Search

LiDAR is a promising tool for fast and accurate measurements of trees. There are several approaches to estimate aboveground woody biomass using LiDAR point clouds. One of the most widely used method consists in fitting geometric primitives (e.g. cylinders) to the point cloud, thereby reconstructing both the geometry and topology of the tree. However, current algorithms are not suited for accurate estimation of the biomass of finer branches, because of the unreliable point dispersions from the movements induced by wind, occlusion in the upper canopy, or the relatively large laser footprint compared to the structure diameter. We propose a new method that couples point cloud-based reconstructions and structural models to estimate accurately the aboveground woody biomass of trees, including finer branches. The model was trained using branch samples from the trees, and accurately predicted the biomass with 1.6% nRMSE at the segment scale from a k-fold cross-validation. It also gave satisfactory results when up-scaled to the branch level with a significantly lower error (13% nRMSE) and bias (-5%) compared to fitting cylinders to the point cloud (nRMSE: 92%), or using the pipe model theory (nRMSE: 31%). The model was then applied to the whole-tree scale, and showed that the sampled trees had more than 1.7km of structures in average, and that 96% of that length was coming from the twigs (i.e.

L'agroforesterie consiste à faire pousser des arbres en association avec des cultures agricoles et / ou en présence de troupeaux d'animaux. Les avantages d'une telle pratique sont nombreux, tant du point de vue de la production agricole que de la protection de l'environnement. Sur le terrain, on observe une meilleure productivité individuelle des arbres agroforestiers mesurée (jusqu'à trois fois le volume raméal des mêmes arbres mais en conditions forestières). La récolte de ces branches est une nécessité pour la bonne gestion agricole. La valorisation de cette biomasse permettrait d'améliorer la rentabilité des parcelles agroforestières et d'inciter les agriculteurs à intégrer davantage les arbres dans leur environnement de production. Et pour les industriels, ce serait également la possibilité d'obtenir plus facilement et durablement des matières premières de qualité dans des démarches agroécologiques reconnues. Le projet Agrobranche visait à étudier la valorisation de cette biomasse dans les domaines des matériaux et de la chimie biosourcés. Agrobranche a permis de quantifier la productivité de ces systèmes et de définir les contours de leur rentabilité. Les partenaires aval ont pu qualifier le potentiel industriel dans les domaines des matériaux composites et panneaux, tout en explorant les principales caractéristiques chimiques des plaquettes agroforestières selon 4 essences principales (noyer, peuplier, châtaignier et chêne), mais aussi à partir de bois déchiqueté issus de plateformes de bois énergie (mix d'essences). Le comportement des matériaux testés est très encourageant, avec des performances souvent égales aux standards issus de matériaux comme l'épicéa forestier ou le miscanthus. Mais l'intérêt, qui a pu être approché dans l'ACV, réside également dans le mode de production. Car produire de la plaquette agroforestière concilie production biomasse et maintien de la production agricole contrairement aux productions biomasses dédiées. Et les services rendus pèsent dans la balance auprès des acteurs de la filière, qu'ils soient les clients finaux comme les intermédiaires.

