Abstract: Plant-soil feedbacks (PSF) are important processes that determine many aspects of ecosystem functioning. In semiarid environments, positive PSF maintain soil fertility and plant performance in a patchy community structure. Although the role of PSF in ecosystem processes is well known, their relative importance in maintaining rangeland productivity is less well understood, nor is the effect of grazing in modulating ecosystem PSF in semiarid environments. The aim of this study was to investigate the importance of PSF in semi-arid rangelands. Specifically, we experimentally tested (1) the effect of grazing on soil fertility depletion and the subsequent effect on plant productivity; (2) the species-specific role of positive PSF in reversing soil nutrient depletion and plant performance. We hypothesised that soils from grazed areas will be nutrient depleted, affecting plant productivity, particularly in unprotected bare soils. Secondly, we expected positive PSF to improve soil fertility and vegetation productivity, and we further investigated species-specific PSF effects along three dominant plant species with different ecological strategies (a N-fixing legume, a stress-tolerant shrub and a competitive grass species). We set up a two-stage experiment in the greenhouse, where in the first stage (F1) the three plant species were sown in pots on four different soils collected in the field: grazed under canopy (GP), grazed in bare soil (GB), ungrazed under canopy (UP) and ungrazed in bare soil (UB). In the second phase (F2), after one year of the experiment, soils from all combinations of field provenance and plant species sown in F1 were sown again with the same three species for another year. We measured soil N, P and organic matter content (initially, after F1 and F2), plant aerial biomass, height and basal diameter (after F1 and F2). Initially, there were differences in soil fertility between soil types, with GB soils being less fertile than GP, UP and UB soils. After F1, fertility differences between treatments were still significant and plant growth and aerial biomass correlated with soil fertility. After F2, differences in soil fertility were buffered, with the GS treatment levelling out with the other treatments. In addition, there was a significant interaction between soil origin and plant species identity sown at F1, which determined soil fertility, plant growth and biomass at F2, suggesting a species-specific PSF. In conclusion, this study suggests that PSF is a relevant ecological process for maintaining rangeland productivity in semiarid environments.
