Search

Ground cover management (GCM) is an important agricultural practice used to reduce weed growth, erosion and runoff, and improve soil fertility. In the present study, an approach to account for GCM is proposed in the modeling of pesticide emissions to evaluate the environmental sustainability of agricultural practices. As a starting point, we include a cover crop compartment in the mass balance of calculating initial (within minutes after application) and secondary (including additional processes) pesticide emission fractions. The following parameters were considered: (i) cover crop occupation between the rows of main field crops, (ii) cover crop canopy density, and (iii) cover crop family. Two modalities of cover crop occupation and cover crop canopy density were tested for two crop growth stages, using scenarios without cover crops as control. From that, emission fractions and related ecotoxicity impacts were estimated for pesticides applied to tomato production in Martinique (French West Indies) and to grapevine cultivation in the Loire Valley (France). Our results demonstrate that, on average, the presence of a cover crop reduced the pesticide emission fraction reaching field soil by a factor of 3 compared with bare soil, independently of field crop and its growth stage, and cover crop occupation and density. When considering cover exported from the field, ecotoxicity impacts were reduced by approximately 65% and 90%, compared with bare soil for grapevine and tomato, respectively, regardless of the emission distribution used. Because additional processes may influence emission distributions under GCM, such as runoff, leaching, or preferential flow, further research is required to incorporate these processes consistently in our proposed GCM approach. Considering GCM in pesticide emission modeling highlights the potential of soil cover to reduce pesticide emissions to field soil and related freshwater ecotoxicity. Furthermore, the consideration of GCM as common farming practice allows the modeling of pesticide emissions in intercropping systems. Integr Environ Assess Manag 2021;00:1–15. © 2021 The Authors. Integrated Environmental Assessment and Management published by Wiley Periodicals LLC on behalf of Society of Environmental Toxicology & Chemistry (SETAC).

Researchers worldwide are expected to design and develop agroecological systems to address major challenges such as increasing biotic pressure and climate change. But to design effective innovations, researchers should integrate new cropping techniques into wider agricultural systems such as cropping systems, innovative farms, alternative food systems, or multifunctional landscapes. This integration requires a long process of exploration in which the object under design can transform and shift to a different organisational scale. In this article, we wish to introduce a new analytical framework highlighting the systemic mechanisms involved in the scale transformations of design objects along agroecological design processes. We conceptualise an agroecological design process as a non-linear unpredictable process in which four components—a science consortium, non-scientific actors in the field, a problem situation and a design object—interact, and co-evolve through time. The scale transformations of the design objects and their drivers are the results of interactions and knowledge flows between these four components. This analytical framework was tested and further elaborated through ex-post analysis of design processes in three contrasting case studies in the context of tropical horticulture. Data were collected through literature review, interviews, and focus group discussions with the researchers involved in the three design processes. Future design methods should take full account of the lengthy, non-linear, and transformational nature of agroecological design processes and the coexistence of different types of knowledge—holistic vs. reductionist, ecosystem focused vs. human-system focused—which can interconnect and nourish each other.

Purpose Current field emission modelling and toxicity characterisation of pesticides suffer from several shortcomings like mismatches between LCI databases and LCIA methods, missing characterisation factors, missing environmental compartments, and environmental impact pathways. The OLCA-Pest project was implemented to address these aspects and to operationalise the assessment of pesticides in LCA. Based on this effort, we propose an approach to integrate pesticide emissions into LCI databases. Methods The PestLCI Consensus Model has been developed in order to estimate emission fractions to different environmental compartments. The initial distribution fractions should be linked to the compartments air, agricultural soil, natural soil, and freshwater. Emissions to off-field surfaces are hereby distributed between agricultural soil, natural soil, and freshwater by using surface cover data. Deposition on the crop surface should be recorded in an emission compartment crop with 13 sub-compartments for crop archetypes for both food and non-food uses. Default emission fractions are provided to calculate the emission fractions for different pesticide application scenarios. Results and discussion A sensitivity analysis shows the effects of the application technique, drift reduction, crop and development stage, field width, and buffer zone on the initial distribution fractions of field-applied pesticides. Recommendations are given for the implementation of a set of default initial distribution fractions into LCI databases, for the organisation of metadata, and for the modelling of pesticide residues in food along the supply chain (processing, storage). Priorities for further research are: improving the modelling of pesticide secondary emissions, further extending emission modeling (e.g. additional application techniques, including cover crops), considering metal-based pesticides in emission models, and systematically assessing human health impacts associated with pesticide residues in food crops. Conclusions The proposed approach allows to preserve the mass balance of the pesticide emitted after application, to make a consistent assessment of ecotoxicity and human toxicity, to define a clear and consistent interface between the LCI and LCIA phases, to estimate initial emission distribution fractions based on existing data, to document metadata transparently and efficiently within crop datasets, and to model the removal of pesticide residues in food during processing.

PurposeIn tropical cropping systems, pesticides are extensively used to fight pests and ensure high crop yields. However, pesticide use also leads to environmental and health impacts. While pesticide emissions and impacts are influenced by farm management practices and environmental conditions, available Life Cycle Inventory (LCI) emission models and Life Cycle Impact Assessment (LCIA) toxicity characterization models are generally designed based on temperate conditions. There is, hence, a need for adapting LCI and LCIA models for evaluating pesticides under tropical conditions. To address this need, we aim to identify the characteristics that determine pesticide emissions and related impacts under tropical conditions, and to assess to what extent LCI and LCIA models need to be adapted to better account for these conditions.MethodsWe investigated the state-of-knowledge with respect to characteristics that drive pesticide emission patterns, environmental fate, human and ecological exposures, and toxicological effects under tropical conditions. We then discuss the applicability of existing LCI and LCIA models to tropical regions as input for deriving specific recommendations for future modeling refinements.Results and discussionOur results indicate that many pesticide-related environmental processes, such as degradation and volatilization, show higher kinetic rates under tropical conditions mainly due to higher temperatures, sunlight radiation, and microbial activity. Heavy and frequent rainfalls enhance leaching and runoff. Specific soil characteristics (e.g., low pH), crops, and cropping systems (e.g., mulching) are important drivers of distinct pesticide emission patterns under tropical conditions. Adapting LCI models to tropical conditions implies incorporating specific features of tropical cropping systems (e.g., intercropping, ground cover management), specific drift curves for tropical pesticide application techniques, and better addressing leaching processes. The validity domain of the discussed LCI and LCIA models could be systematically extended to tropical regions by considering tropical soil types, climate conditions, and crops, and adding active substances applied specifically under tropical conditions, including the consideration of late applications of pesticides before harvest and their effect on crop residues and subsequent human intake.ConclusionsCurrent LCI and LCIA models are not fully suitable for evaluating pesticide emissions and impacts for crops cultivated in tropical regions. Models should be adapted and parameterized to better account for various characteristics influencing emission and impact patterns under tropical conditions using best available data and knowledge. Further research is urgently required to improve our knowledge and data with respect to understanding and evaluating pesticide emission and impact processes under tropical conditions.

