Quantitative Resistance Deployment Can Strengthen Epidemics in Perennial Plants by Selecting Maladapted Pathogen Strains

Abstract

Quantitative resistances are essential tools for mitigating epidemics in managed plant ecosystems. However, their deployment can drive evolutionary changes in pathogen life-history traits, making predictions of epidemic development challenging. To investigate these effects, we developed a demo-genetic model that explicitly captures feedbacks between the pathogen's population demography and its genetic composition. The model also links within-host multiplication and between-host transmission, and is built on the assumption that the coexistence of susceptible and resistant hosts imposes divergent selection pressures on the pathogen population at the landscape scale. We simulated contrasting landscapes of perennial host plants with varying proportions of resistant plants and resistance efficiencies. Our simulations confirmed that deploying resistances with nearly complete efficiency (> 99.99%) effectively reduces the severity of epidemics caused by pathogen introduction and promotes the specialization of infectious genotypes to either susceptible or resistant hosts. Conversely, the use of partial resistances induces limited evolutionary changes, often resulting in pathogen maladaptation to both susceptible and resistant hosts. Notably, deploying resistances with strong (89%) or moderate (60%) efficiencies can, under certain conditions, lead to higher host mortality compared to entirely susceptible populations. This counterintuitive outcome arises from the maladaptation of infectious genotypes to their hosts, which prolongs the lifespan of infected hosts and can increase inoculum pressure. We further compared simulations of the full model with those of simplified versions in which (i) the contribution of infected plants to disease transmission did not depend on the pathogen load they carried, (ii) plant landscapes were not spatially explicit. These comparisons highlighted the essential role of these components in shaping model predictions. Finally, we discuss the conditions that may lead to detrimental outcomes of quantitative resistance deployments in managed perennial plants.

Keywords: evolutionary epidemiology; host–microbe interaction; nested modeling; plant pathogen; resistance deployment strategy; virulence evolution.

FIGURE 1

A quantitative relationship between the pathogen's infection strategy Z and its ability to infect and colonise initially susceptible (black curve) and resistant (gray curve) hosts. The dotted vertical lines represent the optimal trait values conferring the highest chance of infecting and colonizing susceptible ($x={\mathrm{\Theta }}_S$, black line) and resistant ($x={\mathrm{\Theta }}_R$, gray line) hosts, respectively. The introduced pathogen population consistently expressed a mean infection strategy value of 0. Four levels of resistance efficiency, defined by four distinct intensities of local stabilizing selection, were considered. These levels reduced the probability of infection by initial fungal genotypes by nearly 100% (a), 89% (b), 60% (c), and 30% (d) for resistant hosts. For each resistance efficiency, the fitness curves for susceptible and resistant hosts had the same width, but optimal infection strategy values differed. Therefore, any adaptation of the pathogen to one type of host (initially resistant or susceptible) implied a symmetrical maladaptation to the other type.

FIGURE 2

Percentage of healthy compartments, at equilibrium, as a function of the proportion of resistant hosts in the landscape and the efficiency of resistance. In these scenarios, dead host plants were left unreplaced. The first infected host was either susceptible (left column) or resistant (right column). Each point represents the mean of 96 replicates of a scenario. Each light triangle indicates the percentage of infected hosts within a single replicate.

FIGURE 3

Percentage of healthy compartments, at equilibrium, as a function of the proportion of resistant hosts in the landscape and the efficiency of resistance. Here dead plants were systematically replaced by healthy ones. The first infected plant was either susceptible (left column) or resistant (right column). Each point represents the mean of 96 replicates of a scenario. Each light triangle indicates the percentage of infected hosts within a single replicate.

FIGURE 4

Proportion of healthy compartments over the years since pathogen introduction. The first infected plant was either susceptible (left column) or resistant (right column). In these scenarios, dead plants were left unreplaced. Each line denotes the mean of 96 replicates of a scenario.

FIGURE 5

Mean number of active lesions as a function of time elapsed since pathogen introduction. The first infected plant was either susceptible (left column) or resistant (right column). Each line is the mean of 96 replicates of a scenario, involving a given proportion of resistant plants in the landscape.

FIGURE 6

Mean infection strategy value of pathogen populations established on susceptible hosts (left column) and resistant hosts (right column), tracked over the years following pathogen introduction. In this scenario, the first infected host was resistant. The thick lines represent the mean infection strategy values calculated across 96 replicates of a scenario. Each bar indicates the standard deviation of the corresponding mean value. Each light line represents the infection strategy within a single replicate. The opacity of the thick lines and bars was reduced when the number of non‐null replicates fell below 10. The black and gray dashed lines correspond to the optimal infection strategy values that confer the highest chance of infecting susceptible (${\mathrm{\Theta }}_S=0$) and resistant hosts (${\mathrm{\Theta }}_R=6$), respectively.