1. Mycorrhizas are known to increase plant resistance to herbivores and climatic stress. However, it is unknown if particularly hostile environments select for increased investment of plants into mycorrhiza.

2. We studied hostile biotic environments: phylogenetically proximate neighbourhoods known to increase herbivory and pedoclimatic stress. In a common garden, we studied the resistances, tri-trophic interactions and microclimate of 28-years-old oaks (Quercus petraea), descending from provenances of contrasting phylogenetic neighbourhoods.

 3. We found that oaks descending from phylogenetically proximate neighbourhoods had increased ectomycorrhizal enzymatic activities without affecting mycorrhization rate of root tips. We also found that increased ectomycorrhizal enzymatic activity decreased damages by one group of specialist herbivores (leaf-miners), without affecting another (galls) or generalists (ectophages). Consistently, descendants from phylogenetically proximate neighbourhoods showed decreased damage by leaf miners. We finally found that oaks descending from phylogenetically proximate neighbourhoods were most capable of maintaining leaf chlorophyll under heat-induced drought, reflecting their increased ectomycorrhizal enzymatic activities.

4. Synthesis. Overall, oaks living with closely related neighbours can rapidly evolve increased investment into ectomycorrhizas and thereby resistances to herbivores and climatic stress, ultimately facilitating coexistence among closely related neighbours. While phylogenetically diverse forests remain essential for maintaining biodiversity, the selection regimes in less diverse forests can be used as a source of native genotypes resistant to future climate change, rather than using exotic populations and species as a source pool.

Measurement of herbivory

We inferred resistance of trees within the common garden to herbivores from low herbivory. We discriminated three feeding guilds: (1) galls: growth deformities on leaf surface by insect larva; (2) leaf mines: galleries formed between leaf epidermises by insect larva; (3) ectophagy: the partial or complete loss of leaf including epidermis. We assessed herbivory from 10 to 15 leaves, blindly selected from 40 to 60 leaves. To quantify herbivory we first reconstructed the initial leaf shape prior to damages. We took a grid of dots of 1 x 1 cm² and delimited on it the surface of the initial reconstructed leaf. Then we assessed the areas damaged by ectophages or leaf miners by counting the number of dots covering the damaged parts of the leaf.  We also counted the number of leaf mines and galls. For each leaf, we calculated the density of galls and density of leaf mines per cm², the proportion of area damaged by leaf miners, and the proportion of area damaged by ectophages. We then averaged each measure of leaf herbivory across leaves within trees and across trees within provenances. This highly standardized procedure enables us to avoid any bias in our measures. Also, the observer avoided measuring herbivory for the two trees of a provenance in a row to avoid bias.

Measurement of the proportion of ectomycorrhizal root-tips, ectomycorrhizal enzymatic activity and density of root tips

We inferred investment of trees within the common garden into ectomycorrhizas from the proportion of root tips with ectomycorrhizas and from the enzymatic activity of ectomycorrhizal root-tips. We assessed the proportion of ectomycorrhizal root-tips and density of root tips using the ground core taken in May 2018 using photo analyses (see appendix S8) and the ectomycorrhizal enzymatic activity from the ground core taken in September 2018 using high-throughput microplate assays (see appendix S8). Enzymatic activity, proportion of ectomycorrhizal root-tips and root-tip density were then averaged for each tree and averaged across trees within provenances.

Measurement of temperature, altitude and tree height

We inferred drought stress of the trees within the common garden from the air temperature at the surface of the crown at 1 pm (crown temperature from here on), i.e. the warmest period of the day, on the 1st of September 2018. Note that trees had experienced 2 weeks without rain and with high temperatures during the days prior to the assessment of the crown temperature. Air temperature at the surface of plants can be used as a proxy of drought stress 57,58. Air temperature at the surface drives drought stress by increasing saturation water-vapour pressure and thereby saturation deficit of the atmosphere which increases water loss of plants by transpiration. We used the altitude and tree height as a proxy for ground-water access of leaves, i.e. high altitude reduces access to ground-water and height impedes transport of ground water to leaves. We assessed crown temperature and canopy altitude of the common garden from remotely sensed data59 (see appendix S9). Temperatures, heights and altitudes were averaged across trees within provenances.