Gaël Thébaud

French National Research Institute for Agriculture, Food and Environment (INRAE) · PHIM

Improvement of management strategies of epidemics is often hampered by constraints on experiments at large spatiotemporal scales. A promising approach consists of modeling the biological epidemic process and human interventions, which both impact disease spread. However, few methods enable the simultaneous optimization of the numerous parameters of...

Epidemiological models are increasingly used to predict epidemics and improve management strategies. However, they rarely consider landscape characteristics although such characteristics can influence the epidemic dynamics and, thus, the effectiveness of disease management strategies. Here, we present a generic in silico approach which assesses the...

Identifying the key factors underlying the spread of a disease is an essential but challenging prerequisite to design management strategies. To tackle this issue, we propose an approach based on sensitivity analyses of a spatiotemporal stochastic model simulating the spread of a plant epidemic. This work is motivated by the spread of sharka, caused...

Characterising the spatio-temporal dynamics of pathogens in natura is key to ensuring their efficient prevention and control. However, it is notoriously difficult to estimate dispersal parameters at scales that are relevant to real epidemics. Epidemiological surveys can provide informative data, but parameter estimation can be hampered when the tim...

The optimization of management strategies for plant diseases is a difficult task because of the complexity and variability of epidemic dynamics. Thanks to their ability to numerically simulate many scenarios, models can be used to estimate epidemiological parameters, assess the effectiveness of different management strategies and optimize them. Thi...

To avoid the activation of plant defenses and ensure sustained feeding, aphids are assumed to use their mouthparts to deliver effectors into plant cells. A recent study has shown that effectors detected near feeding sites are differentially distributed in plant tissues. However, the precise process of effector delivery into specific plant compartme...

High-throughput sequencing has opened the route for a deep assessment of within-host genetic diversity that can be used, e.g., to characterize microbial communities and to infer transmission links in infectious disease outbreaks. The performance of such characterizations and inferences cannot be analytically assessed in general and are often ground...

In recent decades, a legion of monopartite begomoviruses transmitted by the whitefly Bemisia tabaci has emerged as serious threats to vegetable crops in Africa. Recent studies in Burkina Faso (West Africa) reported the predominance of pepper yellow vein Mali virus (PepYVMLV) and its frequent association with a previously unknown DNA-B component. To...

Sharka is a major viral disease of stone fruits (Prunus) worldwide, reducing both fruit quality and yield. Caused by Plum pox virus (PPV, family Potyviridae, genus Potyvirus), the disease is spread over long distances by the propagation and transport of infected plant material and locally by aphids in a non-persistent manner. Apart from strategies...

[This corrects the article DOI: 10.1371/journal.pcbi.1006085.].

AVIS de l’Agence nationale de sécurité sanitaire de l’alimentation, de l’environnement et du travail relatif à « la stratégie de lutte vis-à-vis de Xylella fastidiosa - phase 2 ». L’Anses a été saisie le 28 septembre 2020 par la Direction générale de l’alimentation pour la réalisation de l’expertise suivante : demande d'appui scientifique et techni...

Inferring the dispersal processes of vector-borne plant pathogens is a great challenge because the plausible epidemiological scenarios often involve complex spread patterns at multiple scales. The spatial genetic structure of ‘Candidatus Phytoplasma prunorum’, responsible for European stone fruit yellows disease, was investigated by the application...

AVIS de l’Agence nationale de sécurité sanitaire de l’alimentation, de l’environnement et du travail relatif à « la stratégie de lutte vis-à-vis de Xylella fastidiosa ». L’Anses a été saisie le 26 novembre 2019 par la Direction générale de l’alimentation pour la réalisation de l’expertise suivante : demande d’avis relatif à la stratégie de lutte vi...

Wheat dwarf virus, transmitted by the leafhopper Psammotettix alienus in a persistent, non-propagative manner, infects numerous species from the Poaceae family. Data associated with wheat dwarf virus (WDV) suggest that some isolates preferentially infect wheat while other preferentially infect barley. This allowed to define the wheat strain and the...

