Abstract Understanding the mechanisms determining population temporal stability is key to explaining species diversity and coexistence, and to maintaining multiple ecosystem functions over time. Functional traits should offer mechanistic insights into the drivers of stability, but generalization of a trait–stability relationship is lacking. Our overarching hypothesis is that traits predicted by bet‐hedging theory to reduce variability in population size, that is, delayed germination and/or bigger seeds and/or traits associated with higher drought resistance (high intrinsic water‐use efficiency, high turgor loss point, slow relative growth rate, high leaf dry matter content, low SLA) lead to higher population temporal stability. We linked these traits to the year‐to‐year variation in winter annuals' abundance of 178 populations (66 species) over 13 years. We tested the generality of these relationships in a biodiversity hotspot along a gradient of rainfall mean, variability and unpredictability defining three climates (Mediterranean, semi‐arid, arid), considering also the spatial micro‐heterogeneity defined by perennial shrubs. Away from shrubs, higher drought resistance was consistently associated with higher population stability in all climates. Under the shrubs, however, this relationship disappeared or was reversed. Delayed germination predicted higher stability in the arid open habitat only. Our results unveiled diverse drought resistance strategies within annuals in drylands, and elucidated their influence on population temporal stability. Our results challenge the long‐standing assumption that delayed germination is the most important bet‐hedging trait for annuals in drylands. Finally, we highlight the crucial need of including local microhabitat conditions in the assessment of population stability. Synthesis . The detection of drought resistance as a key driver of population temporal stability represents a significant advancement in evolutionary theory, trait‐based ecology and our understanding of community stability.
Gresse et al. (Sun,) studied this question.