The growth-differentiation balance hypothesis (GDB), which postulates a physiological trade-off between growth and differentiation (morphological and chemical), has been tested almost exclusively for carbon-based secondary metabolites. Little attention has been paid to N-based compounds. In this study we aimed to test the predictions of the GDB hypothesis under field conditions for growth and pyrrolizidine alkaloid (PA) formation in Senecio vulgaris. We conducted a reciprocal transplant experiment at two sites differing widely in their nutrient supply. These included a conventionally managed vineyard (V) and a strip of local wild flowers between crop fields, which was established to promote species diversity in agroecosystems (C). No fertilizer or pesticides are allowed in such ecological compensation areas. In C, we expected lower growth but higher PA formation than in V. Due to differentiated selection regimes in the two habitat types with regard to nutrient (nitrogen) availability in the soil, we also expected different N-allocation patterns for the genotypes of the two collection sites. Plants of V produced more biomass and were taller than the plants of C. The relatively poor nitrogen conditions in C favored an earlier differentiation towards generative organs. In plants of C, higher concentrations of PAs were found than in plants of V. There exists a close negative correlation between growth and PA formation, indicating a trade-off. The origin of the plant material had only a little effect on PA formation. The observed phenotypic reaction of PA formation in S. vulgaris in the two habitats fits quite well the predictions of GDB theory. It is shown that this general response is overlaid by physiological factors leading to a pattern of PA accumulation, which is not readily predictable by nonmechanistic theories.