Inferring the dispersal processes of vector-borne plant pathogens is a great challenge because the plausible epidemiological scenarios often involve complex spread patterns at multiple scales. European stone fruit yellows (ESFY), a disease caused by ‘Candidatus Phytoplasma prunorum’ and disseminated via planting material and vectors belonging to th...

UMR BFP - Equipe Virologie ; UMR BGPI : Equipe 6

Pathogen sequence data have been exploited to infer who infected whom, by using empirical and model-based approaches. Most of these approaches exploit one pathogen sequence per infected host (e.g. individual, household, field). However, modern sequencing techniques can reveal the polymorphic nature of within-host populations of pathogens. Thus, the...

Cauliflower mosaic virus (CaMV, family Caulimoviridae ) responds to the presence of aphid vectors on infected plants by forming specific transmission morphs. This phenomenon, coined transmission activation (TA), controls plant-to-plant propagation of CaMV. A fundamental question is whether other viruses rely on TA. Here, we demonstrate that transmi...

Multipartite viruses package their genomic segments independently and thus incur the risk of being unable to transmit their entire genome during hostto- host transmission if they undergo severe bottlenecks. In this paper, we estimated the bottleneck size during one infection cycle of Faba bean necrotic stunt virus (FBNSV), an octopartite nanovirus...

Cumulative detected incidence for the kernels tested in the simulation study. For each kernel, the seven tested introduction scenarios are represented by different colours. For each combination of kernel and introduction scenarios, 10 independant simulated epidemics are shown. (TIFF)

Boxplots of the variation among estimated dispersal kernels. Impact of (A) estimation scenario, (B) kernel range, and (C) disease introduction scenario [number of introduction patches (with initial disease prevalence)] on the precision of estimated dispersal kernels. Precision is measured by the span of the 95% credibility interval of Kullback-Leib...

Cumulative detected incidence for the introduction scenarios tested in the simulation study. For each introduction scenario, the number of introduction patches and the corresponding initial disease prevalence are mentionned above the graph. The three tested kernels are represented by different colours. For each combination of kernel and introductio...

Best-fit BWME kernel approximations of 2Dt kernels. The kernels corresponding to 4 values of the shape parameter are represented by their cumulative distribution function F1D (top) and the associated probability density function f1D (bottom) of the distance travelled. Green dashed line: mean distance travelled. (TIFF)

Influence of introduction scenarios on the estimation of a short-range dispersal kernel. For each introduction scenario, 10 epidemics were simulated with a short-range kernel (black dashed curve), and 10 MCMC chains were run per simulated epidemic. The posterior distributions of the kernel obtained under the most exhaustive estimation scheme (Θ4) a...

Influence of introduction scenarios on the estimation of a long-range dispersal kernel. For each introduction scenario, 10 epidemics were simulated with a long-range kernel (black dashed curve), and 10 MCMC chains were run per simulated epidemic. The posterior distributions of the kernel obtained under the most exhaustive estimation scheme (Θ4) are...

Best-fit BWME kernel approximations of exponential-power kernels. The kernels corresponding to 4 values of the shape parameter are represented by their cumulative distribution function F1D (top) and the associated probability density function f1D (bottom) of the distance travelled. Green dashed line: mean distance travelled. (TIFF)

Best-fit BWME kernel approximations of power-law kernels. The kernels corresponding to 4 values of the shape parameter are represented by their cumulative distribution function F1D (top) and the associated probability density function f1D (bottom) of the distance travelled. Green dashed line: mean distance travelled. (TIFF)

Influence of introduction scenarios on the estimation of a medium-range dispersal kernel. For each introduction scenario, 10 epidemics were simulated with a medium-range kernel (black dashed curve), and 10 MCMC chains were run per simulated epidemic. The posterior distributions of the kernel obtained under the most exhaustive estimation scheme (Θ4)...

Cumulative detected incidence at the end of year 22 across the range of detection sensitivities (ρ) tested in the dedicated simulation study. Each polygon represents one peach orchard. All eight simulations start at year 1 from a unique introduction patch with 25% initial prevalence and spread is determined by the long-range kernel. Note that the f...

Influence of detection sensitivity on the distance between simulated and estimated long-range dispersal kernels. For each of the 99 detection sensitivities, a single epidemic was simulated using the long-range kernel. For three levels of prior information, each bar represents a 95% credibility interval on the Kullback-Leibler distance (KLD) between...

