Unexpected long-range edge-to-forest interior environmental gradients

We examined the relationships between distance-to-edge and environmental factors inferred from mean plant indicator values across large distance-to-edge and patch size gradients. Floristic composition, landscape metrics and site variables (climate, soil and forest management) were collected on 19,989 plots in 1,801 forest patches in Northern France using the French National Forest Inventory. Statistical models were applied to mean plant indicator values (MIV) from Ellenberg and Ecoplant databases for soil pH, soil nitrogen (N), soil humidity (F), light (L) and air temperature (Ta) using distance-to-edge and forest patch size as predictors. The five mean indicator values significantly varied with distance-to-edge and MIV pH, N and Ta decreased over distances in excess of 500 m. Consistent very long edge-to-interior gradients were also detected for site variables. The distance-to-edge effect remained significant after controlling for site differences, especially for MIV pH and N. Significant edge-to-core gradients of MIV were detected over much larger ranges than previously recognised. Neither the presence of an ecological boundary between forest and the surrounding matrix, nor microclimate, soil or forest management heterogeneity within forest patches can fully explain this long edge-to-interior gradient observed in MIV. Two hypotheses are discussed for MIV pH and N: (1) soil eutrophication, due to atmospheric N deposition, which could occur deeper into forest-cores than previously acknowledged; (2) land use legacies, as the periphery of ancient forests is more often occupied by recent forests where former agricultural practices have irreversibly modified topsoil properties. Land use history data would help identify the drivers underlying these long-range edge gradients.