• field margin
• Generalized Additive model
• Mediterranena
• Semi-natural vegetation

Although there is a growing scientific concern about the effects that landscape and land use modification may exert on agroecosystems, few papers have taken into account the effect of new types of land use, such as abandonment and urbanisation, on the composition of weed communities. Weed composition has a crucial role in determining the choice of agricultural management practices. Understanding the role played by abandonment and urbanisation in determining weeds species presence can help clarify the dynamics driving weed presence and abundance and the subsequent effects on agricultural management. This information may provide agricultural managers, land planners, and policy maker a useful tool to take decisions at both the farm and landscape scales.
We hypothesised that weed communities respond to landscape dynamics according to the ecological and biological characteristics of the inhabiting species. The use of cross functional response group was proposed to highlight these interactions. The method is based on the use of two or more functional response groups at the same time, to deepen the effect of environmental characteristics on flora composition.
Weed community dynamics were studied in 21 fields located along the coastal plain bordering Tuscany and Liguria (Italy). A land use map was compiled in 2008, soil samples were taken in 2009, field management types (recorded by farmers interviews) and field margin characteristics were recorded in 2009, and weed flora and vegetation samples were collected in 2009 and 2010. Soil characteristics were analyzed in 2010. A seed bank (seedling germination) trial was arranged in 2010, and was carried out until the beginning of 2011.
Multivariate analysis as Generalized Additive Model (GAM), Non metric Multi-Dimensional Scaling (NMDS), Permutational Multivariate Analysis of Variance were performed in order to clarify the effect of landscape modification on weed community composition. Species accumulation curves, Permutation test of multivariate homogeneity of groups dispersions, the main α-diversity indexes (Shannon, Simpson, Species richness) and Equitability were calculated to evaluate the landscape effects on α, β and γ diversity. Effects of farming practices, soil and field margin characteristics on standing vegetation, seed bank and field margin flora were analyzed and subsequently used as co-variables in the analysis of landscape effects on the vegetation.
Weed community composition and diversity were affected by soil characteristics (nitrogen availability, pH, texture and by activate calcium carbonate), by farming practices (weeding techniques, nitrogen distribution, crop sequences, green manure application and irrigation system), by field margin (especially by field margin weed composition) and by landscape parameters.
Urban sprawl and land abandonment resulted as highly connected, in particular old land abandonment (>20 years of patch abandonment) and urban sprawl were found significantly correlated. Land abandonment effects mainly depend on the succession stage of the patches. Both types of abandonment enhanced seed bank species richness. Recently abandoned patches presence at landscape level favoured Competitive-Ruderal Therophytes species, while old abandonment enhanced Competitive Geophytes groups of species. Fragmentation and urban sprawl, enhancing the presence of elongated not tilled areas at landscape scale, determined differentiated effects on connectivity, in accordance with different species attitude on patch persistence and on movement of species through the landscape. Competitive-ruderal, seed dispersed species found in a fragmented landscape a hostile matrix that reduced patch connectivity, while competitive species with vegetate dispersal mechanisms found in a fragmented landscape a more suitable structure to persist, and disperse.
Most landscape parameters had stronger effects on the field standing vegetation at the smallest scale (250 m buffer) while for the seed bank no differences between the various landscape scales were found. In accordance with our findings, field margins play a connection role between landscape and field flora dynamics.
The effect of landscape parameters was clearer when weeds were classified by functional response traits than by taxonomy, and the use of cross functional response groups resulted in most of the cases in better results than simple functional response trait groups.
In the present work the main landscape ecology theories were integrated in an agroecological approach, giving meaningful results regarding the interpretation of a real-world situation.