Simulated planting years and introduction patch locations used in the simulation study. The first map (top left) represents the randomisation of the first planting years of the 553 patches. These years were sampled without replacement from their empirical distribution. The other maps show the location and planting year of each introduction patch in...

Comparison of simulated and estimated dispersal kernels. From left to right: kernels with the minimum, lower quartile, median, upper quartile and maximum Kullback-Leibler (KL) distances (posterior mean), for all chains with non-negligible mean posterior likelihood. Estimations (red) under the most exhaustive scheme (Θ4) are based on simulated epide...

Estimated weights of the (BWME) dispersal kernel for the sharka epidemic. The posterior distribution of the weights (calculated with (Eq 11) for a mixture of 100 exponential kernels) is obtained for κ = 11 (i.e. the number of introduction patches maximising the Fisher information). The plotted posterior distribution of weights (as a function of the...

Comparison of simulated and estimated nuisance parameters. For each combination of short-, medium- and long-range kernels (from top to bottom) and introduction scenarios (colour-coded as in S3, S8, S9 and S10 Figs), 10 epidemics were simulated and 10 MCMC chains were run per simulated epidemic. The curves represent the posterior distribution of the...

Influence of detection sensitivity on the estimation of the long-range dispersal kernel. For each detection sensitivity, a single epidemic was simulated using the long-range kernel (black dashed curve). The posterior distributions of the estimated kernels (obtained from all MCMC chains with non-negligible mean posterior likelihood) are shown for th...

Estimated dispersal density for the sharka epidemic. The posterior distribution of the marginal probability density function, f1D, of the fitted dispersal kernel, obtained for κ = 11 (i.e. the number of introduction patches maximising the Fisher information). The plotted posterior distributions were obtained from 4000 MCMC samples. One line is plot...

(A) Probabilistic framework for statistical inference, (B) prior distributions, (C) Markov chain Monte Carlo, and (D) model selection for κ. (PDF)

During the past decade, knowledge of pathogen life history has greatly benefited from the advent and development of molecular epidemiology. This branch of epidemiology uses information on pathogen variation at the molecular level to gain insights into a pathogen's niche and evolution and to characterize pathogen dispersal within and between host po...

Of worldwide economic importance, Tomato yellow leaf curl virus (TYLCV, Begomovirus) is responsible for one of the most devastating plant diseases in warm and temperate regions. The DNA begomoviruses (Geminiviridae) are transmitted by the whitefly species complex Bemisia tabaci. Although geminiviruses have long been described as circulative non-pro...

The relative durations of the incubation period (the time between inoculation and symptom expression) and of the latent period (the time between inoculation and infectiousness of the host) are poorly documented for plant diseases. However, the extent of asynchrony between the ends of these two periods (i.e., their mismatch) can be a key determinant...

The genetic determinism of viral traits can generally be dissected using either forward or reverse genetics because the clonal reproduction of viruses does not require the use of approaches based on laboratory crosses. Nevertheless, we hypothesized that recombinant viruses could be analyzed as sexually reproducing organisms, using either a quantita...

Many plant epidemics that cause major economic losses cannot be controlled with pesticides. Among them, sharka epidemics severely affect prunus trees worldwide. Its causal agent, Plum pox virus (PPV; genus Potyvirus), has been classified as a quarantine pathogen in numerous countries. As a result, various management strategies have been implemented...

The within-host diversity of virus populations can be drastically limited during between-host transmission, with primary infection of hosts representing a major constraint to diversity maintenance. However, there is an extreme paucity of quantitative data on the demographic changes experienced by virus populations during primary infection. Here, th...

Sharka, caused by Plum pox virus (PPV), is a severe disease of trees belonging to the Prunus family. In various countries the management of sharka epidemics relies on the removal of symptomatic trees. However, this strategy supposes that the infectious trees are symptomatic and vice versa. This assumption implies a synchrony between the date when a...

Biological invasions are the main causes of emerging viral diseases and they favour the co-occurrence of multiple species or strains in the same environment. Depending on the nature of the interaction, co-occurrence can lead to competitive exclusion or coexistence. The successive fortuitous introductions of two strains of Tomato yellow leaf curl vi...