On the request of decision-makers, the Life Cycle Assessment (LCA) framework is increasingly applied in agri-food value chains in developing countries under particularly demanding constraints. Based on the cradle-to-farm-gate LCA results of the tomato in Rwamagana district in Rwanda commissioned by the European Union, the main objectives of the paper were to validate statistically the differences in environmental impacts among expert-based types for this crop based on location and season and identify the key-drivers of these impacts. The study was developed thanks to two intensive field visits and a survey of 15 plots. The functional unit was one kg of tomato at farm-gate. Combining several statistical analyses allowed identifying three groups of plots with contrasting input profile and eco-efficiency. For most impact categories, the first group with mainly plots in marshland during wet season had the lowest impacts and the third group with plots in hillside showed the highest impacts. The yield of tomato being significantly different between marshland and hillside plots was an important driver of these results. The second group composed of plots in marshland during the dry season generally showed intermediate impacts due to the withdrawing of water for irrigation. The second group obtained a higher freshwater ecotoxicity due to a more intensive use of toxic insecticides. The factorial analysis for mixed data (FAMD) confirmed the importance of the location and the season for the eco-efficiency of tomato plots and the hierarchical clustering on principle components (HCPC) separated the tomato plots into three clusters. The generalized linear models (GLMs) validated the differences in the environmental impacts per kg of tomato between the clusters. The principal component analysis (PCA) combined to GLM revealed that only the use of water was significantly different between the three groups. Compared to existing datasets, all groups showed high freshwater ecotoxicity impacts due to the use of toxic insecticides and the excessive use of mancozeb. The third group also showed a high freshwater eutrophication in relation to P losses due to erosion and low yield. From a methodological point of view, we demonstrated in this paper that using expert-based typologies combined with adapted statistical analyses constituted a relevant approach under such circumstances.

Although challenging, private and public decision-makers increasingly demand for quantitative assessments of the environmental performance of value chains in South contexts. This paper presents and critically analyzes a complete LCA study performed with Endpoint indicators for a public decision-maker for the fresh French bean (FB) value chain of Kenya. A cradle-to-market-gate LCA study was done including five main stages: agricultural production, transport by road before pack-house, pack-house, transport by road after pack-house, intercontinental transport by air-freight and using 1 kg of raw French bean processed as functional unit. Supported by local experts, primary data were collected for all inputs and outputs for 33 farms over 5 counties and 2 pack-houses. An expert-based typology defined four farm types: large-farm, medium-farm, small-holder farm (SHF) contracted and small-holder farm scattered. Best available methods for field emissions were used and adapted when possible to local conditions (e.g. P losses). At market-gate, air-freight was identified as main hot-spot pleading for the design of stabilized FB products that could be sea-freighted. At farm-gate, large differences were observed between farm types and fertilizer, water and land use were the key-drivers of their eco-efficiency. Impacts due to pesticides applications were small at Endpoint level but were incomplete. These results should be validated with a greater sample of stakeholders and the scope of the LCA should be extended to the consumption stage. Research is also needed to provide LCA practitionners with operational and reliable tools for a better inclusion of pesticides’ impacts and uncertainty.

© The Author(s) 2019

International audience; Purpose Salinisation is a threat not only to arable land but also to freshwater resources. Nevertheless, salinisation impacts have been rarely and only partially included in life cycle assessment (LCA) so far. The objectives of this review paper were to give a comprehensive overview of salinisation mechanisms due to human interventions, analyse the completeness, relevance and scientific robustness of existing published methods addressing salinisation in LCA and provide recommendations towards a comprehensive integration of salinisation within the impact modelling frameworks in LCA.Methods First, with the support of salinisation experts and related literature, we highlighted multiple causes of soil and water salinisation and presented induced effects on human health, ecosystems and resources. Second, existing life cycle impact assessment (LCIA) methods addressing salinisation were analysed against the International Reference Life Cycle Data System analysis grid of the European Commission. Third, adopting a holistic approach, the modelling options for salinisation impacts were analysed in agreement with up-to-date LCIA frameworks and models.Results and discussion We proposed a categorisation of salinisation processes in four main types based on salinisation determinism: land use change, irrigation, brine disposal and overuse of a water body. For each salinisation type, key human management and biophysical factors involved were identified. Although the existing methods addressing salinisation in LCA are important and relevant contributions, they are often incomplete with regards to both the salinisation pathways they address and their geographical validity. Thus, there is a lack of a consistent framework for salinisation impact assessment in LCA. In analysing existing LCIA models, we discussed the inventory and impact assessment boundary options. The land use/land use change framework represents a good basis for the integration of salinisation impacts due to a land use change but should be completed to account for off-site impacts. Conversely, the land use/land use change framework is not appropriate to model salinisation due to irrigation, overuse of a water body and brine disposal. For all salinisation pathways, a bottom-up approach describing the environmental mechanisms (fate, exposure and effect) is recommended rather than an empirical or top-down approach because (i) salts and water are mobile and theirs effects are interconnected; (ii) water and soil characteristics vary greatly spatially; (iii) this approach allows the evaluation of both on-and off-site impacts and (iv) it is the best way to discriminate systems and support a reliable eco-design.Conclusions This paper highlights the importance of including salinisation impacts in LCA. Much research effort is still required to include salinisation impacts in a global, consistent and operational manner in LCA, and this paper provides the basis for future methodological developments.

Introduction. Fruits are under growing scrutiny regarding their environmental impacts. However, fruit cropping systems have seldom been studied using life cycle assessment (LCA). As part of the Agribalyse® project, the cropping systems for apple and peach in France, clementine in Morocco, and mango in Brazil were evaluated with a cradle-to-farm-gate LCA in order to include the manufacturing, transportation and utilization of all inputs used on the farm. Materials and methods. Representative systems for each fruit were designed, relying mostly on expert knowledge for apple, peach and clementine, and on a detailed survey of eight commercial orchards for mango from the Rio San Francisco Valley in Brazil. Potential environmental impacts were calculated using the ReCiPe Midpoint (H) method. Results and discussion. For most impact categories, apple showed the least impacts, followed by mango, peach and clementine. For marine eutrophication, mango and small citrus had the least impact, followed by apple and peach far above. For ecotoxicity, mango had the least impact followed by apple, peach and clementine far above. For each fruit cropping system, margins of improvement were identified and associated with agronomic recommendations. Generally speaking, the key aspects driving the eco-efficiency of fruit cropping systems at the farm gate were the yield, the fertilizer rates, the use of toxic substances for pest management, the water requirement associated with its local availability and the energy source for irrigation. The ecotoxicity results revealed were uncertain due to the difficulty of determining representative crop protection practices for perennial crops. Conclusion. The Agribalyse® program represented an important step forward for the environmental assessment of fruits. Complementary research is needed to design better crop protection practices, field emissions and water-use impacts. (Resume d'auteur)

