1. Climate change will increase the level of drought stress experienced by plant communities, but the spatial distribution of projected changes in dryness remains highly uncertain. Species can, to some extent, deal with climate uncertainty through natural variation in adaptive responses to environmental heterogeneity and predictability. Biodiversity conservation could thus target populations pre-adapted to climatic heterogeneity to anticipate climate uncertainty. Disentangling evolution of trait means vs. trait plasticity, however, requires a sampling design with genetic replicates grown under distinct environmental conditions. 2. Here, we applied three soil moisture treatments to genetic replicates of Fragaria vesca plants raised from seeds that were sampled in distinct topographical settings, to study adaptive trait and plasticity divergence in response to drought. 3. We demonstrate that various plant traits evolved along distinct topographical gradients. For example, populations on south-exposed slopes retained high levels of both flowering and runner formation under drought stress, while north-faced populations hardly flowered under reduced soil moisture levels. Aspect but not elevation was found to coincide with plant traits, suggesting that micro-environmental variation rather than general clines in elevation drive evolution in mountainous landscapes. Our results also indicate that traits and their plasticity can evolve independently in response to distinct topographical stressors. 4. Synthesis. We conclude that heterogeneous landscapes (i) harbor micro-refugia of adaptive genetic diversity that protect natural populations against environmental change, and (ii) represent invaluable sources of quantitative genetic variation that could support conservation where climate projections are inconclusive.

Phenotypic data were collected in a glasshouse, in which Fragaria vesca plants originating from 11 locations were subjected to three episodes of three soil moisture treatments. The phenotypic variables include dried above-ground biomass, specific leaf area (SLA), the total number of runners produced during the second episode soil moisture treatment (counted and cut each week), number of stomata, and total number of flowers during a third episode of soil moisture treatments. SLA was measured for the most representative leaf per plant (i.e. average in size), as the ratio of “leaf area / dried leaf mass”. Stomata were counted on nail polish leaf prints using a KEYENCE light microscope at 1000 x magnification. For three replicate counts per leaf print, median stomatal density instead of averages were used to minimize the impact of counting errors. We did not measure stomatal size because image quality was insufficient for accurate size estimates. To disconnect runner and flower formation from growth, we divided runner and flower numbers by biomass. Thus, runnering and flowering were expressed per unit dry biomass (g). As a proxy for potential maternal effects, dry above-ground biomass was measured during the first soil moisture treatment (early growth).

For each plant, the genotype, treatment, trait and topographical values are provided. Each plant has an ID representing the genotype, clone and treatment. For example, LC08_13 is genotype 8 from population LC, clone number 13, and is raised in wet conditions. Nheight represetns normalized height, and North_to_South represents orientation relative to the south (0=north, 180=south).