Colonisation and spatial expansion of a plant species is a major phenomenon in ecology, evolutionary biology, population genetics, and biogeography. This phenomenon must also be taken into consideration in management strategies related to spatial expansion of invasive species, species displacement in conjunction with climate change or fragmentation of natural habitats. Extensive empirical and theoretical knowledge is available on the dynamics of colonisation, especially on the persistence of a spreading species, the speed and shape of its colonization front. Reaction-diffusion (R-D) models were largely used to investigate these questions in homogeneous, heterogeneous or fragmented environments as well as in the presence of an Allee effect. Theoretical studies have then extended R-D models to take into consideration the presence of long-distance dispersal (LDD). They proved that LDD had a major impact on the colonisation pattern and made it possible to properly describe the most influential characteristics of the dispersal kernel. In contrast, fewer theoretical studies were devoted to the spatial dynamics of genetic diversity along a colonisation process, despite its importance for population genetics and evolutionary biology. In particular, the consequences of long-distance dispersal and spatial heterogeneity on the spatial pattern of diversity were essentially studied using simulation approaches and less comprehensively than what was done for population dynamic purposes. For example, the effect of the exact shape of the dispersal kernel was not fully investigated: studies on spatial genetic structure during expansion mostly focused on thin-tailed dispersal functions (mixture of two Gaussian, exponential). Further, and although weakly investigated, the spatial structure of diversity is expected to be strongly affected by (i) the fragmentation of the environments into favourable and unfavourable habitats and (ii) the variations of the dispersal function depending on the positions of the individuals relative to the wave front. The originality of the project presented here is to associate 1) a mathematical approach derived from the reaction-diffusion formalism, 2) a pragmatic simulation approach fed by physical models for wind-dispersed seeds and 3) the thorough description of the dynamics and genetics of two colonisation events occurring on Mont-Ventoux since 150 years. The first question tackled will be that of the origin of long-distance dispersers and how mixing of different origins occurs at long distance. First, this question will be considered from the mathematical point of view, using R-D models and integro-difference models. We will focus on the effects of (i) presence/absence of LDD, (ii) heterogeneity/fragmentation of the environment, and (iii) presence/absence of Allee effect. Second we will develop an Eulerian-Lagrangian approach to model wind-dispersed seeds in a heterogeneous environment. We will use this model to evaluate how air flow in the canopy of an expanding tree population generates spatial variations of the dispersal kernel that may affect the mixing of seeds origins at long-distance. Third, we will conduct aerial photograph analyses and experiments using microsatellite markers on a study site on Mont-Ventoux where a cedar colonisation over few kilometres is ongoing since 1860. This will allow quantifying empirically how mixing of origins really occurs at this site and if it affects the spatial pattern of the population. The second question tackled will concern the set-up of the spatial genetic structure (SGS) during the successive dispersal events that generate the colonisation. First, mathematical models (R-D and integro-differences) will be used to focus on the simplified question of the fate of a mutant appearing on a colonisation wave front. Here again, we will theoretically investigate the effects of LDD and heterogeneity of the environment on the possible futures for the mutant. Second, a simulation approach on a grid will be used to investigate this same question in the presence of demographic stochasticity and results will be compared with the deterministic approach. More realistic simulations including the characteristic of tree life-cycles will also be used to investigate more precisely the set-up of a SGS such as those measured from microsatellite data. Third, a landscape genetics approach will be used to study the re-colonisation of beech on the north-face of Mont-Ventoux. We will genotype, position and measure ages of ~1000 adult beech trees in order to retrieve the refugia and colonisation routes explaining the spatial structure of genetic diversity observed today. Finally, we will estimate the SGS for cedar and beech at different distances relative to the colonisation front and for the different cohorts of individuals to understand how SGS evolved with time and space during these two colonisations.

