Face aux besoins croissants pour une production agricole durable, les systems de culture évoluent vers des systèmes qui accomplissent des objectifs environnementaux et agricoles multiples. La recherche en conception de systèmes de cultures (CSC) s'intéresse à l'effet des pratiques et de l'environnement sur les systèmes de culture et leur performance. L'interaction entre production et services ecosystémiques, et la quantification de ces relations, sont un aspect clé de ce domaine de recherche. Une variété d'approches ont été théorisées, tels que l'utilisation de modèles et la mobilisation de connaissances expertes. Les modèles permettent de tester rapidement et à faible coût l'effet de pratiques agricoles dans une variété de conditions, mais l'application de conclusions théoriques à la parcelle peut être limitée par des contraintes locales ainsi que des obstacles à la communication chercheur-agriculteur. Mobiliser les agriculteurs et autres acteurs pertinents pour la CSC peut aider à surmonter ces obstacles ; cependant, cela limite l'innovation au cadre des connaissances expertes.L'objectif de cette thèse est de combiner la modélisation et des méthodes participatives pour une méthode de CSC qui exploite le potentiel de la modélisation numérique tout en s'assurant que les solutions proposées prennent en compte les contraintes environnementales et socioéconomiques. Après avoir revu l'état d'avancement de la recherche en prototypage et en CSC, nous proposons un cadre méthodologique divisé en quatre parties ; a) combiner une typologie des pratiques et un modèle conceptuel pour évaluer la diversité des pratiques, contraintes et trade-offs dans une zone de production ; b) acquérir des données de terrain pour quantifier les trade-offs pertinents entre production et services écosystémiques ; c) sélectionner et préparer un modèle numérique approprié pour simuler les effets des pratiques sur la production et l'apport de services ; et d) évaluer si l'interaction d'agriculteurs avec le modèle numérique peut générer des systèmes de culture potentiels qui répondraient aux objectifs agro-environnementaux posées (apport d'un service écosystémique) ainsi qu'être acceptables pour les agriculteurs qui les adapteraient à l'expérimentation dans leurs parcelles.The systèmes agroforestiers à base de café (cafés/arbres d'ombrage) du Costa Rica central ont étés le système de culture choisi pour répondre à ces questions. Les systèmes agroforestiers offrent de nombreuses occasions d'étudier et évaluer les services écosystémiques apportés, en plus de la production principale. L'association de deux cultures pérennes place l'évaluation de la performance à long terme et de la durabilité des systèmes au centre de la question. La culture du café au Costa Rica fait vivre une part importante de la population, et est aussi basée sur la gestion intensive d'une culture à haute valeur d'exportation, vulnérable aux fluctuations des prix sur le marché mondial ainsi qu'au changements climatiques. Des pentes raides et une saison des pluies importante créent des problèmes d'érosion significatifs ; cependant, certaines pratiques de contrôle de l'érosion (utilisation d'arbres d'ombrage et d'adventices) impactent la production de café. La réconciliation de ces deux aspects nous offrent l'occasion de tester notre cadre méthodologique dans une situation où une solide argumentation technique serait nécessaire pour encourager les expérimentations dans les parcelles. Enfin, le dernier chapitre porte une réflexion d'ensemble sur l'importance de choisir et préparer correctement un modèle agronomomique adéquat, les application potentielles de cette méthodologie, ainsi que les recommandations que nous avons pu effectuer en termes de pratiques de contrôle de l'érosion dans la zone d'étude. / In the face of increasing concerns about sustainability of agricultural production, cropping systemsare evolving towards systems that fulfill multiple agronomic and environmental objectives. Researchin cropping systems design (CSD) is concerned with studying the effect of farming practices oncropping systems and their performance. The interaction between production and ecosystemservices, and quantification of trade-offs between the two, is a key aspect of this research. A varietyof approaches have been theorized, such as use of models and mobilization of expert knowledge.Models allows fast and low-cost testing of the effect of farming practices under a variety ofconditions, but the application of theoretical outcomes to on-farm changes can be limited by localconstraints and researcher-farmer communication. Mobilizing farmers and other relevantstakeholders for CSD can help overcome these obstacles; however this limits innovation to the scopeof expert knowledge.The objective of this thesis is to combine modeling and participatory methods for a CSD frameworkthat harnesses the potential of numerical modeling while ensuring the proposed solutions take intoaccount socioeconomic and environmental constraints. After an overview of current advances inprototyping and CSD, we propose an methodological framework divided into four parts; a) combininga typology of farming practices and a conceptual model to appraise the diversity of farming practices,constraints and trade-offs at the plot scale in a defined production area; b) collection of field data forquantifying relevant trade-offs between production and ecosystem services; c) selecting andpreparing an appropriate numerical model for simulating the effects of farming practices onproduction and provision of ecosystem services; and d) evaluating whether the interaction of farmerswith a numerical model can generate candidate cropping systems that fulfill our agro-environmentalobjectives (provision of ecosystem service) as well as being suitable for the farmers who will adaptthem for on-farm experimentation.The coffee-based agroforestry systems (coffee/shade trees) of central Costa Rica were the chosenproduction system for answering these questions. Agroforestry systems offer plentiful opportunitiesfor valuing ecosystem services in addition to crop production; the combination of two perennialcrops brings long-term performance assessment and sustainability of the system to the heart of thequestion. Coffee cultivation in central Costa Rica concerns a large amount of livelihoods, but is alsobased on intensive management of a highly valued cash crop vulnerable to price fluctuations on theglobal market as well as climate change. Steep slopes and heavy rainfall also cause high levels of soilerosion; yet certain indirect erosion control practices (such as the use of shade trees of weeds) alsohave an impact on coffee production. The reconciliation of these two aspects offers the opportunityto test our methodological framework in situations where precise discussions onproduction/environment trade-offs are needed.Finally, in the last chapter we reflect on the importance of correctly choosing and preparing the rightmodel for the job, potential application of this methodology, as well as the recommendations wereable to make in terms of erosion control practices in the study area.

