Trees affect soil chemistry and nutrient availability via a broad range of processes. Effects can vary dramatically among species, whose distinctive spatial “footprints” can vary for different nutrients. Potentially overlapping effects of neighboring trees in mixed-species stands make footprint shape and interspecific interactions important: If interactions are non-additive, then not only abundance but also spatial configuration influence tree species’ effects on ecosystem properties. We used spatially explicit neighborhood-scale data on tree distributions to fit maximum likelihood models of exchangeable calcium, magnesium, and potassium in surface soils of four sites in northern hardwood forests in northwestern Connecticut, USA. The models incorporated parent material, site, and tree species or functional group configuration to predict availability of the three cations. Site had a stronger effect than species for all cations (there was no species effect for potassium), even after accounting for variation in parent material. Species’ spatial footprints extended further from the stem for calcium than magnesium, which is consistent with the relative importance of litterfall versus stemflow transfer of these nutrients. The magnitude of species effects on calcium and magnesium varied widely. Functional groups made up of species with positive or negative effects provided parsimonious models for magnesium and calcium, and the best model for calcium included a non-additive, antagonistic effect whose strength varied by site. This non-additive effect suggests that the degree of intermingling of tree species from negative- and positive-effect functional groups may influence stand-level availability of calcium, a key nutrient for forest health in these ecosystems.