Purpose: In a context of flourishing eco-labelling programs and environment policy for food products, LCA application to agricultural systems faces the challenges of being operational, accurate and exhaustive. This is particularly challenging for the newly developing LCA and ISO-compliant water footprinting, with many LCIA methods only recently developed, but no dedicated inventory method. To support the inventory of elementary water flows, LCA practitioners have a variety of tools available, ranging from databases (e.g. World Food LCA Database) to complex agro-hydrological models. To allow all LCA practitioners to fulfil their diverse agri-food LCA objectives, a review of available inventory tools for field water flows and recommendations are needed. Methods: The selection of the appropriate method and tool for the inventory of field water flows in agri-food LCA studies depends on the objectives of the LCA study, data and resources available (time and skills). We analysed water inventory and agri-food LCA databases by evaluating the models on which they rely and their input data. Then, we explored the use of agro-hydrological models for LCA aiming at discriminating between different cropping system practices (LCA-based eco-design). Results and discussion: Water inventory and agri-food LCA databases provide estimates of theoretical water consumed by a crop and rely on data and methods that have limitations, making them suitable only for background agricultural LCAs. In addition, databases do not support the application of water availability footprint indicators (assessing quantitative water use and water quality alteration). For the LCA-based eco-design of cropping systems, the inventory of water flows should be based on a model simulating evapotranspiration, deep percolation and runoff accounting for crop specificities, pedo-climatic conditions and agricultural management. In particular, the model should account for possible water, salinity and nutrient stresses; assess evaporation and transpiration separately; and estimate runoff and drainage according to the system specificities. Yield should not be estimated with a model but with primary data. Recommended and default data sources are provided for each input parameter. Conclusions: The FAO AquaCrop model represents a good trade-off between accuracy, simplicity and robustness for LCA-based eco-design of cropping systems. However, this model is not yet applicable for perennial crops. Beyond a single model selection, this is a modelling approach that we characterised in this work.

International audience; AbstractVegetable production in sub-Saharan Africa faces numerous agronomic constraints that will have to be overcome to feed the increasing population and to fight malnutrition. Technology transfer and the adoption of low-tech protected cultivation techniques affordable for smallholders are believed to be able to meet this challenge. Protected cultivation techniques are a set of agricultural practices aimed at artificializing the crop environment through the use of soil covers and/or plant covers to control pests and climatic conditions. Although protected cultivation techniques may increase the yield and quality of vegetable crops and extend their production periods worldwide, the transfer of these techniques in sub-Saharan Africa raises questions about their agronomical performances, their profitability but also their environmental impacts. Are low-tech protected cultivation techniques adapted to the sustainable production of vegetables by smallholders in sub-Saharan Africa? To answer this question, we present an overview of the agronomic, economic, and environmental performances of low-tech protected cultivation techniques in sub-Saharan Africa as reported in the literature. The major conclusions that can be drawn from the review are (1) low-tech protected cultivation techniques are not suitable in all climatic conditions in sub-Saharan Africa and need to be combined with other methods to ensure adequate pest control, (2) the profitability of protected cultivation techniques relies on the capacity to offset increased production costs by higher yields and higher selling prices to be obtained with off-season and/or higher quality products, (3) breaking with existing cropping systems, the lack of technical support and skills, and the limited access to investment funding are major obstacles to the adoption of protected cultivation techniques by smallholders (4) life cycle assessments conducted in northern countries suggested that more efficient use of agricultural inputs would offset the negative impacts of protected cultivation techniques if they are properly managed, but further studies are required to be sure these results can be extrapolated to sub-Saharan Africa context.

International audience; This chapter deals with the application of Life Cycle Assessment to evaluate the environmental sustainability of agriculture and food processing. The life cycle of a food product is split into six stages: production and transportation of inputs to the farm, cultivation, processing, distribution, consumption and waste management. A large number of LCA studies focus on the two first stages in cradle-to-farm gate studies, as they are the stages where most impacts typically occur, due to animal husbandry and manure handling, production and use of fertilisers and the consumption of fuel to operate farm machinery. In the processing step, the raw agricultural product leaving the farm gate is converted to a food item that can be consumed by the user. Distribution includes transportation of the food product before and after processing. In the consumption stage, environmental impacts arise due to storage, preparation and waste of the food. In the waste management stage, food waste can be handled using a number of technologies, such as landfilling, incineration, composting or digestion. A number of case studies are looked at here where the life cycles of typical food products (meat, cheese, bread, tomatoes, etc.), and an entire diet are discussed. Other case studies deal with what LCA can conclude on the differences between conventional and organic farming, and the perceived advantages of local food items. Finally, methodological issues in agricultural LCA are discussed: the choice of functional unit, setting the boundary between technosphere and ecosphere, modelling flows of nutrients and pesticides, and the generally limited number of impact categories included in LCA studies.

Urban agriculture, Life-cycle inventory,Typology,Cropping system data,Treatment frequency index,Nutrient budget; International audience; In sub-Saharan Africa, urban farmers have recently intensified the production of vegetables to cope with the increasing food demand. As a consequence, such an intensification may lead to potential risks for the environment and human health. There is therefore a need for an integrated evaluation of urban agricultural practices. Here, we studied tomato production in Benin cities. We measured performances and the environmental risks. We have monitored 12 cropping systems during 6 months and we calculated the pesticide treatment frequency index (TFI), the nutrient budgets, and the field emissions. Our results show that yields were low and variable, averaging at 9,533 kg.ha−1 and ranging from 0 to 21,163 kg.ha−1. The average TFI for pesticides was 8.9. The maximum TFI of 25 was observed for an insecticide applied weekly at 2.3 times the official rate. We observed an excess of the average nutrient budget of 120 kg N and 84 kg P. ha−1. In conclusion, our study of urban tomato production revealed poor practices and high risks for health and the environment.

