Summary1 A 3-year experiment involving nutrient addition and removal of one of two coexisting dwarf shrub species was conducted in two community types in a subalpine heathland on the northern Apennines (Italy). Vaccinium uliginosum occurred at all sites but was associated with the deciduous Vaccinium myrtillus at more sheltered nutrient-rich sites (HV community), and with the evergreen Empetrum hermaphroditum where the habitat was poorer and more exposed (EV community). Length of current-year shoots and fruit production of each species were determined in each of the 3 years, and standing crop at the beginning and end of the experiment.2 The length of current-year shoots of both deciduous species, but not of the evergreen, varied considerably between years, presumably due to varying temperatures at the beginning of the three growing seasons. Fruit production also varied. 3 Fertilization promoted an increase in the length of V. uliginosum shoots at the HV community but not at the EV community. The removal of V. uliginosum enhanced shoot elongation in V. myrtillus (HV community) but reduced shoot elongation in E. hermaphroditum (EV community). Neighbour removal did not aect shoot length of V. uliginosum at either community. There were few treatment eects on fruiting of these clonal species. 4 The standing crop in untreated stands did not change during the experimental period. Changes in shoot length resulting from environmental manipulations were not accompanied by consistent variation in the standing crop of any species. Standing crop increased only for V. myrtillus after removing V. uliginosum at the HV community (same direction as shoot length). The standing crop of E. hermaphroditum did not change after removing V. uliginosum at the EV community, although shoot length was signi®cantly reduced. The standing crop of V. uliginosum was unaected by neighbour removal and was decreased by nutrient addition at both communities. 5 None of the shrubs appeared able to utilize more abundant resources to increase above-ground biomass. However, increased shoot length in nutrient-rich habitats gave V. uliginosum a superior ability to capture light compared with V. myrtillus, the latter being more successful when the coexisting competitively superior species was removed. This would support Grime's theory, indicating that competition becomes more important as soil resource levels increase. Positive interactions appeared to play a role in the more extreme habitat, where E. hermaphroditum normally bene®ted from the shelter of the V. uliginosum canopy but was able to adapt its architecture when exposed.