International audience; STICS is a dynamic soil-crop model capable of simulating diverse crop rotations over short and long-term (Brisson et al., 2003). Intercropping increases the complexity of the system by adding inter-species competition. Crop models are useful tools for analyzing complex systems, as they allow the user far more control over individual variables than is possible in field experiments. An initial version of the STICS intercrop model was created by Brisson et al. (2004) from the standard version (Brisson et al., 2003). Recently Vezy et al. (2020) improved this initial version [STICS-Intercrop_v2020] by adding new formalisms and replacing some equations which were not found relevant during the first step of testing with our database. The aim of paper is to present the STICS-Intercrop_v2020 functions and to show the quality of simulations obtained for a wide range of winter and spring intercrop mixtures of durum wheat - winter pea and barley - spring pea. STICS-Intercrop_v2020 was tested for two sites in France using various plant densities and N-fertilizer rates in order to determine its relevance and validity domain for simulating intercropping. This research work was done in 3 main steps: 1) improving the existing formalisms and introducing new equations in the model algorithm for improving the light sharing and nitrogen competition between the two species, and then creating the new version STICS-Intercrop_v2020; 2) parameterising the model for sole crops only; and 3) independently evaluating the quality of predictions for intercrops without any re-parametrisation. The results of parametrisation were satisfactory with low Root Mean Square Errors and high Model Efficiencies, illustrating the robustness and accuracy for sole crops, as already shown for many crops, pedoclimatioc conditions and agronomic managements (Brisson et al., 2003). STICS-Intercrop_v2020 was reliably and efficient to simulate inter-specific interactions, development and growth variables, and provided coherent results for predicting yield and grain protein content for the two species, in winter and spring bi-specific cereal-legume intercrops, without any specific parameterization for intercropping. This illustrates the relevance of the formalisms to simulate dynamically both intra and inter-specific plant interactions. These results are very encouraging for using the STICS-intercrop model for future work where virtual experiments will help us to analyse the performance (LER, level of production in low inputs systems, etc.) and resilience of sole crops versus intercrops according to management practices, pedoclimatic variability and climate change scenarios. This work was part of the ReMIX project funded by EU H2020 program.

Ocho años de estudio de la ecofisiología del café, a través de experimentación y de modelación y el monitoreo de los servicios del ecosistema (SE) en una gran finca cafetalera en Costa Rica, revelaron varias recomendaciones prácticas para los agricultores y los formuladores de políticas. El sistema de cultivo estudiado dentro de nuestro observatorio colaborativo (Coffee-Flux), corresponde a un sistema agroforestal (SAF) a base de café bajo la sombra de grandes árboles de Erythrina poeppigiana (16% de la cubierta del dosel). Una gran cantidad de SE y limitantes dependen de las propiedades locales del suelo (en este caso Andisoles), especialmente de la erosión/infiltración, el agua/carbono y la capacidad de almacenamiento de nutrientes. Por lo tanto, para la evaluación de SE, el tipo de suelo es crucial. Una densidad adecuada de árboles de sombra (bastante baja aquí por la condición de libre crecimiento), redujo la severidad de las enfermedades de las hojas con la posibilidad de reducir el uso de plaguicidas y fungicidas. Un inventario simple del área basal en el collar de las plantas de café permitió estimar la biomasa subterránea y la edad promedio de la plantación, para juzgar su valor de mercado y decidir cuándo reemplazarla. Las fincas de café probablemente estén mucho más cerca de la neutralidad de C que lo indicado en el protocolo actual de C-neutralidad, que solo considera árboles de sombra, no los cafetos ni el suelo. Se proponen evaluaciones más completas, que ncluyen árboles, café, hojarasca, suelo y raíces en el balance C del SAF. Los árboles de sombra ofrecen muchos SE si se gestionan adecuadamente en el contexto local. En comparación con las condiciones a pleno sol, los árboles de sombra pueden (i) reducir la erosión laminar en un factor de 2; (ii) aumentar la fijación de N y el % de N reciclado en el sistema, reduciendo así los requisitos de fertilizantes; (iii) reducir la severidad de enfermedades de las hojas; (iv) aumentar el secuestro de C; (v) mejorar el microclima y (vi) reducir sustancialmente los efectos del cambio climático. En nuestro estudio de caso, no se encontró ningún efecto negativo sobre el rendimiento del café.

