Tropospheric ozone acts as a phytotoxin which produces an oxidative stress in plants. Two ways of defense are used, either by preventing ozone input through the regulation of stomatal conductance, or by detoxifying ozone and ROS in cells. It is known that stomatal movements are alt ered by ozone. We performed fumigation experiment on three euramerican poplar genotypes ( Po pulus deltoides x Populus Nigra : ‘Carpaccio’, ‘Cima’ and ‘Robusta’ ), cultivated in pots in phytotronic chambers submitted to 120 ppb ozone or filtered air. We explor ed the effects of ozone on stomatal responses to four environmental parameters (blue light, red light, CO 2 and VPD). We also find out using a porometer that the decrease of stomatal conductance due to ozone is earlier on the adaxial face than on abaxial fa ce. Finally, to better understand these impacts, we studied ultrastructure of guard cells by TEM, stomatal density and size of stomata by SEM, and we performed X - ray microanalysis of guard cells mineral content. These approaches are coupled with the study on microdissected stomata of expression of genes involved in regulation of stomatal movements

The tropospheric level of the phytotoxic air pollutant ozone has considerably increased during the last century and is expected to continue to rise. Long-term exposures of higher plants to low ozone concentrations affect biochemical processes prior to any detectable symptoms of visible injury. On the other hand, the current critical level of ozone used to determine the threshold for damaging plants (biomass loss) is still based on the seasonal sum of the external concentrations of the pollutant above 40 nl l-1 (AOT40). Taking into account the stomatal conductance, a more relevant concept is based upon the actual ozone flux in the leaf through the stomata (cumulative uptake of ozone = CUO). CUO however ignores the internal capacity of leaf defense, which led to the concept of “effective ozone flux”, balance between stomatal flux and the intensity of cellular detoxification, with the aim to propose an improved threshold for ozone risk. Although the direct detoxification of ozone (and ROS issued from its decomposition) can primarily be carried out by cell wall ascorbate, the existing level of this antioxidant is not sufficient to indicate the degree of cell sensitivity. The capacity for regeneration of the antioxidant barrier is needed, implying the knowledge of the increased production of reducing power (NAD(P)H), primary supplier for detoxifying processes. It is made possible through the increased participation of the catabolic pathwaysand associated shunts which can provide NAD(P)H. In addition, the large change in the rubisco/PEPcase ratio, due to a huge increase in activity of the latter enzyme, leads to changes in carbon isotopic discrimination, which could be related to water use efficiency.Some results will be presented knowing that the challenge is to integrate the possible indicators in a leaf model to be used, through an upscaling process, in a tree and forest stand model.

The growing importance of environmental matters in social responsibility of firms has generated manyframeworks of analysis in the economic literature. Among those frameworks, performance evaluationand benchmarking using the non-parametric Data Envelopment Analysis (DEA) have increased at avery fast rate. This PhD research focuses on models that include undesirable outputs such as pollutionin the overall production system, to appraise eco-efficiency of decision making units (DMUs). Besides,the recent awareness on the large contribution of agriculture and particularly livestock farming toglobal warming, has highlighted for this sector the challenge of reaching both economic andenvironmental performances. In this line, the overall objective of this dissertation is to provide atheoretical and empirical background in modelling pollution-generating technologies and to suggesttheoretical improvements that are consistent with the particular case of greenhouse gas emissions inextensive livestock systems. Firstly, we showed that all existing approaches that deal with undesirableoutputs in the non-parametric analysis (i.e. DEA) have some strong drawbacks. However, the modelsgrounded on the estimation of multiple independent sub-technologies offer interesting opportunities.Secondly, I developed a new framework that extends the by-production approach through theintroduction of some explicit dependence constraints that link the sub-technologies in order to build aunified system. Thirdly, an empirical comparison, using a sample of French sheep meat farms, of thisby-production modelling extension with the existing approaches, revealed some inconsistencies ofthese latter. Finally, we expanded this new by-production formulation to account for dynamic aspectsrelated to the presence of adjustment costs. The application to the case of French suckler cow farmsunderlined the necessity of accounting for dynamic aspects and also showed high heterogeneity ininvestment strategies of these farmers. | La prise en compte des problèmes environnementaux dans la responsabilité sociale des entreprises a généré en économie de nombreuses propositions. Parmi elles, le cadre d’analyse basé sur l’évaluation de la performance en utilisant notamment les techniques d’enveloppement des données (DEA) s’est très vite répandu dans la littérature théorique comme empirique. Ce travail de thèse s’inscrit dans cette logique en mettant l’accent sur la modélisation des technologies polluantes. Par ailleurs, la question des changements climatiques et de la forte contribution de l’agriculture et en particulier de l’élevage dans les émissions de gaz à effet de serre (GES) impose à ce secteur de relever aujourd’hui en plus du défi économique celui de l’amélioration de sa performance environnementale. L’objectif général de cette recherche doctorale est donc de fournir un nouveau cadre d’analyse théorique et empirique dans la modélisation des technologies polluantes afin d’évaluer l’éco-efficience des systèmes productifs, en particulier le cas des émissions de GES en élevage extensif de ruminants. Dans un premier temps, nous montrons les limites théoriques et méthodologiques des modèles existants. Néanmoins, nous insistons sur le fait que les approches basées sur l’estimation de plusieurs sous-technologies indépendantes pour prendre en compte les différents processus présents dans les systèmes productifs sont très prometteuses. Dès lors dans un deuxième temps, nous proposons une nouvelle extension de la méthode « by-production » qui repose sur l’introduction d’interconnections entre les différentes sous-technologies impliquées afin de construire un système plus unifié. Dans un troisième temps, une comparaison empirique utilisant des données d’exploitations de viande ovine de notre extension avec les approches existantes a révélé certaines incohérences de ces dernières. Enfin pour aller plus loin, nous élargissons dans un quatrième temps notre approche afin de prendre en compte les aspects dynamiques et notamment la présence de coûts d’ajustement. Les résultats de l’analyse empirique entreprise avec des données d’exploitations bovines allaitantes (viande) ont révélé la nécessité de prendre en compte ces aspects,mais ont aussi révélé la forte hétérogénéité existante dans les stratégies d’investissements deséleveurs.