Abstract
BACKGROUND AND AIMS: The stress gradient hypothesis (SGH), the view that competition prevails in undisturbed and productive environments, and shifts to facilitation in disturbed or stressful environments, has become a central paradigm in ecology. However, an alternative view proposes that the relationship between biotic interactions and environmental severity should be unimodal instead of monotonic. Possible causes of discrepancies between these two views were examined in the high elevation desert of the arid Trans-Himalayas. METHODS: A putative nurse species and its associated plant community was surveyed over its entire elevation range, spanning from alpine to desert vegetation belts. The results were analysed at the community level (vegetation cover and species richness), considering the distinction between the intensity and the importance of biotic interactions. Interactions at the species level (pairwise interactions) were also considered, i.e. the variation of biotic interactions within the niche of a species, for which the abundance (species cover) and probability of occurrence (presence/absence) for the most widespread species along the gradient were distinguished. KEY RESULTS: Overall, facilitation was infrequent in our study system; however, it was observed for the two most widespread species. At the community level, the intensity and importance of biotic interactions showed a unimodal pattern. The departure from the prediction of the SGH happened abruptly where the nurse species entered the desert vegetation belt at the lowest elevation. This abrupt shift was attributed to the turnover of species with contrasting tolerances. At the species level, however, facilitation increased consistently as the level of stress increases and individuals deviate from their optimum (increasing strain). CONCLUSION: While the stress gradient hypothesis was not supported along our elevation gradient at the community level, the strain gradient hypothesis, considering how species perceive the ambient level of stress and deviate from their optimum, provided a parsimonious explanation for the outcome of plant-plant interactions at both scales.