International audience; Functional-structural plant modelling approaches (FSPM) open the way for exploring the relationships between the 3D structureandthe physiological functioningof plantsin relation to environmental conditions.FSPMscan beparticularly interesting when dealing with perennial crops like oil palm, for which research on innovative management practices requires long and expensive agronomic trials. The presentstudy is part of the PalmStudio project, whichaimsat developing a FSPMforoil palmcapable of conducting virtual experiments to test the relevance of innovative management practicedsand/or design ideotypes.We propose amethodologicalapproach which integrates architectural responses to planting densityin an existing oil palmFSPM(Perez et al.2018a b). Combining standard field phenotyping with Lidar-based derived measurements, wemanage toevaluate the phenotypic plasticity of the main parameters required for the calibration of the 3D plant model.LiDAR scans were processedusing the PlantScan3D software (Boudon et al. 2014)to derivephenotypic traits of leaf geometry that were compared to labour-intensive measurements. Density-based allometriesof leaf geometry and biomassarethenderived from theobserved variations in phenotypic traits and integrateintothe FSPM.Our results illustratethe accuracy and the efficiency of Lidar-based phenotypingofleaf geometrical traits. In average,we find less than 3%of difference in leaf dimensions (i.e.rachis length) in comparison with traditional hand-made field measurements.The fastand efficient measurements ofusuallylabour-intensive traits such as leaf curvatureallowed estimating the plasticity of leaf geometry in response to density. We find that the main traits affected by density were leaf dimensions(up to 15% and 25% of increase in rachislength and petiolelengthrespectively)and curvature(15% of increase in leaf erectness-related parameter), whereas other structural traits like the number of leaflets per leaf remained unchanged. Simple density-based allometric relationships were thenmodelled andcombined with the existing allometric-based 3D oil palm model VPalm (Perez et al.2018a). These data also enablethe development and the integrationin VPalm ofa biomechanical model simulating leaf curvature.The methodology presented in this study paves the way for a rapid integration of phenotypic plasticityin FSPMs. OurFSPM is now able toestimatehow planting density affectnot only plant architecture but alsofunctional processes such as carbon assimilation and transpiration.Ongoing research aims at coupling the current FSPM with a carbon allocation model (Pallas et al.2013)tosimulate the retroactions of functioning processes on plant architecture together with environmental and agronomic conditions

International audience; Introduction Light modelling at the scale of organs is essential to account accurately for the complex interactions between biophysical processes such as photosynthesis, stomatal conductance and energy balance. Yet, the calculation of radiative exchanges at fine scales is computationally-intensive and it remains a hindrance to a widespread use of FSPMs despite advances in light modelling using either radiosity (Chelle and Andrieu, 1998) or raytracing (Bailey, 2018). This study shows that simplifications based on the discretization of radiative fluxes allow processing radiative exchanges in a natural environment while maintaining good accuracy on the simulation of biophysical processes such as carbon assimilation.

National audience; Introduction STICS is a soil-crop model capable of simulating crops in succession (Brisson et al., 2003). Intercropping occurs when multiple species are grown simultaneously on the same field. There has been a growing interest in adapting this traditional technique for modern agriculture as a way of ecological intensification, especially for combining leguminous and cereal crops in order to reduce N inputs and potential environmental damage through N losses. Intercropping adds complexity to the system by adding inter-species competition. Crop models are useful tools for analyzing complex systems, as they allow the user far more control over individual variables than is possible in field experiments. A first version of the STICS intercrop model was created by Brisson et al. (2004) and was recently improved by Vezy et al. (2020). The aim of this study was to calibrate and evaluate this improved STICS-Intercrop model by simulating a winter and a spring intercrop mixture: durum wheat-winter pea and barley-spring pea. Materials and Methods: The data set used for modelling comprised of four years of wheat (Triticum turgidum L.) and pea (Pisum sativum L.) field data from Auzeville, France with multiple levels of nitrogen fertilizer, and four years of barley (Hordeum vulgare L.) and pea field data from Angers, France (Corre-Hellou, 2005), which in some years included two levels of nitrogen fertilizer and two different plant densities of the intercrops. The sole crop trials were used for calibration and the intercrop trials for evaluation, except for a subset of intercrop data that was used to calibrate the parameters unique to the intercrop model. The assumption was that parameters common to both sole and intercropping, such as plant-soil interactions and phenology, would be the same for both. The optimization method used for calibration was based on Wallach et al. (2011). The parameters were broken down into 15 groups (16 for pea including nitrogen fixation) for calibration, each corresponding to a different process. Results and Discussion: The model calibration process is still ongoing. The root mean square error (RMSE) for shoot biomass was 1.92 t/ha for winter pea and 1.37 t/ha for durum wheat. The RMSE for grain yield was 1.84 t/ha for spring pea and 1.15 t/ha for barley. Overall the model captured the dominancy of one species quite well, however the accuracy has to be increased. The phenology and height were correctly simulated. Some of the discrepancies could be due to biological stresses that STICS does not capture. Some parameters for the pea-wheat model have not been calibrated yet, so the RMSE is likely to improve. Conclusions: Intercrop systems are difficult to model in comparison to sole crops due to their complex interactions, but the STICS model, after having completed the calibration, could be a useful tool for better understanding the biological functions of this management practice.