Northeastern Thailand is an essential production area for high-quality fragrant rice for both domestic use and export. While rain-fed conditions still largely prevail, plans to extend irrigation are being drafted. This paper compares the advantages of rice production under irrigation and rain-fed conditions in both environmental and economic terms. Indicators of techno-economic performances were combined with environmental impact indicators based upon life cycle assessment, energy and water use analyses. Data were collected in 2010 at the farm level in 43 households of Lam Sieo Yai Basin in North-Eastern Thailand, according to 3 cropping systems, namely wet-season rain-fed (Rw), wet-season irrigated (Iw) and dry-season irrigated (Id) systems. Eco-efficiency indicators were calculated as per impact category. Wide-ranging techno-economic performances and environmental impacts were observed, while cropping practices were found to be homogeneous. Differentiation of systems originated mostly from differences in yield, which were mostly impacted by water supply. Yields vary from 2625 kg/ha in the Iw system to 2375 in the Rw system and 2188 in the Id system. The results highlight the low performances of Id systems in both techno-economic and environmental terms. Id systems require mostly blue water, while the two other systems rely primarily on green water. Id systems also require more energy and labour, due to increased water management needs. Overall, the productivity of most production factors was found to be higher in Rw and Iw systems; this results in return on investment being slightly higher in the Iw system compared to the Rw system (0.12 kg/THB and 0.11 kg/THB, respectively) and is lowest in the Id system (0.1 kg/THB) where THB is Thai baht, currency of Thailand. In Id systems, farmers need to produce twice as much rice (0.41 kg) to obtain 1 THB of net income, compared to 0.23 and 0.25 kg for Iw and Rw respectively. Emissions proved relatively similar across all 3 systems, with the exception of CH 4 , which was markedly lower in Rw systems due to specific water and organic residue management. Id systems systematically emitted more nitrates, phosphates and pesticides into water sources. Rw systems showed the lowest environmental impacts per ha and per kg of paddy rice produced. GWP 100 was higher in Id systems (5.55 kg CO 2 -eq per kg of rice) compared to Iw (4.87) and Rw (2.97). Finally, Rw systems were found to be more eco-efficient in most impact categories, including Global Warming Potential. The total value product per kg of CO 2 -eq emitted is 4, 2.5 and 2.2 THB in Rw, Iw, and Id systems respectively. This paper further discusses the results in view of contrasting perspectives, including societal objectives, farmer income and environmental integrity, and possible irrigation development in Northeastern Thailand.

North-eastern Thailand is an essential production area for high-quality fragrant rice for domestic use and export. While rainfed conditions largely prevail, plans to extend irrigation are drafted. This article compares paddy rice production under irrigation and rainfed conditions. Techno-economic performances were analysed jointly with environ- mental impacts, based upon life cycle analysis, and energy and water use analyses. Data were collected in 2010 from 45 diverse rice cropping systems in the Lam Sieo Yai Basin, according to three systems, namely wet-season rain-fed (Rw), wet-season irrigation (Iw), and dry-season irrigation (Id) systems. Wide-ranging performances and impacts were observed, while cropping practices were relatively homogeneous. Differentiation of systems originated mostly from differences in yield, which were largely impacted by water supply. The results highlight the low performances and high impacts of Id systems. They require mostly blue water, while the two other systems rely primarily on green water. Id systems also require more energy and labour, due to increased water management needs. The productivity of most production factors was higher in Rw and Iw systems. Emissions proved relatively similar across systems, with the exception of CH 4 , which was markedly lower in Rw systems due to specific water and organic residue management. Id systems systematically emitted more nitrates, phosphates, and pesticides. Rw systems showed the lowest environmental impacts per ha and per kg of paddy rice produced. The average Global Warming Potential was 2.97 kg CO 2 -eq per kg rice in Rw systems, 4.87 in Iw systems, and 5.55 in Id systems. This article further discusses the results in view of contrasting perspectives, including societal objectives, farmer income and environmental integrity, and possible irrigation development in north-eastern Thailand.

Purpose As highlighted in recent reviews, there is a need to harmonise the way life cycle assessment (LCA) of perennial crops is conducted. In most published LCA on perennial crops, the modelling of the agricultural production is based on data sets for just one productive year. This may be misleading since performance and impacts of the system may greatly vary year by year. The purposes of this study are to analyse how partial modelling of the perennial cycle through non-holistic data collection may affect LCA results and to make recommendations. Methods Three modelling choices for the perennial crop cycle were tested in parallel in two contrasted LCA case studies: oil palm fruits from Indonesia, and small citrus from Morocco. Modelling choices tested were as follows: (i) a chronological modelling over the complete crop cycle of orchards, (ii) a 3-year average from the productive phase, and (iii) various single years from the productive phase. In both case studies, the system boundary was a cradle-to-farm gate with a functional unit of 1 kg fresh fruits. LCA midpoint impacts were calculated with ReCiPe 2008 in Simapro©V.7. We first analysed how inputs, yields and potential impacts varied over time. We then analysed process contributions in the baseline model, i.e. the chronological modelling, and finally compared LCA results for the various perennial modelling choices. Results and discussion Agricultural practices, yields and impacts varied over the years especially during the first 3–9 years depending on the case study. In both case studies, the modelling choices to account or not for the whole perennial cycle drastically influenced LCA results. The differences could be explained by the inclusion or not of the yearly variability and the accounting or not of the immature phase, which contributed to 7–40 or 6.5–29 % of all impact categories for oil palm fruit and citrus, respectively. Conclusions The chosen approach to model the perennial cycle influenced the final LCA results for two contrasted case studies and deserved specific attention. Although data availability may remain the limiting factor in most cases, assumptions can be made to interpolate or extrapolate some data sets or to consolidate data sets from chronosequences (i.e. modular modelling). In all cases, we suggest that the approach chosen to model the perennial cycle and the representativeness of associated collected data should be made transparent and discussed. Further research work is needed to improve the understanding and modelling of perennial crop functioning and LCA assessment.

A practical challenge in LCA for comparing pesticide application in different agricultural practices is the agreement on how to quantify the amount emitted, while only the amount applied to the field is known. Main goal of this paper is to present an international effort carried out to reach agreement on recommended default agricultural pesticide emission fractions to environmental media. Consensual decisions on the assessment framework are (a) primary distributions are used as inputs for LCIA, while further investigating how to assess secondary emissions, (b) framework and LCA application guidelines and documentation will be compiled, (c) the emission framework will be based on modifying PestLCI 2.0, (d) drift values will be provided by German, Dutch and other drift modelers, (e) pesticide application methods will be complemented to develop scenarios for ropical regions, (f) climate, soil and application method scenarios will be based on sensitiv ity analysis, (g) default emission estimates for LCA will be derived from production-weighted averages, and (h) emission fractions will be reported spatially disaggregated. Recommendations for LCA practitioners and database developers are (a) LCA studies should state whether the agricultural field belongs to technosphere or ecosphere, (b) additional information needs to be reported in LCI (e.g. pesticide mass applied), (c) emissions after primary distribution and secondary fate processes should be reported, (d) LCIA methods should allow for treating the field as part of technosphere and ecosphere, (e) fate and exposure processes should be included in LCIA (e.g. crop uptake), (f) default emission estimates should be used in absence of detailed scenario data, (g) and all assumptions should be reported. The recommended pesticide emission fractions results and recommendations are presented and disseminated to strive for broad acceptance at a dedicated stakeholder workshop back-to-back with the current LCA Food 2016 conference in Dublin.

