In face of the current environmental challenges, developing multifunctional cropping systems is increasingly needed, and crop variety mixtures are particularly interesting since they can deliver diverse services including grain production, yield stability, N₂O production regulation, disease control, and reduction of N-fertilizer losses. However, the relationships between intraspecific diversity and ecosystem multifunctionality are poorly understood so far, and practitioners lack science-based guidance to design mixtures. We used a pool of 16 bread wheat varieties classified into 4 functional groups based on 26 below- and aboveground functional traits, to conduct a field trial (88 large plots cultivated with single varieties or mixtures of 2, 4, or 8 varieties), quantifying 15 provisioning and regulating services for each plot. To assess yield stability between local conditions and years, the trial was replicated at 4 other locations and for 2 years, using 2 managements each time. We analyzed how variety number and functional groups predicted the variance in services, and applied in an innovative manner the RLQ co-inertia analysis to relate the (variety × traits) matrix Q to a (plot × services) matrix R, using a (plot × variety) composition matrix L as a link. Our results show that using variety mixtures allowed delivery of baskets of services not reachable when cultivating single varieties, and that mixtures mitigated tradeoffs between different pairs of services. Variety number or functional groups poorly predicted the variance in services, but the RLQ approach allowed the identification of groups of plots delivering consistent baskets of services. Moreover, we demonstrated for the first-time significant relationships between specific baskets of services and bundles of variety traits. We discuss how our results increase our understanding of intraspecific diversity–agroecosystem multifunctionality relationships, and propose the next steps using our new approach to support practitioners for designing variety mixtures that provide particular baskets of services.

面对当前的环境挑战，越来越需要开发多功能种植系统，而作物品种混合物尤其令人感兴趣，因为它们可以提供多种服务，包括粮食生产、产量稳定、N ₂ O生产调节、病害控制和减少N-肥料损失。然而，迄今为止，人们对种内多样性和生态系统多功能性之间的关系知之甚少，实践者缺乏基于科学的指导来设计混合物。我们使用了 16 个面包小麦品种，根据 26 个地下和地上功能性状分为 4 个功能组，进行了田间试验（88 个大地块，种植单一品种或 2、4 或 8 个品种的混合物），量化了每个地块有 15 个配置和调节服务。为了评估当地条件和年份之间的产量稳定性，该试验在其他 4 个地点重复进行了 2 年，每次使用 2 名管理人员。我们分析了品种数量和功能组如何预测服务的方差，并以创新的方式应用 RLQ 协同惯性分析，将（品种 × 性状）矩阵Q与（图 × 服务）矩阵R联系起来，使用（图× 品种）组成矩阵L作为链接。我们的结果表明，使用品种混合物可以提供培育单一品种时无法达到的一揽子服务，并且混合物可以减轻不同服务对之间的权衡。品种数量或功能组很难预测服务的差异，但 RLQ 方法可以识别提供一致服务篮子的地块组。此外，我们首次证明了特定的服务篮子和品种特征之间的显着关系。我们讨论了我们的结果如何增进我们对种内多样性与农业生态系统多功能关系的理解，并提出了使用我们的新方法支持从业者设计提供特定服务篮子的品种混合物的后续步骤。