Agro-silvo-pastoralism is a highly representative Land Use in Africa, often presented as a strategical option for ecological intensification of cropping systems towards food security and sovereignty.We set up a new long-term observatory (“Faidherbia-Flux”) to monitor and model microclimate, energy and C balance in Niakhar (central Senegal, rainfall ~ 500 mm), dominated by the multipurpose tree Faidherbia albida (12.5 m high; 7 tree ha-1; 5% canopy cover). Faidherbia is an attractive agroforestry tree species in order to partition fluxes, given that it is on leaf during the dry season (October-June) and defoliated during the wet season, just when crops take over. Pearl-millet and groundnut crops were conducted during the wet season, following annual rotation in a complex mixed mosaic of ca. 1 ha fields.Early 2018, we installed an eddy-covariance (EC) tower above the whole mosaic (EC1: 20 m high). A second EC system was displayed above the crop (EC2: 4.5 m if pearl-millet, 2.5 m if groundnut) in order to partition ecosystem EC fluxes between tree layer and crop+soil layers. Sap-flow was monitored from April 2019 onwards in 5 faidherbia trees (37 sensors).The ecosystem displayed moderate but significant daily CO2 and H2O fluxes during the dry season, when faidherbia (low canopy cover) was in leaf and the soil was evaporating. At the onset of the rainy season, the soil bursted a large amount of CO2. Just after the growth of pearl-millet in 2018, CO2 uptake by photosynthesis increased dramatically. However, this was largely compensated by high ecosystem respiration. Surprisingly in 2019, although the crop was turned to groundnut, the fluxes behaved pretty much the same as with pearl millet in 2018: comparing annual balances between 2018 and 2019 we obtained [454, 513] for rainfall (P: mm yr-1), [3500, 3486] for potential evapotranspiration (ETo: mm yr-1), [0.13, 0.15] for P/ETo, [470, 497] for actual evapotranspiration (E: mm yr-1), [2809, 2785] for net radiation (Rn: MJ m-2 yr-1), [1686, 1645] for sensible heat flux (H: MJ m-2 yr-1), [-3.2, -2.8] for net ecosystem exchange of C (NEE: tC ha-1 yr-1), [-11.8, -11.1] for gross primary productivity (GPP: tC ha-1 yr-1) and [8.6, 8.3] for ecosystem respiration (Re: tC ha-1 yr-1). The energy balance (Rn-H-LE) was nearly nil indicating that the EC system behaved reasonably. E was very close to P, indicating that little or no water would recharge the deep soil layers.Now comparing the dry (2/3 of the year) and wet (1/3) seasons: surprisingly, NEE was more effective during the dry season [-3.9, -1.7]. This was the result of Re being much lower on a daily basis as well as cumulated over the entire seasons [57, 84], whereas GPP was similar [-10.8, -12.1].We found a good match between E measured above the whole ecosystem (EC1), and the sum of tree transpiration (T, measured by sapflow) + E measured just above crops + soil (EC2) throughout the wet and dry seasons.The “Faidherbia-Flux” observatory is registered in FLUXNET as SN-Nkr and is widely open for collaboration.

