Testing the environmental filtering concept in global drylands

International audience

إنجليزي. 1. The environmental filtering hypothesis predicts that the abiotic environment selects species withsimilar trait values within communities. Testing this hypothesis along multiple – and interacting –gradients of climate and soil variables constitutes a great opportunity to better understand and predictthe responses of plant communities to ongoing environmental changes.2. Based on two key plant traits, maximum plant height and specific leaf area (SLA), we assessedthe filtering effects of climate (mean annual temperature and precipitation, precipitation seasonality),soil characteristics (soil pH, sand content and total phosphorus) and all potential interactions on thefunctional structure and diversity of 124 dryland communities spread over the globe. The functionalstructure and diversity of dryland communities were quantified using the mean, variance, skewnessand kurtosis of plant trait distributions.3. The models accurately explained the observed variations in functional trait diversity across the124 communities studied. All models included interactions among factors, i.e. climate–climate (9%of explanatory power), climate–soil (24% of explanatory power) and soil–soil interactions (5% ofexplanatory power). Precipitation seasonality was the main driver of maximum plant height, andinteracted with mean annual temperature and precipitation. Soil pH mediated the filtering effects ofclimate and sand content on SLA. Our results also revealed that communities characterized by a lowvariance can also exhibit low kurtosis values, indicating that functionally contrasting species canco-occur even in communities with narrow ranges of trait values.4. Synthesis. We identified the particular set of conditions under which the environmental filteringhypothesis operates in drylands world-wide. Our findings also indicate that species with functionallycontrasting strategies can still co-occur locally, even under prevailing environmental filtering. Interactionsbetween sources of environmental stress should be therefore included in global trait-basedstudies, as this will help to further anticipate where the effects of environmental filtering will impactplant trait diversity under climate change.