We postulated that soil nutrient heterogeneity arises not only through physical and biological processes in the soil, but also through emergent rocks diverting precipitation containing nutrients to the surrounding soil. To test this idea-which we call the 'funnelling effect' of such rocks-we placed ion-exchange resin in small boxes beside rocks and in open soil on a pristine glacial forefield site in Switzerland, and measured the amounts of NH4 (+), NO3 (-), NO2 (-) and PO4 (3-) that were adsorbed. We also placed resin bags beneath PVC funnels of different sizes so that we could calibrate the natural funnelling effect of rocks. We obtained strong linear relationships between nitrogen (N) adsorbed and rain-collecting area of both rocks and funnels. Although the mean rain-collecting area of rocks was only 0.02 m(2), mean N adsorption was around 10 times higher within 1 cm of rocks than further away. In contrast, phosphorus (P) was not concentrated beside rocks, so that N:P stoichiometry varied spatially. Rumex scutatus and Agrostis gigantea plants that rooted beside rocks had significantly higher foliar N concentrations than those growing further away, in line with the resin data. However, the two species showed differing responses in foliar P and N:P. We propose that R. scutatus benefits from the increased N supply by increasing its uptake of soil P, while A. gigantea is unable to do so. This study clearly demonstrates that aboveground rain-funnelling structures can produce spatial heterogeneity in N supply, thereby creating a diversity of nutritional niches for plants.