International audience; Abstract. The mechanistic model GO+ describes the functioning and growth of managed forests based upon biophysical and biogeochemical processes. The biophysical and biogeochemical processes included are modelled using standard formulations of radiative transfer, convective heat exchange, evapotranspiration, photosynthesis, respiration, plant phenology, growth and mortality, biomass nutrient content, and soil carbon dynamics. The forest ecosystem is modelled as three layers, namely the tree overstorey, understorey and soil. The vegetation layers include stems, branches and foliage and are partitioned dynamically between sunlit and shaded fractions. The soil carbon submodel is an adaption of the Roth-C model to simulate the impact of forest operations. The model runs at an hourly time step. It represents a forest stand covering typically 1 ha and can be straightforwardly upscaled across gridded data at regional, country or continental levels. GO+ accounts for both the immediate and long-term impacts of forest operations on energy, water and carbon exchanges within the soil–vegetation–atmosphere continuum. It includes exhaustive and versatile descriptions of management operations (soil preparation, regeneration, vegetation control, selective thinning, clear-cutting, coppicing, etc.), thus permitting the effects of a wide variety of forest management strategies to be estimated: from close to nature to intensive. This paper examines the sensitivity of the model to its main parameters and estimates how errors in parameter values are propagated into the predicted values of its main output variables.The sensitivity analysis demonstrates an interaction between the sensitivity of variables, with the climate and soil hydraulic properties being dominant under dry conditions but the leaf biochemical properties being most influential with wet soil. The sensitivity profile of the model changes from short to long timescales due to the cumulative effects of the fluxes of carbon, energy and water on the stand growth and canopy structure. Apart from a few specific cases, the model simulations are close to the values of the observations of atmospheric exchanges, tree growth, and soil carbon and water stock changes monitored over Douglas fir, European beech and pine forests of different ages. We also illustrate the capacity of the GO+ model to simulate the provision of key ecosystem services, such as the long-term storage of carbon in biomass and soil under various management and climate scenarios.

Introduction STICS is a soil-crop model capable of simulating crops in succession (Brisson et al., 2003). Intercropping occurs when multiple species are grown simultaneously on the same field. There has been a growing interest in adapting this traditional technique for modern agriculture as a way of ecological intensification, especially for combining leguminous and cereal crops in order to reduce N inputs and potential environmental damage through N losses. Intercropping adds complexity to the system by adding inter-species competition. Crop models are useful tools for analyzing complex systems, as they allow the user far more control over individual variables than is possible in field experiments. A first version of the STICS intercrop model was created by Brisson et al. (2004) and was recently improved by Vezy et al. (2020). The aim of this study was to calibrate and evaluate this improved STICS-Intercrop model by simulating a winter and a spring intercrop mixture: durum wheat-winter pea and barley-spring pea. Materials and Methods: [br/]The data set used for modelling comprised of four years of wheat (Triticum turgidum L.) and pea (Pisum sativum L.) field data from Auzeville, France with multiple levels of nitrogen fertilizer, and four years of barley (Hordeum vulgare L.) and pea field data from Angers, France (Corre-Hellou, 2005), which in some years included two levels of nitrogen fertilizer and two different plant densities of the intercrops. The sole crop trials were used for calibration and the intercrop trials for evaluation, except for a subset of intercrop data that was used to calibrate the parameters unique to the intercrop model. The assumption was that parameters common to both sole and intercropping, such as plant-soil interactions and phenology, would be the same for both. The optimization method used for calibration was based on Wallach et al. (2011). The parameters were broken down into 15 groups (16 for pea including nitrogen fixation) for calibration, each corresponding to a different process. Results and Discussion: The model calibration process is still ongoing. The root mean square error (RMSE) for shoot biomass was 1.92 t/ha for winter pea and 1.37 t/ha for durum wheat. The RMSE for grain yield was 1.84 t/ha for spring pea and 1.15 t/ha for barley. Overall the model captured the dominancy of one species quite well, however the accuracy has to be increased. The phenology and height were correctly simulated. Some of the discrepancies could be due to biological stresses that STICS does not capture. Some parameters for the pea-wheat model have not been calibrated yet, so the RMSE is likely to improve. Conclusions: Intercrop systems are difficult to model in comparison to sole crops due to their complex interactions, but the STICS model, after having completed the calibration, could be a useful tool for better understanding the biological functions of this management practice.