LCAs (life cycle assessments) are often based on average data to produce a generic evaluation of a good or service. However, ignoring variability and induced uncertainty of LCA results reduces their significance, especially when dealing with agricultural processes that present high natural fluctuations. The objective of the study was to explore the robustness of LCA results when accounting for variable emissions data, illustrated by the case of slurry application techniques. Four application techniques were compared: band spreading, broadcast spreading, harrowing after surface application and direct injection. On the basis of the normalisation results, acidification, eutrophication and global warming potentials were selected. To estimate field nitrogen emissions, an original approach was developed based on relative nitrogen loss factors for each technique from a literature review. The calculated field emissions from different soil and climate conditions were considered equally probable and were propagated into a range of LCA result using the Monte Carlo method. Injection and harrowing both showed reduced acidification and eutrophication potentials compared to band spreading and broadcast spreading but had larger global warming potentials, which could be particularly important with injection. Harrowing consequently appeared as the best compromise. Despite the large range of LCA results, robust conclusions could be drawn. To achieve a more refined comparison between the techniques, the use of process-based models in contrasted situations is suggested.

New Zealand (NZ) dairy farms used to be the lowest input and most efficient dairy farms of the world. However, intensification of the traditional pasture-based system has occurred over the last decade and has not always been accompanied by increased efficiency. The purpose of this paper is to produce an updated reference of the eco-efficiency of NZ dairy farm systems and to analyse the implications of intensification on their eco-efficiency. Results for an average NZ dairy farm system were compared with those for three dairy farmlet systems representing a wide range in intensification practices. A low input system (LI) (no N fertiliser, no brought-in feed supplement, stocking rate of 2.3 cows/ha) was compared with an N-fertilised farm system (NF) (170 kg fertiliser-N/ha/year, 3 cows/ha) representing a first level of intensification and with an N-fertilised and maize silage supplemented system (NFMS) (170 kg fertiliser-N/ha/year, 13 t DM maize silage/ha/year, 5.2 cows/ha), representing a possible future intensification option. Their eco-efficiency in terms of milk production and land use was compared using Life Cycle Assessment (LCA) methodology. NZ dairy farm systems rely on favourable temperate climate conditions and long-term perennial ryegrass/white clover pasture, to achieve eco-efficient milk production and land use compared to European systems. However, intensification of NZ dairy farms was shown to be detrimental to their eco-efficiency in terms both of milk production and land use functions and could greatly reduce their advantage compared to European systems. The eco-efficiency of LI was very high whatever the functional unit which is remarkable from an LCA perspective. NF and NFMS had a similar eco-efficiency except for energy use which corresponded to the most critical hot spot of NFMS. All studied NZ systems presented some areas for improvement where some new technologies available for dairy farms might play a promising role in the future. Finally, it should be highlighted that the comparison with European studies based on the literature available needs validation through a more comprehensive study using harmonised methodology and assumptions across countries.

As one of the most comprehensive environmental assessment methodologies, life cycle assessment enables evaluation of the environmental impacts of anthropogenic activities along a supply chain. Its implementation raises many scientific questions. In the case of tropical cropping systems, researchers are working to understand and model environmental emissions based on the diversity of environments and systems. They are also focusing on the relationship between emissions and impacts. Cropping system life cycle assessments show that the impact on climate change varies by crop, environment and type of practice. Life cycle assessment can help guide production methods so as to reduce their environmental impacts. But the choices are not always clearcut.

AGRIBALYSE is a French research program dedicated to producing Life cycle inventories (LCI) of agricultural products, based on a strong partnership of 14 research and technical institutes. It provides a homogenous and consensual LCI database to support environmental labelling policies and to help the agricultural sector to improve its practices. The public LCI database of more than 100 products and its detailed methodology report were published in 2014. The database mainly contains LCIs for average French products, the functional unit is the product mass (kg) and the perimeter cradle-to-farm-gate. The datasets go along with a detailed methodology report and a project report, both available at: (www.ademe.fr/agribalyse). Several key objectives for a follow-up project have already been identified: better estimate of uncertainties, accounting for soil carbon dynamics, impacts on biodiversity and impacts of water consumption. Enlargement of the database with additional products and products from innovative systems is also demanded, as well as stronger connections with international programs. These issues should be addressed in a future AGRIBALYSE 2 program.

International audience; Land application of manures and slurries are a major source of pollution such as water contamination by nitrates and greenhouse gas emissions. NH3 and N2O emissions can be lowered by suitable spreading techniques. However, a comprehensive review of the impact of slurry spreading techniques is lacking. For that we developed a model, named OSEEP, that simulates the effect of slurry incorporation, slurry spatial distribution, and soil compaction on NH3 and N2O emissions. OSEEP integrates a soil compaction model, a hydraulic pedotransfer function, a NH3 volatilization model, and a crop model. We ran OSEEP for five sites in France for 7 years. Four techniques were simulated: broadcast spreading, band spreading, incorporation after surface spreading, and injection. We tested various sizes of slurry tankers and tractors. We calculated NH3 and N2O relative emissions from various spreading techniques with respect to band spreading. Results show good agreement between model calculation and published field data. We found that slurry applied by a self-propelled 15-m3 tanker with extra large tires led to an increase of 20 % in N2O emission, by comparison with a 15-m3 tanker trailed by a tractor. Hence, we show that soil compaction should be taken into account to optimize trade-offs between NH3 and N2O emissions.

The environmental impact of imported fresh agricultural products, such as off-season vegetables transported over long distances, is under growing scrutiny. We hypothesised that the environmental Life Cycle Assessment (LCA) ranking between local and imported vegetables might change depending on the impact category considered. We focused on the case study of off-season tomatoes produced in Morocco under unheated greenhouses in a water-scarce area, which covers 68% of the fresh tomatoes imported to France. First, we performed a cradle-to-market gate LCA of the Moroccan production using primary data based on a field survey. Second, we applied the same Life Cycle Impact Assessment (LCIA) method to published cradle-to-farm-gate results of the French tomato cropping system, which also provides off-season tomatoes to the French market and which is characterised by heated greenhouses with a high level of inputs. In addition to typical environmental impact categories, the freshwater use impact was included. The ranking between imported and local tomatoes was different depending on the impact category. Freshwater use had greater impacts under the Moroccan arid climate: 28.0 L H 2 O eq kg −1 of Moroccan tomato and 7.5 L H 2 O eq kg −1 of French tomato. Conversely, the higher level of artificialisation of the French production resulted in greater impacts on total energy consumption, global warming, and eutrophication, even including transport to France for the Moroccan tomato. This reveals a trade-off between freshwater use impacts and the usual/other impacts, mostly energy-related. At the farm gate, we found that the Moroccan tomato water consumption highly contributed to the total damages to Human Health (14%), and Ecosystems (20%) (contribution to Resources depletion was only 2%). Therefore, ignoring the impacts of freshwater use in LCA also underestimates the damages. Moreover, we showed that the assessment of freshwater use impacts and damages still has shortcomings, leading to an underestimation of the impact for the Moroccan tomato case. These results emphasised the importance of considering all of the impact categories when performing an agricultural LCA and the need for a more comprehensive method for assessing the impacts of freshwater use. In particular, the use of an operational tool for estimating water and solute fluxes at the field level is recommended to feed freshwater impact assessment methods.