Advances in sequencing technology coupled with new integrative approaches to data analysis provide a potentially transformative opportunity to use pathogen genome data to advance our understanding of transmission. However, to maximize the insights such genetic data can provide, we need to understand more about how the microevolution of pathogens is...

FMDV complete genomes for the 2001 dataset. (FASTA)

Simulated outbreak. Trees with the five highest posterior probabilities (coloured disks) and true transmissions (black circles). (TIF)

2007 UK epidemics. Posterior distributions (histograms) of parameters. Top four panels: ; dotted-dashed curve: prior distribution; solid line: posterior median; dotted lines: posterior quantiles 0.025 and 0.975. Bottom left: transmission kernel, depending on parameter ; solid curve: posterior median; dotted-dashed curves: posterior quartiles 0.25 a...

2001 epidemics, Darlington cluster without premise B. Posterior distributions (histograms) of parameters. Top four panels: ; dotted-dashed curve: prior distribution; solid line: posterior median; dotted lines: posterior quantiles 0.025 and 0.975. Bottom left: transmission kernel which depends on parameter ; solid curve: posterior median; dotted-das...

Epidemiological data for the 2001 dataset. (TXT)

Epidemiological data for the 2007 dataset. (TXT)

Simulated outbreak. Posterior distributions of infection times (top) and latency durations (bottom left) for the simulated outbreak. In both panels, vertical solid lines indicate the true values. In the top panel, vertical dashed lines indicate the virus observation times. (TIF)

Genetic network, based on statistical parsimony, implemented in the software package TCS [16]. Full dots represent observed genomes, while empty dots represent unsampled genomes (for these last ones, timing is arbitrary), links represent single mutations. Top panel: subset of the Darlington cluster, 2001 UK FMDV epidemics [13]; bottom panel: 2007 U...

FMDV complete genomes for the 2007 dataset. (FASTA)

Example of transmissions between four spatially-confined premises . Bold lines: time intervals appearing in Equation (2) in Text S1, over which the true conditional distributions of observed sequences can be computed. (TIF)

Simulated outbreak. Posterior distributions (histograms) of parameters. Top four panels: parameters ; dashed line: true value; dotted-dashed curve: prior distribution; solid line: posterior median; dotted lines: posterior quantiles 0.025 and 0.975. Bottom left: transmission kernel, depending on parameter ; dashed curve: true kernel; solid curve: po...

2001 epidemics, Darlington cluster without premise B. Trees with the highest posterior probabilities (coloured disks). (TIF)

Spatial representation of the tree with the highest posterior probability, for different parametrisations of the prior distribution for the veterinarian assessment of the age of the oldest lesion on a premise. Left column: 2007 epidemics, right column: cluster in the 2001 epidemics. Top: prior variance of equal to . Center: prior variance of set to...

Posterior distributions (histograms) - simulated outbreak with 100 farms. Posterior distributions of mean latency duration (; left) and mean transmission distance (; right); dashed lines: true values; dotted-dashed curves: prior distributions; solid lines: posterior medians; dotted lines: posterior quantiles 0.025 and 0.975. (TIF)

2007 UK epidemics. Trees with the five highest posterior probabilities (coloured disks). (TIF)

2007 UK epidemics. Posterior distributions of infection times (top) and latency durations (bottom left). In the top panel, vertical dashed lines indicate the virus observation times. (TIF)

2001 epidemics, Darlington cluster including premise B. Top left: Posterior probabilities of transmissions (dots with varying size). Top right: Tree with the highest posterior probability mapped in space (arrows). Bottom: Trees with the five highest posterior probabilities (coloured disks). (TIF)

Transmissions for the simulated outbreak with 100 farms. True transmissions are indicated with circles; dot sizes are proportional to posterior probabilities of transmissions. (TIF)

Additional criteria to assess the performance of the estimation algorithm over three series of 100 simulations (test, 2007, 2001). Criteria are the coverages by the 95% posterior intervals of the infection times, the times at which the premises became infectious, the transmission parameters (source strength and dispersion parameter) and the latency...