The mechanistic model GO+ describes the functioning and growth of managed forests based upon biophysical and biogeochemical processes. The biophysical and biogeochemical processes included are modelled using standard formulations of radiative transfer, convective heat exchange, evapotranspiration, photosynthesis, respiration, plant phenology, growth and mortality, biomass nutrient content, and soil carbon dynamics. The forest ecosystem is modelled as three layers, namely the tree overstorey, understorey and soil. The vegetation layers include stems, branches and foliage and are partitioned dynamically between sunlit and shaded fractions. The soil carbon submodel is an adaption of the Roth-C model to simulate the impact of forest operations. The model runs at an hourly time step. It represents a forest stand covering typically 1 ha and can be straightforwardly upscaled across gridded data at regional, country or continental levels. GO+ accounts for both the immediate and long-term impacts of forest operations on energy, water and carbon exchanges within the soil–vegetation–atmosphere continuum. It includes exhaustive and versatile descriptions of management operations (soil preparation, regeneration, vegetation control, selective thinning, clear-cutting, coppicing, etc.), thus permitting the effects of a wide variety of forest management strategies to be estimated: from close to nature to intensive. This paper examines the sensitivity of the model to its main parameters and estimates how errors in parameter values are propagated into the predicted values of its main output variables.The sensitivity analysis demonstrates an interaction between the sensitivity of variables, with the climate and soil hydraulic properties being dominant under dry conditions but the leaf biochemical properties being most influential with wet soil. The sensitivity profile of the model changes from short to long timescales due to the cumulative effects of the fluxes of carbon, energy and water on the stand growth and canopy structure. Apart from a few specific cases, the model simulations are close to the values of the observations of atmospheric exchanges, tree growth, and soil carbon and water stock changes monitored over Douglas fir, European beech and pine forests of different ages. We also illustrate the capacity of the GO+ model to simulate the provision of key ecosystem services, such as the long-term storage of carbon in biomass and soil under various management and climate scenarios.

Huit ans de travaux de recherche sur les services écosystémiques dans une grande ferme caféière du Costa Rica (observatoire collaboratif Coffee-Flux, en système agroforestier à base de caféiers sous de grands arbres d'Erythrina poeppigiana, surface projetée de couronne de l'ordre de 16 %) ont suggéré plusieurs applications pour les agriculteurs et les décideurs. Il est apparu que de nombreux services écosystémiques dépendaient des propriétés du sol (ici des Andisols), en particulier de l'érosion, de l'infiltration, de la capacité de stockage de l'eau et des éléments nutritifs. Nous confirmons qu'il est essentiel de lier les services hydrologiques et de conservation au type de sol en présence. Une densité adéquate d'arbres d'ombrage (plutôt faible ici) permet de réduire la sévérité des maladies foliaires avec, en perspective, une réduction de l'usage de pesticides-fongicides. Un simple inventaire de la surface basale au collet des caféiers permet d'estimer la biomasse souterraine et la moyenne d'âge d'une plantation de caféiers, ce qui permet d'évaluer sa valeur marchande ou de planifier son remplacement. Le protocole de calcul actuel pour la neutralité carbone des systèmes agroforestiers ne prend en compte que les arbres d'ombrage, pas la culture intercalaire. Dans la réalité, si on inclut les caféiers, on se rapproche très probablement de la neutralité. Des évaluations plus complètes, incluant les arbres, les caféiers, la litière, le sol et les racines dans le bilan en carbone du système agroforestier sont proposées. Les arbres d'ombrage offrent de nombreux servies écosystémiques s'ils sont gérés de manière adéquate dans le contexte local. Par rapport aux parcelles en plein soleil, nous montrons qu'ils réduisent l'érosion laminaire d'un facteur 2, augmentent la fixation de l'azote (N2) atmosphérique et le pourcentage d'azote recyclé dans le système, réduisant ainsi les besoins en engrais. Ils réduisent aussi la sévérité des maladies foliaires, augmentent la séquestration de carbone, améliorent le microclimat et atténuent substantiellement les effets des changements climatiques. Dans notre étude de cas, aucun effet négatif sur le rendement n'a été enregistré.

