Ergot, caused by the fungal pathogen Claviceps purpurea (Fr.) Tul., attacks the floral organs of many grassy species resulting in sclerotia production rather than grain. Infection causes reduced yields, downgrading, and poisoning if consumed by humans or animals. Few recent studies have been conducted on ergot in wheat (Triticum spp.), and prevention is the only means of control. The objectives of this study were to determine if western Canadian spring wheat differed in reaction to infection with C. purpurea and if levels of inoculum would affect disease intensity in a field setting. Three variables were measured for the field experiments to determine disease reaction, including percent sclerotia by weight, number of sclerotia per spike, and weight per sclerotium. In the first experiment, nine wheat cultivars were tested using three inoculum levels. No significant differences were detected among inoculum levels. In the second and third experiments, ninety-two cultivars were studied in field and controlled conditions. Honeydew production, sclerotial size, and the percent of florets aborted were added as variables in the growth chamber experiment. Pearson correlations were calculated using cultivar means for the field and controlled environments. Results indicate that there are differences in disease reaction among cultivars and market classes, but these differences varied depending on the evaluation method used. In the field, CWAD wheat had the smallest sclerotia, but had more per spike compared to the CWRS and CWES market classes. There were no significant differences among these market classes for percent sclerotia by weight. In the growth chamber, CWAD wheat generally had the lowest ergot infection levels. When comparing the market classes within T. aestivum (CWRS, CPS, and CWES), there were no significant differences except for honeydew production. The correlation between environments was not significant for any of the variables, suggesting alternate resistance mechanism expression. In the field, reduced infection may be due to an escape mechanism, while artificial inoculation in a controlled environment may detect a physiological resistance mechanism. However, a group of cultivars with Grandin parentage showed promising results in both environments, and might confer resistance that could be integrated into disease resistance breeding programs.

Le développement de variétés performantes de blé dur (Triticum turgidum durum) doit prendre en compte la corrélation négative entre rendement et teneur en protéines (GPC). Une meilleure compréhension de ce phénomène nécessite une connaissance de l'élaboration des composantes de ces variables. Au cours de cette thèse nous nous sommes focalisés sur l'étude du déterminisme de paramètres foliaires pendant le remplissage du grain (morphologie, résorption d'azote/senescence) ainsi que leurs relations avec les caractéristiques post-récolte (rendement, GPC). La population d'étude est constituée de 282 lignées recombinantes issues d'un croisement demi-diallèle entre quatre lignées élites. Les données proviennent d'expérimentations menées en milieu contrôlé (3 expérimentations) et en plein champ (2 essais).En milieu contrôlé, plusieurs variables (teneur en azote, masse surfacique) ont été prédites par spectrométrie infrarouge à raison de mesures régulières au cours de la période d'intérêt. La résorption d'azote de la dernière feuille a été modélisée pour chaque génotype dont les paramètres ont été analysés. Un lien négatif fort a été établi entre la date d'initiation de la résorption (t0) en temps thermique post-floraison et sa durée et un lien positif entre cette date clé et le poids individuel des grains. A l'échelle de la parcelle, le suivi dynamique de la senescence par la mesure d'évolution d'un indice de végétation, le NDVI (Normalized Difference Vegetation Index) a montré une relation forte entre vitesse d'évolution de la senescence et son ordonnée à l'origine, suggérant que la fin de la senescence était un phénomène très contraint dans les conditions environnementales de plein champ.En absence de stress azoté, l'analyse des 4 géniteurs a montré que des combinaisons linéaires de paramètres foliaires permettaient d'expliquer environ 2/3 des variations de la production en grains d'un épi et de la GPC. En considérant l'ensemble de la population, les effets antagonistes d'une résorption tardive sur la production de grains et la teneur en protéines ont été établis. Une carence en azote appliquée post-floraison provoque une résorption précoce et longue, ce qui semble d'avantage affecter la GPC que le rendement. Une carence en azote appliqué pré-floraison réduit la quantité d'azote disponible dans les feuilles drapeau (et par conséquent leur potentiel photosynthétique) ainsi que le nombre de grains alors que les apports tardifs en azote réalisés dans ce traitement vont retarder la résorption donc augmenter le poids individuel des grains et la GPC. L'analyse