The oomycete Hyaloperonospora arabidopsis (= parasitica) is a natural pathogen of the Brassicaceae Arabidopsis thaliana, its specific host. This biotrophic parasite causes the loss of some rosette leaves, but only exceptionally kills host plants when they are infected after the true leaves have appeared. The major symptom of the infection is the production of conidiophores on the surface of leaves a few days after infection, giving this parasite its name "downy mildew". These conidiophores bear packets of spores that are the asexual stage of H. arabidopsis and can be transmitted to other plants. The parasite also reproduces sexually via oospores that remain within leaves until host death, and then can reinfect seedlings the next season. . Reproduction by oospores is critical for the survival of this pathogen between the active growing seasons of its host but plays no role in within-season dynamics. Asexual reproduction via conidiospores, on the other hand, is responsible for dissemination and epidemic dynamics within populations and seasons. Therefore we concentrated on only the asexual stage of this pathogen. The seven strains used in the experiments were of three different origins. Three of them, Emco, Emwa and Noco, were "laboratory strains" originated from isolates collected more than ten years ago and since maintained artificially as asexual cultures on specific A. thaliana lines . They were provided by the Sainsbury Laboratory (John Innes Center, Norwich, U.K.). A second type of strains, Ors 3 and Ors 5, were collected in spring 2004 from conidiospores on infected host plants of the same population on the campus of Université Paris-Sud, Orsay, France. The last strains, Fri 3 and Fri 5, were obtained from oospores of two infected plants also sampled in spring 2004 in a population in Fribourg, Switzerland. In both Fribourg and Orsay populations the sampled plants were situated a few meters apart. For Fribourg strains seedlings were experimentally infected with each natural oospore isolate and a single infected seedling per isolate was retained as spore source for subsequent infections. Conidiospores were multiplied for 22 asexual generations for the Orsay strains and 6–7 asexual generations for the Fribourg strains before the experiment. Thus they are likely to represent only a single genotype per strain. These wild strains were tested for infectivity profiles on a series of host lines. All strains, except the two Fribourg ones were distinguishable from each other.
Of the six A. thaliana lines used as hosts in the experiments, five were generated from at least two generations of selfing of plants issued from seed collected in wild populations across Europe: Finland [Fin], England [Gb], Pyrenees [Pyr], Sweden [Sue], and Czech Republic [Tch]. The sixth line was the registered ecotype Tsu, originally sampled in Japan. Our purpose was to test a maximum number of parasite and host strains in a complete matrix of interactions. Major genes for resistance against this parasite are, however, common  and it was difficult to find a large range of ecotypes susceptible to all parasites. Therefore we included a host ecotype (Fin) that was resistant to Emco but susceptible to the other six strains. All other hosts were susceptible to all seven parasite strains.
Controlled cross inoculation experiment
Every host line was subjected to eight treatments, an inoculation with a spore suspension of each of the seven H. arabidopsis strains or a mock inoculation with pure water (control treatment). Five replicates were carried out for each of these 48 combinations. The seeds of each host lines were sown the same day in 5 × 5 × 5 cm compost pots then randomized and placed in the dark at 5–6°C ten days in order to synchronize germination. Seedlings were then grown up to 4–6 leaves in the greenhouse with natural photoperiod (23°C day – 15°C night). Supernumerary seedlings were removed after germination in order to keep only the most central plant in each pot. After 18 days in the greenhouse all the plants receiving the same parasite treatment were regrouped and inoculated with a 8 μL drop of spore suspension for the two largest leaves and a 4 μL drop on all smaller leaves of the corresponding parasite strain, or of pure water for the control treatment. The seven spore suspensions were diluted with water to 6 × 104 spores per mL . Each plant was then placed in its own closed transparent plastic cylinder to prevent contamination and maintain high hygrometry, and then randomly arranged in a growth chamber (10:14 light-dark photoperiod, 16°C ± 3°C average temperature and hygrometry around 95–100%). From the seventh day after inoculation the plants were individually checked twice a week during four weeks, which corresponds to the usual duration of the symptoms in such conditions. At each observation the number of infected leaves and their position on the plant was recorded, and the plant was allowed to transmit spores to a group of three new healthy seedlings at the 4–6 leaves stage of the same host line ("test plants"), following the protocol described in . All the pots containing the test plants inoculated during a same transmission event were kept together in six trays with plastic covers in the same growth chamber as the observed plants. The daily transmission success was then estimated by counting the number of infected leaves on the three test plants eight days after the transmission event. The total transmission success ("transmission") over the whole infection period was estimated as the asymptote of the sigmoid curve fitted to the cumulated daily transmission data of the eight transmission events. When zero or only one out of the eight transmission events led to a successful infection of the test plants, the variable transmission was arbitrary given the value of the cumulative transmission at the eighth day. The total number of infected leaves of each plant was estimated by summing all newly infected leaves over the eight observations. After 35 days in the growth chamber the plants were moved again into the greenhouse (natural photoperiod 23°C day – 15°C night) to complete their life cycle. At that time the covers of the plastic cylinders were removed in order to lower the hygrometry. Plants were watered ad libitum until flowering and then harvested regularly as fruits matured, in order to collect all the seeds before they fell from open fruits. The total weight of all the seeds produced by each plant (measured to the precision of 1/1000 g) was used as our estimate for host fitness. Because A. thaliana is annual, this variable represents its lifespan investment in reproduction.
As there were doubts about a possible contamination of three plants belonging to the control treatments of the respectively Fin, Pyr and Tch host lines, these plants were not included in the statistical analyses. Statistical analyses were performed with JMP version 5.1.2 (SAS institute, Cary, NC). We used nested analyses of variances (ANOVA) to examine the effects of Origin (Laboratory strains, Fribourg and Orsay), Parasite strain (nested within Origin), Host line and the interactions between Host line and Origin, and Host line and Parasite strain on transmission and the number of infected leaves of all the inoculated treatments. ANOVAs were then performed independently for each of the three different origins to test for Host line, Parasite line and Interaction effects on the same variables. As the distribution of the variable transmission was very asymmetrical due to a large number of zeros this variable was subsequently modified with a square root transformation ((Transmission)0.5) in all the ANOVA analyses to obtain residuals as close as possible to a normal distribution. An ANCOVA was carried out to test the relationship between host and parasite fitness for each specific host and parasite combination using mean seed production and transmission of each of the 48 combinations, including the controls. We chose to analyze seed production of control and infected combinations itself instead of a measure of virulence estimated from the difference between healthy and infected plants because this maximizes the information used and permits each host line its own starting point specific to its fecundity. In this analysis, the "Parasite treatment" variable included the three parasite types (Laboratory, Fribourg and Orsay) and Controls. We tested the Host line and Parasite treatment as main effects (the parasite line effect nested within Parasite treatment was removed as it explained almost none of the variance in the model), transmission as covariable and the effects of the interaction term between this covariable and the two main effects. This model enabled us to compare the slopes of the relationship for the different host lines (Host line × Transmission term) and the different parasite origins (Parasite treatment × Transmission term). Note that in these interaction terms the covariable was centered at zero by subtracting its mean so that means rather than intercepts were compared for the main effects.