Second-generation biofuels could provide renewable energy while reducing the globaleconomy dependence on oil and mitigating climate change. However, their greenhouse gasemission balances, as well as their energy and environmental balances, are discussed,especially when they are produced from agricultural feedstock. The use of agriculturalfeedstock for energy purposes also raises the issue of competition with food production. Inthis context, this work contributes to the assessment of the sustainability of Miscanthus xgiganteus, a perennial C4 crop candidate to the production of second-generation ethanol. Theobjectives of this work are (i) to achieve a multicriteria evaluation of cropping systems basedon M. Giganteus using data collected in farmers’ fields and (ii) to compare these croppingsystems with cropping systems including other resources Agricultural candidates for biofuelproduction. The main contributions of this work are (i) the study of the variability of yieldsand winter nitrate losses in a network of commercial fields located in Burgundy (France), (ii)the characterization by modeling of M. giganteus long-term yield evolution and (iii) theintegration of these findings in a process of cropping systems design and assessment aimingat comparing M. giganteus with other feedstock candidate to the production of bioethanol.The study of M. giganteus in farmers’ fields shows that the high variability of yields andnitrate losses is linked to (i) crop age, (ii) soil type and (iii) the type of field (i.e. culturalhistory, size, shape, and environment). Contrasting yield scenarios, built by combining datacollected in commercial fields with a long-term yield evolution model, show that thesensitivity of assessment results regarding yields depends on the assessment field. Theinsertion of M. giganteus in a cropping system can significantly improve the greenhouse gasemission balance as well as the environmental balance, compared with a cropping systembased on a short cropping sequence. Economic results depend strongly on M. giganteus yield.Other agricultural feedstocks are also interesting, especially on soils where the yield potentialof M. giganteus is low: this is particularly the case of alfalfa stems, which can be used forsecond-generation ethanol production. / Second-generation biofuels could provide renewable energy while reducing the globaleconomy dependence on oil and mitigating climate change. However, their greenhouse gasemission balances, as well as their energy and environmental balances, are discussed,especially when they are produced from agricultural feedstock. The use of agriculturalfeedstock for energy purposes also raises the issue of competition with food production. Inthis context, this work contributes to the assessment of the sustainability of Miscanthus xgiganteus, a perennial C4 crop candidate to the production of second-generation ethanol. Theobjectives of this work are (i) to achieve a multicriteria evaluation of cropping systems basedon M. Giganteus using data collected in farmers’ fields and (ii) to compare these croppingsystems with cropping systems including other resources Agricultural candidates for biofuelproduction. The main contributions of this work are (i) the study of the variability of yieldsand winter nitrate losses in a network of commercial fields located in Burgundy (France), (ii)the characterization by modeling of M. giganteus long-term yield evolution and (iii) theintegration of these findings in a process of cropping systems design and assessment aimingat comparing M. giganteus with other feedstock candidate to the production of bioethanol.The study of M. giganteus in farmers’ fields shows that the high variability of yields andnitrate losses is linked to (i) crop age, (ii) soil type and (iii) the type of field (i.e. culturalhistory, size, shape, and environment). Contrasting yield scenarios, built by combining datacollected in commercial fields with a long-term yield evolution model, show that thesensitivity of assessment results regarding yields depends on the assessment field. Theinsertion of M. giganteus in a cropping system can significantly improve the greenhouse gasemission balance as well as the environmental balance, compared with a cropping systembased on a short cropping sequence. Economic results depend strongly on M. giganteus yield.Other agricultural feedstocks are also interesting, especially on soils where the yield potentialof M. giganteus is low: this is particularly the case of alfalfa stems, which can be used forsecond-generation ethanol production.