Goal, Scope and Background Calculating LCA outcomes implies the use of parameters, models, choices and scenarios which introduce uncertainty, as they imperfectly account for the variability of both human and environmental systems. The analysis of the uncertainty of LCA results, and its reduction by an improved estimation of key parameters and through the improvement of the models used to convert emissions into regional impacts, such as eutrophication, are major issues for LCA.

Purpose Recent life cycle assessment studies for vegetable products have identified the agricultural stage as one of the most important contributors to the environmental impacts for these products, while vegetable production systems are characterized by specific but also widely diverse production conditions. In this context, a review aiming at comparing the potential impacts of vegetable products and analyzing the relevance of the methods and data used for the inventory of the farm stage appeared necessary. Methods Ten papers published in peer-reviewed scientific journals or ISO-compliant reports were selected. First, a presentation of the selected papers was done to compare the goal and scope and the life cycle inventory data to the related sections in the ILCDHandbook. Second, a quantitative review of input flows and life cycle impact assessment (LCIA) results (global warming, eutrophication, and acidification) was based on a cropping system typology and on a classification per product group. Third, an in-depth analysis of the methods used to estimate field emissions of reactive nitrogen was proposed. Results and discussion The heated greenhouse system types showed the greatest global warming potential. The giant bean group showed the greatest acidification and eutrophication potentials per kilogram of product, while the tomato group showed the greatest acidification and eutrophication potentials per unit of area. Main sources of variations for impacts across systems were yields and inputs variations and system expansion rules. Overall, the ability to compare the environmental impact for these diverse vegetable products from cradle-toharvest was hampered by (1) weaknesses regarding transparency of goal and scope, (2) a lack of representativeness and completeness of data used for the field stage, and (3) heterogeneous and inadequate methods for estimating field emissions. In particular, methods to estimate reactive nitrogen emissions were applied beyond their validity domain. Conclusions and recommendations This first attempt at comparing the potential impacts of vegetable products pinpointed several gaps in terms of data and methods to reach representative LCIA results for the field production stage. To better account for the specificities of vegetable cropping systems and improve the overall quality of their LCA studies, our key recommendations were (1) to include systematically phosphorus, water, and pesticide fluxes and characterize associated impacts, such as eutrophication, toxicity, and water deprivation; (2) to better address space and time representativeness for field stage inventory data through better sampling procedures and reporting transparency; and (3) to use best available methods and when possible more mechanistic tools for estimating Nr emissions.

Purpose Perennial crops globally provide a lot of fruit and other food products. They may also provide feedstock for bioenergy and have been, notably to this end, the subject of several LCA-based studies mostly focusing on energy and GHG balances. The purpose of this review was to investigate the relevance of LCAs on perennial crops, especially focusing on how the perennial crop specificities were accounted for in the farm stage modelling. Methods More than 100 papers were reviewed covering 14 products from perennial crops: apple, banana (managed over several years), orange and other citrus fruits, cocoa, coconut, coffee, grape fruit, Jatropha oil, kiwi fruit, palm oil, olive, pear and sugarcane. These papers were classified into three categories according to the comprehensiveness of the LCA study and depending on whether they were peer-reviewed or not. An in-depth analysis of the goal and scope, data origin for farming systems, modelling approach for the perennial cropping systems and methods and data for field emissions helped reveal the more critical issues and design some key recommendations to account better for perennial cropping systems in LCA. Results and discussion In the vast majority of the reviewed papers, very little attention was paid on integrating the perennial cropping cycle in the LCA. It is especially true for bioenergy LCA-based studies that often mostly focused on the industrial transformation without detailing the agricultural raw material production, although it might contribute to a large extent to the studied impacts. Some key parameters, such as the length of the crop cycle, the immature and unproductive phase or the biannual yield alternance, were mostly not accounted for. Moreover, the lack of conceptual modelling of the perennial cycle was not balanced by any attempt to represent the temporal variability of the system with a comprehensive inventory of crop managements and field emissions over several years. Conclusions According to the reviewed papers and complementary references, we identified the gaps in current LCA of perennial cropping systems and proposed a road map for scientific researches to help fill-in the knowledge-based gaps. We also made some methodological recommendations in order to account better for the perennial cycle within LCA considering the aim of the study and data availability.

LCA outputs are often presented as point estimates measuring potential impacts although average impacts values may be misleading to rank different options, especially in the case of agricultural products. In an LCA study comparing different slurry application techniques, NH3 and N2O emissions have been estimated through two approaches, experimental data collected from the literature and mathematical simulations over different soil and climate conditions. Both approaches lead to similar ranges of emissions; however the simulation-based approach allows us to construct a probability distribution of emissions whereas the limited number of experimental studies leads only to the definition of a range of emissions. A better knowledge of the variability of emissions helps the practitioner to sort alternatives and to detect situations where they are not discernable. Moreover the knowledge of the distribution and of its most impacting sources of variability leads to the definition of more informative and significant typologies., Sorties ACV sont souvent présentées comme des estimations ponctuelles de mesure des impacts potentiels, bien que les valeurs moyennes des impacts peuvent être trompeuses pour classer les différentes options, en particulier dans le cas des produits agricoles. Dans une étude ACV comparant les différentes techniques d'application du lisier, les émissions de NH3 et de N2O ont été estimés selon deux approches, les données expérimentales recueillies dans la littérature et des simulations mathématiques sur un sol différent et les conditions climatiques. Les deux approches conduisent à des plages d'émissions similaires, mais l'approche basée sur la simulation nous permet de construire une distribution de probabilité des émissions alors que le nombre limité d'études expérimentales ne mène qu'à la définition d'une gamme d'émissions. Une meilleure connaissance de la variabilité des émissions permet au praticien de tri des alternatives et de détecter les situations où ils ne sont pas discernables. Par ailleurs la connaissance de la distribution et de la plupart de ses sources impact de la variabilité conduit à la définition de typologies plus informatif et significatif.