The adaptation of semi-arid crops to climate changes is theoretically possible in agroforestry systems (AFS), provided that the trees exert little competition, or even increase the land equivalent ratio (LER). However, the measurement of LER requires sole crop plots which are usually not available in AFS. We designed a geostatistical method to estimate the distance of influence of trees on crops and selected sole crop plots at even higher distances, in order to compute LER. We monitored microclimate, net primary productivity (NPP), CO2 and H2O fluxes in a semi-arid agro-silvo-pastoral system (Niakhar, Senegal), dominated by the multipurpose Faidherbia albida (FA) tree. Undercrops were mainly millet and peanut, under annual rotation. Just before harvest, we scanned a ca. 1 ha millet plot under FA with UAV photogrammetry (drone) in RGB, thermal infrared and multispectral bands. At harvest, we collected 12 subplots, distributed either below the crown of FA, or at 2.5 x the crown radius (CR), or at 5 CR. We separated all organs. The whole millet root system (0-200 cm) was sampled also in 2 m trenches. Whole plot harvest allowed extrapolating yield from subplots, through UAV images. Millet yield per unit ground area was about 3 times higher below FA. The geostatistical method indicated that NDVI was not affected by the FA trees at 5 times the crown radius. LER of millet was computed 1.4, confirming the positive effect FA at distances up to 18 m. The “Faidherbia-Flux” observatory is open for collaboration.

Although large amounts of potassium (K) are applied in tropical crops and planted forests, little is known about the interaction between K nutrition and water supply regimes on water resources in tropical regions. This interaction is a major issue because climate change is expected to increase the length of drought periods in many tropical regions and soil water availability in deep soil layers is likely to have a major influence on tree growth during dry periods in tropical planted forests. In this study, we described a modeling approach to quantify water fluxes in a Eucalyptus throughfall exclusion experiment in Brazil to gain insight into the combined effects of K deficiency and rainfall reduction (37% throughfall exclusion) on the water used by the trees, soil water storage and water table fluctuations over the first 4.5 years after planting. Although the mean water withdrawal from depths of over 10 m amounted to only 5% of canopy transpiration in K-fertilized plantation with undisturbed rainfall (+K+W), the proportion of water taken up near the water-table was much higher during dry periods. Under contrasted K availability, water withdrawal was more superficial for -K than for +K. Under rainfall exclusion, water was withdrawn in deeper soil layers for -W than for +W, particularly over dry seasons. A comparison of canopy transpiration in each plot with the values predicted for the same soil with the water content maintained at field capacity, made it possible to calculate a soil-driven tree water stress index for each treatment. The soil-driven tree water stress index was 166% higher over the first 4.5 years after planting for -W than for +W, 76% lower for -K than for +K, and 14% lower for -K-W than for +K+W. Over the study period, deep seepage was higher by 371 mm yr-1 (+122%) for -K than for +K and lower by 200 mm yr-1 (-66%) for -W than for +W. Deep seepage was lower by 44% for -K-W than for +K+W. At the end of the study period, the model predicted a higher water table for -K (10 mbs for -K+W and 16 mbs for -K-W) than for +K (16 mbs for +K+W and 18 mbs for +K-W). Our study suggests that the depth of the soil should be a major criterion for the selection of future afforestation areas and that flexible fertilization regimes could contribute to adjusting the local trade-off between wood production and demand for soil water resources in planted forests.