génétique en modèle mixte a permis de prendre en compte un apparentement entre lignées, calculé grâce aux pedigrees ou aux marqueurs. Ces deux estimateurs de l'apparentement sont très fortement corrélés ; l'ordonnée à l'origine de la relation (0.3) pouvant être interprétée comme une mesure de l'identité par état de la population de base à partir de laquelle les géniteurs sont issus. Par conséquent, les variances additives issues d'un apparentement marqueurs sont systématiquement supérieures à celles d'un apparentement pedigree. Les valeurs d'héritabilité des variables étudiées caractérisant la morphologie de la feuille drapeau à floraison et des paramètres de résorption d'azote sont du même ordre de grandeur que celles du rendement d'un épi et de la teneur en protéines des grains.En conclusion, nous discutons des stratégies de sélection autour de la corrélation négative rendement/GPC en lien avec la résorption d'azote et la senescence, de l'intérêt de l'utilisation du modèle dit « animal » pour les analyses génétiques et la recherche de QTL. / Performing cultivar development of durum wheat (Triticum turgidum durum) have to take in consideration the negative correlation existing between yield and grain protein content (GPC). A best comprehension of this phenomenon needs the knowledge of these variables elaboration. During this thesis, we focused on the study of foliar parameters determinism during the grain filling (morphology, nitrogen resorption/senescence) and their relations with post-harvest parameters (yield, GPC). The population was formed of 282 recombinant lines coming from a half-diallel cross between 4 elite lines. Data are from experimentations on controlled environment (3 experimentations) and on field (2 experimentations).In controlled environment, various variables (nitrogen content, mass area) were predicted using near-infrared spectroscopy and performing regular measures during the period of interest. Nitrogen resorptions of the last leaf were modeled for each genotype and parameters were analyzed. A strong negative correlation was established between the resorption initiation date (t0) in thermal time post-flowering and its duration and a positive correlation was established between t0 and the grain individual weight. At the plot scale, the dynamic monitoring of senescence through measures of NDVI (Normalized Difference Vegetation Index) evolution showed a strong relation between the senescence evolution rate and its intercept, suggesting the senescence end was a very constraint phenomenon in field.Without nitrogen stress, the analysis of the 4 genitors indicated that linear combinations of foliar parameters could explain around 2/3 of one-spike grain production and GPC variations. Considering the entire population, the antagonist effects of a late resorption on grain production and on the GPC were established. A post-flowering nitrogen stress causes an early and long resorption which seems to affect more GPC than yield. A pre-flowering nitrogen stress reduces nitrogen quantity available in flag leaves (and consequently their photosynthetic potential) and the grain number whereas late nitrogen inputs of this environment delay resorption and raise grain individual weight and GPC.The genetic analysis in mixed model allowed taking into account relatedness among lines, calculated with pedigrees or markers. Those 2 relatedness estimators are strongly correlated; the regression intercept (0.3) can be interpreted as an identity-by-state measure of the base population from which genitors are derived. Consequently, additive variances from a marker-relatedness model are always superior to additive variances form a pedigree-relatedness model. Heritability values of the study variables (flag leaf morphology and nitrogen resorption parameters) have the same order of magnitude than those of one-spike yield and GPC.Field experimentations allowed highlighting the flag leaf weight as a contributor organ to yield and GPC. Nitrogen balance showed a link between post-flowering nitrogen fluxes and nitrogen available at flowering. Genotype-phenotype association's research was realized through a simple-marker analysis for all the experimentations and a Composite Interval Mapping analysis only for data obtained in greenhouses. A stable QTL was detected on the 3 controlled environments for t0. QTL comparison on controlled environment QTL and field QTL highlighted various area containing QTL on certain chromosomes: an area of foliar nitrogen content at flowering QTL on the 1A, an area of GPC QTL, Thousand Kernel Weight (TKW) QTL and grain number QTL on the 2A, an area of foliar morphology QTL on the 2B, an area of TKW QTL on the 3B and an area of senescence variables QTL on the 7B.In conclusion, we discuss about breeding strategies around the negative correlation yield/GPC in relations with nitrogen resorption and senescence and about the interest to use the “animal” model for genetic analysis and QTL research.