International audience; LCA is the evaluation of all environmental impacts of a function (product or service) on natural capital, human health and resource use. However, current sustainability definitions are considerably broader in both environmental and social aspects than is commonly treated in LCA. The egg of sustainability model which places ontological priority on the environment, then social systems, then economics, is perhaps the most appropriate to effectively address sustainability issues. However, in real-world examples, due to short-term reactions to economic and social pressures, environmental components are frequently given the lowest priority. If properly researched and developed, LCA could be an effective tool to identify practical ways to address sustainability issues in real-world food production. In particular, development of regionally specific indicators which have quantifiable levels of uncertainty, approximate reality and incorporate social and economic aspects would be important future progressions to enhance the contribution of LCA to sustainable food production.

Background, aim and scope As a food exporting nation, New Zealand recognises that the Global Warming Potential (GWP) impact of agriculture has become important to food customers. Food production policy and industry analysts make GWP decisions based on greenhouse gas inventory and life cycle assessment (LCA) results. For decision making, the level of confidence associated with information is important. However, treatment of uncertainty has been problematic in LCA, especially in agricultural systems. In this paper, the GWP of 1 kg of milk was used as a case study to test the feasibility of quantifying uncertainties by Monte Carlo simulation in an LCA applied to an agriculture product. The study also contributes to the development of good practice and has implications for the incorporation of uncertainties into decision making. Materials and methods We distinguished between three sources of variation. First, there is variability amongst basic units such as dairy cattle, soils and farm characteristics which may be quantified by the standard deviation (SD). Second, there is uncertainty about true population means, which is typically provided by a sample and can be measured by the standard error of the mean (SEM). Third, choices, such as the time horizon for computing GWP, can strongly affect the LCA outcomes. The first two sources were analysed by compiling input variable statistics and undertaking Monte Carlo numerical simulations. The third source of variation was quantified by sensitivity analysis. Results Up to the farm-gate stage, the mean GWP of 1 kg of milk (computed over 100 years) was 0.96 kg CO2-eq. The associated SD was 38% of the mean when using the SD of input variables (and called "variability") and 7% when using the SEM (and called "uncertainty"). The GWP was most sensitive to uncertainty of pasture dry matter intake by grazing cattle. The second and third key input variables were the cattle excreta nitrous oxide emission factor and the enteric fermentation methane emission factor, respectively. Changing the GWP from a 100-year computation to one of 20 years corresponded to GWP increasing by 92%, while for a change from 100 to 500 years GWP declined by 54%. Discussion Data compilation for the uncertainty analyses was challenging because the measurements available were made over smaller time and space scales than ideal, so observations had to be generalised and data gaps filled by expert judgement. Uncertainty analysis using the SEM of input variables was considered most adapted in LCA, so it is recommended as best practice. Identification of the key parameters responsible for uncertainty in the LCA revealed knowledge gaps where research should be directed, such as for methane digestion and nitrous oxide emissions from N excreted by ruminants. Moreover, richer information for those key parameters could be used to build a typology of more meaningful simulations instead of a single, virtual average for analysing environmental impacts of the agricultural system. Conclusions The use of Monte Carlo simulations for uncertainty and sensitivity analyses of LCA estimates for an agricultural product was feasible and recommendations were made. Developing a typology of realistic simulations based on the key parameters identified in the sensitivity analysis could provide decision makers with more information. Furthermore, in comparative LCA studies, a probabilistic framework provides further information including the statistical significance of differences between technological options. This would represent considerable progress for the decision-making process. Recommendations and perspectives We recommend that uncertainty information such as SEM, when available, be part of inventory data for agriculture systems in public reports and databases including assessment of its statistical meaning and consistency.

International audience; The comparison of the environmental performance of contrasting agricultural production systems using life cycle assessment (LCA) requires very specific data for each production mode, especially for direct emissions on the farm. In a case study of pig production, three contrasting production scenarios were defined: conventional good agricultural practice (GAP), organic agriculture (OA), and red label (RL). An expert panel produced methods for the estimation of emissions for key-processes of these scenarios. Clear differences were obtained for these emissions between the three scenarios. Finally, the summation of the main emissions for the entire life cycle allowed the identification of weak points for each scenario, which can be explained by the differences obtained for the emissions of the key-processes studied. (C) 2006 Elsevier Ltd. All rights reserved.

It is not known to what extent the outcome of studies assessing the environmental impacts of agricultural systems depends on the characteristics of the evaluation method used. The study reported here investigated five well-documented evaluation methods (DIALECTE, Ecological Footprint, Environmental Management for Agriculture, FarmSmart, Life Cycle Assessment) by applying them to a case study of three pig farm scenarios. These methods differ with respect to their global objective (evaluation of impact versus evaluation of adherence to good practice), the number and type of environmental issues they consider, the way they define the system to be analysed, the mode of expression of results (for the farm as a whole, per unit area or per unit product) and the type of indicators used (pressure, state or impact indicators). The pig farm scenarios compared were conventional good agricultural practice (GAP), a quality label scenario called red label (RL) and organic agriculture (OA). We used the methods to rank the three scenarios according to their environmental impacts. The relative ranking of the three scenarios varied considerably depending on characteristics of the evaluation method used and on the mode of expression of results. We recommend the use of evaluation methods that express results both per unit area and per unit product. Environmental evaluation methods should be used with great caution, users should carefully consider which method is most appropriate given their particular needs, taking into consideration the method's characteristics.

The challenge for environmental assessment tools, such as Life Cycle Assessment (LCA) is to provide a holistic picture of the environmental impacts of a given system, while being relevant both at a global scale, i.e., for global impact categories such as climate change, and at a smaller scale, i.e., for regional impact categories such as aquatic eutrophication. To this end, the environmental mechanisms between emission and impact should be taken into account. For eutrophication in particular, which is one of the main impacts of farming systems, the fate factor of eutrophying pollutants in catchments, and particularly of nitrate, reflects one of these important and complex environmental mechanisms. We define this fate factor as: the ratio of the amount of nitrate at the outlet of the catchment over the nitrate emitted from the catchment's soils. In LCA, this fate factor is most often assumed equal to 1, while the observed fate factor is generally less than 1. A generic approach for estimating the range of variation of nitrate fate factors in a region of intensive agriculture was proposed. This approach was based on the analysis of different catchment scenarios combining different catchment types and different effective rainfalls. The evolution over time of the nitrate fate factor as well as the steady state fate factor for each catchment scenario was obtained using the INCA simulation model. In line with the general LCA model, the implications of the steady state fate factors for nitrate were investigated for the eutrophication impact result in the framework of an LCA of pig production. A sensitivity analysis to the fraction of nitrate lost as N2O was presented for the climate change impact category. This study highlighted the difference between the observed fate factor at a given time, which aggregates both storage and transformation processes and a "steady state fate factor", specific to the system considered. The range of steady state fate factors obtained for the study region was wide, from 0.44 to 0.86, depending primarily on the catchment type and secondarily on the effective rainfall. The sensitivity of the LCA of pig production to the fate factors was significant concerning eutrophication, but potentially much larger concerning climate change. The potential for producing improved eutrophication results by using spatially differentiated fate factors was demonstrated. Additionally, the urgent need for quantitative studies on the N2O/N-2 ratio in riparian zones denitrification was highlighted. (c) 2006 Elsevier B.V. All rights reserved.

