Defoliation through herbivory is well known to affect target plants and their associated belowground properties, but the response of plants and their soil environment to defoliation of their neighbours is less well understood. We performed a controlled shade‐house experiment involving three plant species that colonize New Zealand floodplains during primary succession, i.e. a palatable N2‐fixing shrub (Carmichaelia odorata), a palatable deciduous small tree (Fuchsia excorticata) and a less palatable evergreen tree (Weinmannia racemosa). All species were grown in large pots for 40 months both singly and in two species pairs, and either one or both of the species grown in pairs were clipped to simulate herbivory. Responses of growth and foliar nutrient status to clipping varied strongly among species, with Carmichaelia having the largest response and Fuchsia having the smallest. Carmichaelia also enhanced soil microbial biomass and activity, and foliar N concentrations of Weinmannia. However, this did not translate to a net positive effect; instead Carmichaelia competitively reduced growth and foliar P concentrations of both other species. Most effects of Carmichaelia on the soil microflora, and growth and nutrient status of its neighbours, disappeared when Carmichaelia was clipped. Further, the effect of clipping Carmichaelia had a stronger impact on growth, soil activity and nutrient status of the other two species than did the clipping of those species. These results contradict expectations that N2‐fixing plants should promote growth of other species in pioneer communities or that defoliation of N2‐fixers exacerbate positive effects; in our study, defoliation of Carmichaelia merely mitigated the negative effects that it had on other species. They also suggest that interplay of competition and differential herbivory among coexisting plants has important implications for soil microflora and processes, relative nutrient acquisition and stoichiometry of coexisting plant species, and potentially plant community development.