Spatial and temporal aridity gradients provide poor proxies for plant-plant interactions under climate change: a large-scale experiment

1. Plant-plant interactions may critically modify the impact of climate change on plant communities. However, the magnitude and even direction of potential future interactions remains highly debated, especially for water-limited ecosystems. Predictions range from increasing facilitation to increasing competition with future aridification. 2. The different methodologies used for assessing plant-plant interactions under changing environmental conditions may affect the outcome but they are not equally represented in the literature. Mechanistic experimental manipulations are rare compared with correlative approaches that infer future patterns from current observations along spatial climatic gradients. 3. Here, we utilize a unique climatic gradient in combination with a large-scale, long-term experiment to test whether predictions about plant-plant interactions yield similar results when using experimental manipulations, spatial gradients or temporal variation. We assessed shrub-annual interactions in three different sites along a natural rainfall gradient (spatial) during 9 years of varying rainfall (temporal) and 8 years of dry and wet manipulations of ambient rainfall (experimental) that closely mimicked regional climate scenarios. 4. The results were fundamentally different among all three approaches. Experimental water manipulations hardly altered shrub effects on annual plant communities for the assessed fitness parameters biomass and survival. Along the spatial gradient, shrub effects shifted from clearly negative to mildly facilitative towards drier sites, whereas temporal variation showed the opposite trend: more negative shrub effects in drier years. 5. Based on our experimental approach, we conclude that shrub-annual interaction will remain similar under climate change. In contrast, the commonly applied space-for-time approach based on spatial gradients would have suggested increasing facilitative effects with climate change. We discuss potential mechanisms governing the differences among the three approaches. 6. Our study highlights the critical importance of long-term experimental manipulations for evaluating climate change impacts. Correlative approaches, for example along spatial or temporal gradients, may be misleading and overestimate the response of plant-plant interactions to climate change.