Florian Hartig

← Publications

Identifying local drivers of a vector-pathogen-disease system using Bayesian modeling

Bernd Panassiti, Florian Hartig, Johannes Fahrentrapp, Michael Breuer & Robert Biedermann

Basic and Applied Ecology, 18, 75–85 (2017)
Cite this
@article{panassiti2017identifying,
  author = {Bernd Panassiti and Florian Hartig and Johannes Fahrentrapp and Michael Breuer and Robert Biedermann},
  title = {Identifying local drivers of a vector-pathogen-disease system using Bayesian modeling},
  journal = {Basic and Applied Ecology},
  volume = {18},
  pages = {75--85},
  year = {2017},
  doi = {10.1016/j.baae.2016.11.005},
}

DOI: 10.1016/j.baae.2016.11.005
Citation statistics: 16 (Google Scholar) · 13 (OpenAlex), as of 29 October 2026

Google Scholar citations per year, as of 21 October 2026 (yearly counts are fetched separately from the total)

Funding

This study was supported through funding by Association of German Wine Scientists.

View article (DOI)

Abstract

‘Bois noir’ is a phytoplasma-mediated grapevine yellows disease that causes great economic damage in European vineyards. Previous studies have examined habitat relationships on a regional scale, which help to better understand the large-scale epidemiology. Local drivers, such as micro-habitat preferences of the vector (Hyalesthes obsoletus, a cixiid planthopper), or local interactions with reservoir host plants, however, are still poorly understood, although this knowledge is crucial for developing site-specific management strategies. Here, we examined the local environment-species relationships of a phytoplasma-mediated grapevine disease on a scale of 15 m in a 2.9 ha vineyard using: (i) data on elevation and habitat types; (ii) cover of host plants Urtica dioica and Convolvulus arvensis over three seasons, (iii) vector monitoring over four seasons; (iv) genetic tests for phytoplasma presence in the vector; and (v) inspection of 6056 grapevine plants for visual symptoms of the ‘bois noir’ disease. The data were analyzed in a joint causal model that describes the interplay between vector, pathogen, disease and environment, estimated with Bayesian inference. Our results indicate that surrounding natural and semi-natural vegetation (fallow land, forest and managed agricultural land) and high density of the major host plant U. dioica are associated with an increase in vector population densities. Higher vector population densities at low availability of U. dioica were associated with higher phytoplasma infection rates in the vector. The prevalence of disease symptoms in grapevine plants was nonetheless more affected by grapevine cultivar and higher elevation than by the estimated availability of infected vectors. The results of our local analysis support current bois noir management recommendations stating that (1) removal of the host plant U. dioica should be best carried out in either spring or autumn; and (2) grapevine cultivars are unequally susceptible. Moreover, we provide evidence that U. dioica control before the flight period may result in low U. dioica densities and high H. obsoletus population densities, causing an increase in vector infection rates and disease pressure.

What the paper shows and why it matters

(AI-generated)

Crop diseases caused by phytoplasma bacteria depend on several players: an insect that carries the pathogen (the vector), the plants that harbour it, and the crop. This makes them hard to predict. Habitat models describe the large-scale preferences of such disease systems but, as the paper puts it, may miss small-scale biotic interactions and micro-habitat effects. Analysing the chain from environment to symptoms step by step does not use the data optimally, and path analysis, which fits chains jointly, is limited for binary and count data. The problem the paper takes on is estimating, at the scale of a single vineyard, how environment, host plants, vector, pathogen and disease depend on one another.

The paper argues that fitting vector counts, pathogen infection in the vector and disease symptoms in one Bayesian model can identify the local drivers of a plant disease, and applies this to bois noir, a phytoplasma disease of grapevine, in a 2.9 ha vineyard in south-western Germany. The planthopper vector, Hyalesthes obsoletus, was more abundant near fallow land, forest and managed grassland and where its host plant, stinging nettle (Urtica dioica), was dense. The share of planthoppers carrying the phytoplasma depended on an interaction between the two: it was higher where planthoppers were dense and nettle scarce. The model would also predict this for the opposite combination, which the paper judges implausible and poorly covered by data. Disease symptoms were associated with cultivar and elevation, though the cultivar estimates were uncertain, but not detectably with the estimated density of infected planthoppers. The paper suspects the analysis lacked the power to detect such a link.

The paper states that the results support existing advice: control nettle in spring or autumn, and expect cultivars to differ in susceptibility. It argues that removing nettle before or during the planthoppers’ flight period may leave many planthoppers on a few remaining, possibly infected plants and so raise disease pressure; it advises delaying host-plant management until the flight ends. Limits remain: no link between infected planthoppers and symptoms was found, and the advice rests on one vineyard and an unevenly covered interaction, not on a removal trial. Murolo et al. (2020) cite the finding on low nettle density and high planthopper abundance in discussing how weed management can drive planthoppers onto grapevines.

References

  • Murolo et al. (2020). Spatial pattern of Bois noir: case study of a delicate balance between disease progression and recovery. Scientific Reports 10, 9801. doi:10.1038/s41598-020-66210-7