INTRODUCTION Ultramafic (serpentine) soils typically host a dis- tinctive flora and vegetation, largely affected by the edaphic and physical characters of these soils (Brooks 1987; Baker et al. 1992; Roberts & Proctor 1992). In Tuscany, central Italy, serpentine vege- tation has been investigated over a long period and with different approaches (e.g. Messeri 1936; Pichi Sermolli 1948; Chiarucci et al. 1998b). The appar- ent lack of vegetation dynamics on these soils has been ascribed to the negative effects of soil metal. Recent studies provide evidence of a higher content of exchangeable metals in the soil under- lying the later successional stages, suggesting that factors other than soil metals are limiting. Chiarucci et al. (1998a, 1998b) hypothesized that the physical conditions present at a site (e.g. slope, solar heating) and the nutrient content of the soil have a greater influence on the species com- position than the negative affect of soil metal content. These studies were carried out with explo- rative aims and the sampling sites were placed in areas with physiognomically homogeneous plant cover. Although obtained with subjective sam- pling, the relationships were found at different spatial scales: 4 m 2 plot as grain and a single site as extent by Chiarucci et al. (1998a); 100 m 2 plot as grain and the whole Tuscany as extent by Chiarucci et al. (1998b). Here we use an objective sampling method to test the hypothesized relative strength of the relationship between vegetation composition and environmental factors and a finer spatial scale to reduce the effects of within-plot heterogeneity. Large parts of the serpentine outcrops of Tuscany have been planted with pines. These plantations have a negative affect on serpentine vegetation, by modifying the plant community structure and species composition, in particular by reducing the abundance of endemic species in favor of common and/or more competitive species (Chiarucci & De Dominicis 1995; Chiarucci et al. 1996, 1998a). The spread of pines introduced in the last few decades in the Upper Tiber Valley threaten grasslands Ecological Research (2001) 16, 627–639 A test of vegetation–environment relationship in serpentine soils of Tuscany, Italy Alessandro Chiarucci,* Duccio Rocchini, Claudio Leonzio and Vincenzo De Dominicis Dipartimento di Scienze Ambientali, Università di Siena Via P.A. Mattioli 4, 53100 Siena, Italy The present study evaluates the relative importance of environmental factors in affecting the species composition and abundance of the plant communities on ultramafic soils in Tuscany, Italy. We used rigorous sampling techniques to test hypotheses generated from exploratory studies performed previously. Vegetation–environmental relationships were analyzed using 50 plots, each 1m 2 , ran- domly located throughout a 22-ha area in the Upper Tiber Valley. We confirm that the exchange- able fraction of nickel in the soil is almost never high enough to affect the vegetation. However, physical factors (e.g. substrate setting and elevation) are important in controlling the distribution of plant species. Tree cover (almost exclusively due to the introduced plantation pines) also had a significant affect on the vegetation composition and on soil features such as the C/N ratio. Other important factors significantly related to the gradients in vegetation composition (e.g. rockiness and total soil nitrogen) are interpreted as factors related to the vegetation composition through a posi- tive feedback mechanism. Key words: endemic plants; monitoring; positive feedback; restoration; ultramafic soils. *Author to whom correspondence should be addressed. Email:chiarucci@unisi.it Received 28 February 2001. Accepted 27 June 2001.