The increasing frequency of drought events under climate change poses significant challenges for the selection of plant communities in ecological restoration. This study introduces a trait-based, multi-objective optimisation framework that identifies plant community compositions balancing physiological adaptation to drought with functional trait diversity.

This approach integrates climate-derived ecological targets and trait-based community assembly principles. Climatic constraints are incorporated by defining targets for the community-weighted mean of P50, a physiological trait indicative of xylem vulnerability to embolism, using site-specific drought intensity derived from the Standardised Precipitation Evapotranspiration Index (SPEI). Functional diversity was quantified using Rao’s quadratic entropy to capture trait dissimilarity across communities.

The optimisation procedure, implemented via the Non-dominated Sorting Genetic Algorithm II (NSGA-II), generates a set of Pareto-optimal solutions that represent trade-offs between aligning the community-weighted mean P50 with the climatic target and maximising functional diversity.

The model was applied to a Mediterranean forest (Palo Laziale, Italy) under two drought scenarios: Near Normal and Extra Dry. Results show: I) increasing aridity reduces the solution space; II) in more arid conditions, species with moderate drought tolerance and distinct functional profiles are favoured; III) despite the stronger climatic filtering, drier conditions tend to promote communities with more balanced and diversified species contributions, suggesting a shift from single-species dominance; IV) kernel density estimation and convergence analyses across 500 replicated runs confirmed the stability and recurrence of optimal configurations, particularly under extreme climatic stress.

This framework offers a flexible and reproducible tool to support ecological restoration planning under climate change. By explicitly modelling ecological trade-offs and incorporating climate-responsive trait targets, it enables the identification of multiple alternative plant assemblages that are both resilient and ecologically plausible.