To be fully operational for assisting decisions at global, national and local levels, environmental modelling require precise spatially referenced soil information as inputs. The current soil database that exist in the world are neither exhaustive nor precise enough for being used efficiently for this purpose. An alternative is the digital soil mapping (DSM) which can be defined as 'the creation and population of spatial soil information systems by numerical models inferring the spatial and temporal variations of soil types and soil properties from soil observation and knowledge and from related environmental variables'. To face the lack of soil data, The Digital Soil Mapping working group of IUSS recently proposed (www.globalsoilmap.net/) to make a new digital soil map of the world using state-of-the-art and emerging technologies for soil mapping and predicting soil properties at fine resolution (100 m). This new global soil map is expected to assist better decisions in a range of global issues like food production, climate change and environmental degradation. The submitted project is designed to initiate a French contribution to this global initiative. It aims to conduct a 'proof-of-concept' test of the proposed digital soil map in Tunisia. The innovative aspect of this French contribution is the extensive use of spatial spectroscopy as a new source of data that is expected to provide direct estimates of some soil surface key properties (carbon content, soil texture, calcium carbonate,…) that can considerably increase the accuracy of the Digital Soil Mapping outputs. The main study area of the proposed research project is centred around the Lebna catchment, in Tunisia. This 1200 km² area (36°52'N ; 10°52'E) constitutes a suitable proof of concept zone that may anticipate the digital soil mapping of the northern Africa and middle East area. The proposed scientific program is broken into three work packages and to three distinct scientific domains: soil property mapping by hyperspectral imagery (WP1), Digital Soil Mapping of soil properties (WP2) and Environmental Modelling using DSM outputs (WP3). The Workpackage 1 aims to predict all possible soil surface properties from hyperspectral imagery over large areas. It includes the elaboration of spectrotransfert functions for predicting simultaneously several soil properties from Visible and Near Infrared spectrum of soil. These spectro-transfert functions will then be adapted for the fine-scale soil properties mapping using airborne vis-NIR data, over bare soils. Finally, the case of mixed surface with soil but also vegetation and rock will be adressed following methods such as Independent Component Analysis for extracting soil spectrum from mixed surface spectrum. The Workpackage 2 aims to perform the digital soil mapping of surface and sub-surface soil properties from the hyperspectral imagery predictions made in the previous work package. Soil surface properties will be predicted by extending the hyperspectral imagery results obtained in the previous step to the areas that remain inaccessible because of a too large amount of perturbing components. This suppose to develop a specific conditional simulaion approach. Although hyperspectral imagery cannot directly estimate sub-surface soil properties we expect that using this new variable as input of DSM function together with relief indicators will greatly improve the prediction of these properties. This is the joint use of this two new set of spatial data that will be innovative here. The general objective of the work package 3 is to determine the added value of Digital Soil Mapping over currently available soil databases for the environmental modelling. We have selected for that two case studies of environmental modelling that have been already undertaken in previous research programs in Tunisia and that will be re-applied in this program: i) the Spatialized assessment of soil vulnerability to erosion and ii) the Mapping wheat yield in water limited situations. The consortium of the project include three French teams (LISAH, EMMAH and BioSP) three Tunisian teams and an Australian one. It covers all the competences required by the objective of the project.

Plant viruses are responsible for a significant proportion of crop diseases and are very difficult to combat due to the scarcity of effective countermeasures, placing them among the most important agricultural pathogens. Most emerging infectious diseases of plants are caused by viruses, being emergence provably favored by climate change, increased seasonal weather instability, and conditioned by intensive global trade. Viral diseases can affect food quality as well as reducing yields, yet quality is also affected by measures such as spraying with pesticides to kill off insect vectors. With no doubt, the best strategy for the control of plant viruses consist of the use of crop cultivars or varieties that are genetically resistant. Thus, introgressions of durable resistance to plant viruses into elite cultivars as well as the identification of new sources of resistance are major goals in plant research. About half of the ~200 known virus resistance genes in plants are recessively inherited, suggesting that recessive resistance is more common for viruses than for other plant pathogens. The use of such genes is a very important tool in breeding programs. Based on previous knowledge on recessive resistance to plant viruses generated by the participating teams, we propose the identification of new plant susceptibility factors whose modification would confer recessive virus resistance. We aim at transferring this knowledge to at least three important crop species: barley, tomato and melon.