Frontiers in Plant Science (Aug 2020)

Can Spore Sampler Data Be Used to Predict Plasmopara viticola Infection in Vineyards?

  • Chiara Brischetto,
  • Federica Bove,
  • Luca Languasco,
  • Vittorio Rossi

DOI
https://doi.org/10.3389/fpls.2020.01187
Journal volume & issue
Vol. 11

Abstract

Read online

Grapevine downy mildew (DM) is caused by the dimorphic oomycete Plasmopara viticola, which incites epidemics through primary and secondary infection cycles that occur throughout the season. The secondary infection cycles are caused by the sporangia produced on DM lesions. The current research examined the relationship between numbers of airborne sporangia and DM development on grape leaves to determine whether spore sampler data can be useful to predict the potential for secondary infections of P. viticola. Three years (2015–2017) of spore sampler data confirmed that sporangia are a common component of the airborne microflora in a DM-infested vineyard and that their numbers depend on weather conditions. For a total of 108 days, leaf samples were collected from the vineyard at 2- to 3-day intervals and incubated under optimal conditions for P. viticola infection. The numbers of airborne sporangia sampled on 1 to 7 days before leaf sampling were significantly correlated with the numbers of DM lesions on the leaves. The best correlation (r=0.59), however, was found for the numbers of viable airborne sporangia (SPV), which were assessed by using equations driven by the vapour pressure deficit. In Bayesian and ROC curve analyses, SPV was found to be a good predictor of P. viticola infection of grape leaves, with AUROC=0.821 and false positive predictions mainly occurring at low SPV. A binary logistic regression showed that a threshold of 2.52 viable sporangia m-3 air day-1 enables a prediction of no infection with a posterior probability of 0.870, which was higher than the prior probability of 0.574. Numbers of viable sporangia in the vineyard air is therefore a useful predictor of infection and especially of no infection. The predictor missed some observed infections, but these infections were not severe (they accounted for only 31 of 374 DM lesions).

Keywords