivers using logistic regressions, and quantified pure and joint effects by deviance partitioning. Results Over the whole period, the three main LULCs were spatially structured according to land accessibility and soil productivity. LULC was driven more by socioeconomic than biophysical drivers in 1860, but the pattern was reversed in 2010. A widespread forest recovery mainly occurred on steeper slopes, far from houses and close to pre-existing forest, due to traditional practice abandonment. Forest recovery was better explained by biophysical than by socioeconomic drivers and was more dependent on Electronic supplementary material The online version of this article (

Questions: Land use legacies in current forest understorey vegetation, thoroughly studied in temperate regions, were investigated in a Mediterranean context. We tested the effect of three historical variables on current forest plant communities and traits: forest temporal continuity (ancient: forested before 1860, recent: reforested after 1860, and very recent forest: reforested after 1958) and type of land use in 1860 and 1958 (forest, pasture or arable land). Location: The Regional Natural Park of Luberon (southeastern France). Methods: We used a comprehensive vegetation plot database (473 species in 1,429 plots). Species' response to historical variables was tested with logistic regressions, and the relationship between plant traits and historical variables was analysed with RLQ and fourth-corner analyses. Results: Among all studied species, 250 responded to forest temporal continuity, 208 to 1860 land use, and 246 to 1958 land use. Species associated with ancient forests were more frequently forest specialists or forest edge species, shade-tolerant and perennials, while species associated with recent and very recent forests were more frequently annuals, anemochorous and heliophilous species. Species exhibited different traits and ecological preferences according to the type of land use prior to forest: therophytes were more frequent on former arable land while chamaephytes were more frequent on former pasture. Trait responses to 1860 and 1958 land uses were globally consistent. Conclusions: The effect of forest temporal continuity and past land use on forest understorey communities was consistent with other studies in northern Europe or northern America, which suggests that the same ecological processes apply in temperate lowland and Mediterranean regions. This study highlights a succession of plant communities in the long term and different trajectories of succession according to the type of former agricultural use. The long-term legacies of past land use in current 2 of 13 |

HAL is a multidisciplinary open access archive for the deposit and dissemination of scientific research documents, whether they are published or not. The documents may come from teaching and research institutions in France or abroad, or from public or private research centers. L'archive ouverte pluridisciplinaire HAL, est destinée au dépôt et à la diffusion de documents scientifiques de niveau recherche, publiés ou non, émanant des établissements d'enseignement et de recherche français ou étrangers, des laboratoires publics ou privés.

QuestionsHow does connectivity affect animal-dispersed plant assemblages in woodlots of agriculturedominated landscapes? Is this effect dependent on zoochorous dispersal modes? Location Long Term Socio-Ecological Research (LTSER) site of "Zone Atelier Armorique" (ca. 150 km 2 ), Brittany (Western France) Accepted Article MethodsWe sampled 26 small post-agricultural woodlots embedded in an agricultural matrix (1-7 ha).We assessed the connectivity of each woodlot using habitat reachability metrics (i.e., Proportion of Total Potential Dispersal Flux in the Landscape) calculated along a range of dispersal distances from 100 to 1500 m. This metric was based on graph theory, and was calculated as a function of landscape matrix permeability to animal movement (i.e., seed plant dispersal vectors). We analysed the composition of plant assemblages in the core and margins of woodlots. We calculated the proportion of species richness and cover of zoochorous species within these assemblages and for specific zoochorous modes (i.e., epi-, endo-and dyszoochorous species). We analysed the effects of Proportion of Total Potential Dispersal flux at different dispersal distances and woodlot size on the functional dispersal structure of the plant community using linear models. ResultsWe demonstrated that overall plant zoochorous assemblages only depend on woodlot size in the range of dispersal distances tested. The proportion of richness of zoochorous species within the community increased with habitat size. Connectivity influenced only the relative proportions of the different zoochorous modes, except for the dyszoochorous group, which was primarily influenced by woodlot size. Both epizoochorous and endozoochorous species were impacted by the interaction between connectivity and woodlot area. Connectivity increased the proportion of epizoochorous species richness and the proportion of endozoochorous cover in assemblages in big woodlots. These effects were detected at short dispersal distances. Our results were supported in both total and core plant assemblages. ConclusionThis study indicates that both habitat isolation and decrease in habitat size contribute towards shaping the functional structure of the plant community, but they act at different levels of dispersal mode. Our results suggest that in these fragmented landscapes long-distance dispersal has been lost either because of the selection of small-sized animal movements or of the rareness of plant species needing large habitat size. Maintaining woodlot connectivity in the woodland surroundings seems to be sufficient to preserve short-distance dispersal though a more thorough restoration of landscape connectivity is necessary to recover the functional composition of the community with the whole set of dispersal patterns.