Climate change significantly impacts crop productivity, with projections indicating yield gains in northern regions and losses in the Mediterranean region, even at 1.5°C warming. While market spillover effects can mitigate some negative impacts, high risks remain, particularly concerning climate variability and extremes. Effective adaptation strategies are crucial for enhancing climate resilience in agriculture. Assessing impacts and designing adaptation strategies benefits from improved process-based crop models, although limitations and uncertainties remain. Quantifying the added value of climate services in adaptation is challenging, but crucial. This study focuses on the effectiveness of climate services in informing optimal variety choice at sowing for durum wheat, a major crop in the Euro-Mediterranean region. Previous studies have focused on climate change impacts on wheat, but this study uniquely simulates an idealized climate service and quantifies its benefits on wheat productivity. A user scoping workshop of the EU-H2020 MedGOLD project highlighted variety selection at sowing as a key decision needing support from sectoral climate services. Such services, based on seasonal climate predictions and local information, can provide risk estimation and advice farmers on minimizing risk for the growing season. The study explores the effectiveness of informed variety selection under near-future climate conditions (2021-2040, RCP8.5 scenario), evaluating the benefits of an idealized agro-climate service in reducing negative effects on mean wheat yield and stability. Other adaptation mechanisms are not considered to isolate the effect of the climate service.

The introduction thoroughly reviews existing literature on the impacts of climate change on agriculture, specifically focusing on wheat production. It highlights previous research on the effects of climate change on wheat yields in various regions, including the diverging impacts in different latitudinal zones. The review also acknowledges the limitations of current crop models and the challenges in quantifying the value of climate services in adaptation. The authors cite relevant studies that have explored climate change impacts, adaptation strategies, and the use of climate services in agriculture, providing a strong foundation for their research.

The study analyzed 191 durum wheat accessions to represent genetic diversity, identifying three families and defining 18 ideotypes. Crop yields were simulated under historical and projected climate conditions using the ECroPS crop growth model, driven by five regional climate models (RCMs) from the EURO-CORDEX initiative under the RCP8.5 scenario. The RCMs' daily temperature, precipitation, and solar radiation data were bias-adjusted using quantile mapping. ECroPS simulates water limitation, heat stress, and CO2 fertilization effects, but excludes nutrient limitations and pest/disease impacts. Soil data and parameters were derived from the European Commission Joint Research Centre's MARS system. The 18 ideotypes, representing a range of growing cycles, were used to assess the benefits of targeted climate services. An idealized climate service was implemented, applying a probability of success (prediction skill) to select the optimal variety based on projected climate conditions for each year. This probability, representing the service's accuracy, varied from 10% to 70%. Simulations were conducted with and without the climate service to compare impacts on mean yield, interannual variability, and a resilience index (μ²/σ²). Statistical tests (Anderson-Darling) were used to assess yield differences. The supplementary material contains additional detail on the RCMs used and other methodologies employed.

Without a climate service, yield reductions ranged from -7.8% to -5.8% across ideotypes, with shorter-cycle ideotypes less affected. Spatial heterogeneity in yield losses highlighted the need for local adaptation solutions. Interannual variability increased by 7-12% without adaptation. Integrating the idealized climate service yielded significant positive effects, offsetting negative impacts on mean yield at 40% prediction skill and producing yield gains up to 5.3% at 70% skill. However, this came with a 25% increase in interannual variability for lower-skill services. Only services with 70% or higher prediction skill achieved a balance between mean yield gains and reduced variability increases. Regional differences were observed; some regions benefit more from longer-cycle varieties while others favor shorter ones. The resilience index, accounting for both mean yield and interannual variability, showed higher model coherence, with longer-cycle varieties often optimizing resilience.

The findings confirm climate change's negative impacts on durum wheat yields but demonstrate that targeted adaptation, specifically through a climate service guiding variety selection, can mitigate these impacts. The study highlights a trade-off between mean yield improvement and increased interannual variability. High prediction skill (≥70%) in the climate service is crucial to achieve both yield gains and reduced variability, supporting the need for high accuracy in climate services for effective adaptation. The results suggest the potential role of variety mixtures in reducing climate change impacts, even with lower-skill climate services. The study emphasizes the need for a dynamic adaptation strategy that involves continuous monitoring, feedback, and re-implementation.

This study demonstrates that integrating an idealized agro-climate service to inform variety selection significantly improves the climate resilience of durum wheat in the Euro-Mediterranean region. High prediction skill (≥70%) is essential to balance yield gains with reduced interannual variability. The findings underscore the importance of dynamic adaptation strategies, emphasizing continuous monitoring and collaboration with stakeholders. Future research should focus on expanding model ensembles to incorporate more detailed traits, and developing integrated farm system models to consider the full range of farm management decisions.

The study used an idealized climate service that simplifies the complexities of real-world climate prediction and farmer decision-making. Spatial and temporal dependencies in farmer choices, variety restrictions, and access to varieties were not fully considered. The model excluded nutrient limitations, pest/disease impacts, and other factors that could influence yield. The use of a limited number of ideotypes might not fully capture the spectrum of potential durum wheat varieties. The results are specific to the RCP8.5 scenario and the selected RCMs; other scenarios might lead to different outcomes.