Variation in vegetative and reproductive performance and leaf mineral composition among 25 populations of Colchicum autumnale (meadow saffron) from soils derived from six parent materials (limestone, marl, sandstone, greenstone, melaphyre and serpentine) in southwestern Poland has been investigated. The plant size (PS), total le-af area (TLA), leaf shape (LS), number of fruits per plant (NFP), number of seeds per plant (NFP), total weight seed per plant (TWSP) were estimated, and concentrations of seventeen elements (N, P, K, Ca, Mg, Na, S, Fe, Mn, Cu, Zn, Pb, Cd, Ni, Co, Cr, Mo) were analyzed in leaf and soil samples. In soil samples, also soil pH, organic matter content and sand, silt and clay content were determined. All soils (except melaphyre soil) contained elevated levels of Cr. Concentrations of soil Ni and Zn, Pb, Cd in serpentinite soil and polluted marl soils (respectively) were significantly higher than those of other examined soils. Meadow saffron leaves from all sites (except marl sites) contained elevated levels of Cr, Co and Ni. Statistical analysis, carried out with principal component analysis (PCA) revealed that a good correlation exists between the element content in leaves and plant performance traits and soil environment. The leaf Ca content and NFP were correlated and were much higher in populations from metal (except Cr) and nutrient-poor marl soils than in those from metal (except Ni in serpentine soil)and nutrient moderate rich soils derived from melaphyre, greenstones, serpentinite, limestones and sandstones and in those from metal-rich and nutrient-poor polluted marl soils. Meadow saffron plants tended to take up higher amounts of N, P, K, Mn, Cu and Na and lower amounts of S and had much higher TLA in populations from sandstone soils than those in populations from serpentinite, melaphyre, greenstone, marl and limestone soils. The leaf Co, Ni, Cr, Mg concentrations were correlated and were much higher in populations from serpentinite, melaphyre, greenstone, polluted marl and limestone soils than those from marl and sandstone soils. The pattern of variation in NFP and TLA across the different soil types was the opposite of that for leaf N, Cu, Na, Ni, Mg and S concentrations. Redundancy analysis (RDA) enables identification of the soil variables that best explain the variance pattern of plant response. The variation explained by the soil variables (15 soil elements and soil pH) was high (79%). The forward selection of soil variables identified soil Co, Mo, Ni, Ca and Zn concentrations as significantly influencing the ordination plant traits. The variation explained by these selected variables was 55%. Thus the five soil variables appeared to be the main factors determining the pattern variation of vegetative, reproductive and nutrient traits of Colchicum autumnale.
