Marie Gosme

Diversification of farming systems at plot scale is increasingly seen as an essential option to meet agriculture sustainability challenges and support agroecological transition. Significant progress has been made in recent years on methods for designing diversified cropping systems based on annual crops. However, few methods have been tested for designing diversified cropping systems based on perennial crops, and in particular perennial crop-based agroforestry systems, despite their important dietary and economic roles in the world. In this study, we argue that the characteristics of perennial crops induce specific structural and functional changes that impact the design process. We identified four key concepts for the design of diversified perennial crop-based agroforestry systems: (i) the targeted productivity must be considered over the long term and at the plot level, and encompass not only the productive and non-productive stages of perennial crops but also of associated plants; (ii) these systems will eventually face shocks and crisis in their lifetime, so they should be designed with resilience and adaptability in mind; (iii) initial designs will necessarily evolve throughout the long lifespan of perennial crops, and so a combination of de novo design from scratch and step-by-step design approaches for regeneration or transformation will be required; and (iv) due to the high complexity of these systems and the difficulty of setting up trials, the design process should draw from a range of knowledges from academics and practitioners and hybrid them. These four key features show that perennial crops offer distinct opportunities for the design of sustainable diversified cropping systems, but also face specific challenges that require more transdisciplinary research.

Agroforestry systems (AFSs) enhance biodiversity, productivity, and climate resilience, yet their inherent complexity challenges traditional modeling approaches. This study presents an integrated framework combining process-based modeling (PBM), machine learning, and life cycle assessment (LCA) into a user-friendly decision support system (DSS) for AFS analysis and optimization. Multi-scenario simulations using Hi-sAFe (16 configurations varying in latitude and orientation) for maize–poplar systems produced detailed outputs, including land equivalent ratios (LER up to 1.29), crop yield (ranging from 4.3 to 8.1 t ha ⁻¹ ), and N ₂ O emissions (1.6–9.4 kg N ha ⁻¹ ). We trained Random Forest models on PBM diagnostics, microclimate, and design covariates; post-hoc SHapley Additive exPlanations (SHAP) identified maximum air temperature (Tmax), distance to tree rows, and photosynthetically active radiation (PAR) as the dominant predictors of yield and LER. To address data gaps, long short-term memory (LSTM) and TrAdaBoost-LSTM models were used for microclimate forecasting, achieving high fidelity ( R ² = 0.88) in capturing hourly temperature and humidity trends. Outputs have been integrated into an LCA workflow, producing CO ₂ -equivalent and nitrogen footprint metrics that feed into a functional web-based DSS. This platform enables robust, transparent, and scalable AFS evaluation, supporting stakeholders with real-time scenario exploration and scientifically grounded decision-making tools. Furthermore, the framework enables greenhouse gas (GHG) accounting in AFSs by integrating modeled soil organic carbon dynamics and N ₂ O emissions. A Germany-wide assessment illustrates the significant GHG reduction potential achievable through large-scale implementation of AFSs.

Selecting the right tree species is a crucial step in designing sustainable and effective agroforestry systems. To support this process, several decision support systems (DSS) have been developed in various countries to help farmers and advisors choose appropriate species. However, these tools are often limited in reach—typically used only within the country where they were created—and tend to focus on different aspects of adaptation to local conditions or achieving specific farming objectives. In this context, we introduce a new framework for agroforestry tree species selection. Its goals are twofold: (i) to compile and organize the knowledge embedded in existing tree selection tools, and (ii) to offer an intuitive, user-friendly graphical interface—AgroforesTreeAdvice. This tool allows users to input local parameters such as soil type, climate, biotic factors, and farm-level or socio-economic constraints, alongside production goals (e.g., timber, fruits) and ecosystem service objectives (e.g., soil conservation, carbon storage). The unified system successfully integrates eight existing agroforestry DSS, improving their (i) findability by centralizing them in one location, (ii) accessibility through a standardized and user-friendly interface, (iii) interoperability by enabling API-based queries across all tools, and (iv) reusability for future DSS development. Finally, we present early user feedback from pilot deployments in three countries, highlighting the tool’s benefits and outlining next steps toward further harmonization of agroforestry databases.

Agroforestry could be a major strategy to adapt agriculture to climate change, thanks to the microclimate effects of trees and improved infiltration. However, the experimental validation of these claims is scarce. In this methodological review, we discuss options for the experimental simulation of drought conditions in agroforestry field experiments, comparing it with strategies adopted in natural, agricultural, or forestry ecosystems. We classify rainout shelters used in field experiments according to mobility, completeness of rain interception and height of rainout shelter. We show that specificities of agroforestry systems create constraints and require compromises in the design and operation of rainout shelters. We conclude that large rainout shelters, which induce drought for both the trees and the crops while limiting artifacts and biases, would be most relevant for studying the resistance of agroforestry systems to drought. Unfortunately, the review of rainout shelters already used in agroforestry systems reveals a lack of rainout shelters capable of intercepting rain on both trees and crops, achieving total rain interception, while being relatively low-cost and manageable by a small team. Therefore, we benchmark three novel rainout shelter designs that we tested in a mature agroforestry system under Mediterranean climatic conditions. We discuss their advantages and disadvantages in terms of both scientific and operational aspects. While compromises had to be done between experimental design, risks of artifact/bias, effectiveness, ease of installation, operation and maintenance, and agricultural management, these prototypes are starting points for achieving well-performing rainout shelters and testing the effects of drought in agroforestry experiments.

