Temporal environmental variation tips the balance between facilitation andinterference in desert plants

Recently, numerous studies have pointed to the importance of positive inter actions in natural communities. There is now a broad consensus that the bal ance between negative and positive interactions should shift along environm ental gradients, with competition prevailing under environmentally benign c onditions and positive interactions dominating under harsh conditions. A co mmonly cited example of the importance of facilitation in harsh environment s is the preference of desert annual plants for the areas under the canopy of shrubs. The recognition of apparently positive effects of desert shrubs on annuals, however, has been mostly based on density measurements, while f itness parameters of the understory plants have been ignored. Also, the tem poral consistency of such effects has not been previously tested. Based on conceptual ideas about the balance between interference and facilitation, w e predicted that positive effects of the shrubs on the understory should do minate in dry years, while in favorable years, negative effects would be st ronger. We tested our hypothesis by measuring the direction and magnitude o f the shrub effect on demographic responses of four desert annual plant spe cies during four consecutive seasons of differing rainfall. The results con tradicted our initial hypothesis. Depending on the species, the effect of t he shrubs shifted from either negative to neutral or from neutral to positi ve with increasing annual rainfall. However, this trend was stronger for th e effect of shrubs on plant reproductive success than on their densities. O ur data highlight the importance of measuring fitness parameters in studies of plant-plant interactions. We suggest that the negative effects of shrub s on plant fitness were due to rainfall interception, while positive effect s were related to increased nutrient availability beneath shrubs. However, the mechanisms by which the shrubs and annuals interact can only be resolve d using an experimental approach. Our results contradict previous hypothese s about the relative importance of positive and negative interactions along environmental gradients. A simple conceptual model summarizing the propose d role of rainfall in determining the direction of shrub effects on their u nderstory annuals is presented.