Different types of assessment tools have been developed in Europe with the purpose of determining the environmental impact of various livestock production systems at farm level. The assessment tools differ in terms of which environmental objectives are included and how indicators are constructed and interpreted. The paper compares typical tools for environmental assessment of livestock production systems, and recommends selected indicators suitable for benchmarking. The assessment tools used very different types of indicators ranging from descriptions of farm management and quantification of input to estimates of emissions of, e.g., nitrate and ammonia. The indicators should be useful in a benchmarking process where farmers may improve their practices through learning from farms with better agri-environmental performance. An example of this is given using data on P-surplus on pig farms. Some indicators used the area of the farm as the basis of the indicator—e.g. nitrogen surplus per hectare—while others were expressed per unit produced, e.g. emission of greenhouse gasses per kilogram milk. The paper demonstrates that a comparison of organic vs. conventional milk production and comparison of three pig production systems give different results, depending on the basis of the indicators (i.e. per hectare or per kilogram product). Indicators linked to environmental objectives with a local or regional geographical target should be area-based—while indicators with a global focus should be product-based. It is argued that the choice of indicators should be linked with the definition of the system boundaries, in the sense that area-based indicators should include emissions on the farm only, whereas product-based indicators should preferably include emissions from production of farm inputs, as well as the inputs on the actual farm. The paper ends with recommendations for choice of agri-environmental indicators taking into account the geographical scale, system boundary and method of interpretation. D 2005 Elsevier B.V. All rights reserved.

Scenario-based environmental assessment of farming systems: the case of pig production in France

The inventory model 'PestLCI Consensus', originally developed for temperate conditions, estimates initial pesticide emission fractions to air, to off-field surfaces by drift deposition, and to field crop and field soil surfaces according to crop foliar interception characteristics. Since crop characteristics and application techniques differ in tropical conditions, these aspects need to be included in the model in support of evaluating pesticide emissions under tropical conditions. Based on published literature, a consistent set of crop foliar interception fractions was developed as function of crop characteristics and spraying techniques for tropical crops. In addition, we derived drift deposition fractions from published drift experiments specifically conducted under tropical conditions. Finally, we compiled a consistent set of pesticide emission fractions for application in life cycle assessment (LCA). Foliar interception fractions are strongly influenced by the spraying technique, particularly for hand-operated applications. Drift deposition fractions to off-field surfaces were derived for air blast sprayer on papaya and coffee, for boom sprayer on bean and soybean, for aerial application on soybean, sorghum, millet, corn and cotton, and for hand-operated application on cotton. Emission fractions vary for each combination of crop and application method. Drift deposition curves for missing crop-application method combinations can only partly be extrapolated from the set of considered combinations. Overall, our proposed foliar interception fractions and drift deposition fractions for various crops grown under tropical conditions allow to estimate pesticide emissions in support of assessing the environmental performance of agrifood systems in LCA with focus on tropical regions.

The general objective of the thesis is to develop and validate an assessment framework of the environmental efficiency of slurry application techniques based on Life Cycle Assessment (LCA). A first LCA showed that the impact of the manufacturing and the use of slurry spreaders were negligible compared to those induced by the nitrogenous emissions (NH3, N2O & NO3-) from the fertilized soil. These emissions depend on soil, climate and agricultural practices and are thus eminently site-specific. This is challenging for LCA, which is a method based on the absence of spatial differentiation. To account for the variability of nitrogenous emissions at field level in the LCA of the slurry application function, two approaches were followed: (A) by using experimental data from the literature which allowed to order four application techniques in a robust way, to quantify the variability of emissions and to discuss the uncertainty induced on LCA final results. (B) by using simulated data with models from soil physics, agro-climatology, and agronomy models. A coherent simulation tool was developed including the effects on nitrogenous emissions from i) the incorporation of slurry into the soil, ii) the soil compaction by the spreaders, iii) the spatial distribution of the slurry over the field. These effects were simulated in 5 sites in France over 7 years. The range of emissions for each technique estimated with both approaches was comparable. The use of models allowed to assess the contribution of each elementary effect on nitrogenous emissions and to order the applications techniques in a range of contrasted situations. Finally, a general methodology for the inclusion of variability in the environmental inventory for agricultural products is proposed., L'objectif général de la thèse est de développer et valider un référentiel d'évaluation de la performance environnementale des techniques d'épandage basé sur l'Analyse de Cycle de Vie (ACV). Une première ACV a montré que la fabrication et le fonctionnement des machines avaient une contribution négligeable par rapport aux impacts générés par les substances azotées (NH3, N2O et NO3-) émises par le sol fertilisé. Ces émissions sont conditionnées par le sol, le climat et les pratiques culturales et ont un caractère éminemment local. Cela pose un défi à l'ACV, méthode globale basée sur une absence de différentiation spatiale des systèmes étudiés. Pour prendre en compte la variabilité des émissions azotées à la parcelle dans l'ACV de la fonction d'épandage, deux approches ont été suivies: (A) l'utilisation de données expérimentales issues de la bibliographie qui a permis de hiérarchiser de manière robuste quatre techniques d'épandage, de quantifier la variabilité des émissions et de discuter l'incertitude induite sur les résultats de l'évaluation. (B) l'utilisation de données simulées avec des modèles de physique des sols, d'agro-climatologie et d'agronomie. Un outil de simulation intégrateur a été construit pour modéliser les effets sur les émissions azotées de i) l'enfouissement de l'effluent dans le sol, ii) du tassement du sol par les engins agricoles, et iii) la répartition spatiale de l'effluent sur la parcelle. Ces effets ont été simulés sur 5 sites en France pendant 7 années. La variabilité des émissions estimées par les deux approches est comparable. L'utilisation de modèles a permis d'évaluer la contribution des effets élémentaires des épandeurs sur les émissions azotées et de hiérarchiser la performance des techniques d'épandage dans une gamme de situations contrastées. Finalement, une méthodologie permettant la prise en compte de la variabilité dans l'inventaire environnemental des produits agricoles est proposée.