Despite its potential for fostering farm sustainability, the adoption of agroforestry faces context-dependent challenges, among which the (perceived) shortage of decision-supporting tools and barriers hindering the assessment of economic, environmental, and social benefits. The process of digitalization offers significant opportunities to enhance sustainability, but it remains crucial to foster a human-centric, fair, and sustainable approach. In the context of the DigitAF Horizon Europe project, we present the results of a multi-stakeholder questionnaire aimed at understanding the perceptions of stakeholders regarding agroforestry and digitalization, as well as the needs of these stakeholders for a successful implementation of this agricultural practice. In the questionnaire, there was a specific focus on the need for and the conditions for the use of digital tools and models, such as generalized digital tools, applications and mapping, climate and weather forecasting and recording, farm management and decision support, and agroforestry and environmental tools. The purpose of this survey was to provide insights to inform agroforestry actors and to foster collaborative initiatives that enhance the potential of digital tools to support the design, implementation, and maintenance of effective and sustainable agroforestry in the European context. Our questionnaire was completed by stakeholders from seven European countries, including farmers, academics, policy actors, farm advisors, and actors in the value chain with an interest in agroforestry. Stakeholders from six living labs, representing Czechia, Finland, Germany, Italy, the Netherlands, and the UK, were involved in the appraisal, along with a multi-stakeholder group from Belgium. Respondents used data and digital tools for various purposes in farming systems and were interested in their potential to improve agroforestry including animal, tree, and crop performance, management guidance, system design, and tree species selection. Our survey revealed that the perceived usefulness of digital tools for agroforestry was substantially higher than stakeholders' awareness of existing tools, indicating a need for better promotion and development of user-friendly, accessible solutions. Additionally, significant obstacles to agroforestry adoption, such as large up-front investments, administrative burdens, and fear of reduced CAP support, were identified, emphasizing the necessity for targeted support and policy improvements. Moving forward, efforts should focus on developing targeted solutions to promote agroforestry according to stakeholder perception, and user-friendly digital tools tailored to the needs and preferences expressed by stakeholders, while also increasing knowledge sharing and capacity building among practitioners and researchers.

Agroforestry systems are complex due to the diverse interactions between their elements, and they develop over several decades. Existing numerical models focus either on the structure or on the functions of agroforestry systems. However, both of these aspects are necessary, as function influences structure and vice versa. Here, we present a representation of agroforestry systems based on combinatorial maps (which are a type of multidimensional graphs), that allows conceptualizing the structure–function relationship at the agroecosystem scale. We show that such a model can represent the structure of agroforestry systems at multiple scales and its evolution through time. We propose an implementation of this framework, coded in Python, which is available on GitHub. In the future, this framework could be coupled with knowledge based or with biophysical simulation models to predict the production of ecosystem services. The code can also be integrated into visualization tools. Combinatorial maps seem promising to provide a unifying and generic description of agroforestry systems, including their structure, functions, and dynamics, with the possibility to translate to and from other representations.

Introduction: Climate change (CC) and the increased occurrence of extreme climatic events pose a serious threat to crop yields and their stability worldwide. This study analyzed the CC mitigation potential of an alley cropping system on crop physiological stresses and growth as compared to a monoculture system. Materials and Methods: Growth of winter durum wheat, cultivated alone (agriculture) and in combination with hybrid walnut (agroforestry), was simulated with the Hi-sAFe agroforestry model, as driven by business-as-usual Intergovernmental Panel on Climate Change (IPCC) projections, split into three scenarios, representing Past (1951–1990), Present (1991–2030), and Future (2031–2070) climatic conditions. Crop growth and the occurrence of thermal, nitrogen, and water stresses were analyzed. Results: Cold-related stresses were modest in Past and almost disappeared over time. Heat, drought, and nitrogen stresses increased about twofold from Past to Future, but were reduced by 20–35% in agroforestry, already with medium-sized trees (diameter at breast height (DBH) of about 10–15 cm). Crop yields in agriculture increased from Past to the end of Present and then remained stable. This moderately decreased with tree age in agroforestry (especially in Future). Discussion: The impact of CC on the crop was buffered in agroforestry, especially for the most extreme climatic events. The mitigation of crop microclimate and the increased stability of crop yields highlight the potential of agroforestry as a CC adaptation strategy.