1. Afforestation is globally increasing to produce timber and pulp wood, but also to enhance ecosystem services such as carbon sequestration, nutrient retention or groundwater recharge. In China, large areas have been and will be afforested in order to compensate for the negative impacts of former clear-cuttings and to make use of the ecosystem services associated with afforestation. In order to further optimize these services with regard to balanced nutrient (particularly nitrogen) cycles, it is important to know whether the use of mixtures of native tree species in afforestation projects promotes the acquisition and retention of nitrogen compared with the currently established large-scale monocultures.

2. To test the effect of species richness on system N retention and tree sapling N uptake, we conducted a 15N tracer experiment in a young tree plantation. To this end, saplings of four abundant early successional tree species were planted in monocultures, in two- and four-species mixtures and as single trees.

3. Nitrogen retention increased with higher species richness due to enhanced N pools in sapling biomass. These species richness effects strengthened over time.

4. Species-specific differences in 15N recoveries over time revealed below-ground niche differentiation with regard to N uptake, which is likely to result in complementary resource use among coexisting species.

5. Synthesis and applications. This study provides evidence that mixed afforestation promotes N retention from the sapling stage. To further improve ecosystem services associated with afforestation, we strongly suggest the use of mixtures of native tree species instead of monocultures. Mixtures of four species may reduce system N losses and thus may lessen groundwater contamination due to N leaching. We encourage further investigations to find optimal species combinations that promote a wide range of ecosystem services related to more closed nutrient cycles and minimized soil erosion. In our study, the plantations' capability to retain N could be optimized by means of both increasing tree species richness and by choosing the optimal species combinations.