The relative importance of facilitation and competition between pairwise plants across abiotic stress gradients as predicted by the stress-gradient hypothesis has been confirmed in arid and temperate ecosystems, but the hypothesis has rarely been tested in tropical systems, particularly across nutrient gradients. The current research examines the interactions between a pioneer shrub Rhodomyrtus tomentosa (the nurse plant) and seedlings of a transplanted native woody Schima superba (the target species) in a tropical system in which position on a slope corresponds with a nutrient gradient; high soil nutrients at the slope bottom and relatively low soil nutrients at the slope top. In contrast, soil physical traits were more favorable for seedling growth under the shrub than in open spaces. The effect of R. tomentosa on S. superba survival was positive (facilitation) at the top of the slope, as indicated by the relative interaction index (RII), but negative in the bottom (competition). RII indicated a positive effect on seedling height at the top of the slope but was not at the bottom. Seedling survival was positively related to soil nutrient level and negatively related to soil acidity, but seedling growth of S. superba seemed to be enhanced by the shrub canopy. Thus, the results seem to support stress-gradient hypothesis in terms of target species survival but not growth. We suggest using the shrub as a nurse plant in forest restoration in tropical degraded land with caution because not all of its effects on target species are positive.