In this thesis, we used two complementary mathematical models to simulate the future behavior of coffee plantations under climate change (1979 - 2100). We studied their carbon, water, and energy balances to better understand and predict the effects of these changes on coffee production. The addition of shade trees above the coffee layer leadto higher yield compared to full sun management under increased temperature.However, coffee yield was predicted to decrease compared to current levels by 2100,whatever the shade tree species or management.; Dans le cadre de cette thèse, nous avons utilisé deux modèles mathématiques complémentaires pour simuler le comportement futur des plantations de café sous conditions actuelles ainsi que sous changements climatiques (1979 -2100). Nous avons étudié leurs bilans de carbone, d'eau et d'énergie pour mieux comprendre et prévoir les effets des changements sur la production de café. Comparativement à une plantation en plein soleil, l'ajout d'arbres d'ombrage au dessus des caféiers pourrait permettre d'augmenter les rendements lorsque la température augmente. Cependant, les rendements en grain de caféiers à l'horizon 2100 sont prédits inférieurs aux rendements actuels quelle que soit l'espèce d'arbres d'ombrage ou sa gestion.

Eight years of monitoring ecophysiology and ecosystem services (ES) in a large coffee farm of Costa Rica yields a range of practical applications for the farmer and stakeholders, thanks to numerous scientific actors and disciplines contributing to our collaborative observatory (Coffee-Flux). • A lot of ecosystem services depend on the soil properties, such as runoff/infiltration, water and nutrient storage capacity. It is essential to relate hydrological and soil conservation services to the soil type, since this might have even more importance than the crop itself for ES. Regarding the use of fertilizer, we show that some soils may have a large storage capacity, allowing producing coffee at normal yields with just a reduced, or even a minimum amount of fertilizers, for instance when the economic conditions are unfavorable. Also, due to the soil variability within the farm, it is possible to adjust fertilization to micro-local conditions and reduce the total expenses and risks of leaching of N to the environment. VNIRS and MIR are promising broadband tools for screening the variability in soils. Adjusting N fertilizer to the optimum will also considerably reduce the N2O emissions and improve the GHG balance of the farm. • Pesticides-fongicides: we show that an adequate amount of shade trees allows reducing the severity of the whole complex of leaf diseases. This also should reduce expenses and impacts on the ecosystem. • Roots: a simple survey of basal area at collar allows estimating the belowground biomass and the average age of a plantation, to judge of its market value and to decide when to replace it. • Also starch plays a key role in the trophic equilibrium between the perennial parts of the coffee plant (aerial stump, belowground stump, coarse roots) and its ephemeral parts (resprout, leaves, fruits, fine roots). Coffee plants accumulate starch in the stumps by the end of the life of their resprout, as a strategy for survival. Breeding plants with less starch build-up capacity would probably allow increasing the fraction of productive years during the lifespan of the resprouts. • Coffee farms are probably much closer to C neutrality than currently admitted using the C-Neutrality protocol. We stress the prevailing role of coffee plants + litter + soil in the ecosystem C balance. If those are excluded from the calculations as done so far, coffee farms are GHG sources, by definition. We argue that either full assessments (as proposed here, at the ecosystem level, including trees, coffee, litter, soil and roots) or consensus on “sequestration factors” (the counterpart of emission factors) would allow performing a more realistic assessment of the GHG balance. • Finally, we bring new data confirming that shade trees offer numerous ecosystem services, when adequately managed for the local context. As compared to full sun conditions, they may (i) reduce laminar erosion by a factor of ca. 2, (ii) increase the atmospheric N2 fixation and the % of N recycled into the system, thus reducing the fertilizer requirements, (iii) reduce the severity of the leaf disease complex, (iv) increase C sequestration, (v) improve the microclimate, and (vi) be a large part of the solution to face climate changes. All this is possibly without negative effects on profitability or yield, if managed properly. In our particular case-study, we encount.