Validation of a model for maturation of Erysiphe necator chasmothecia

Powdery mildew of grapevine, caused by the ascomycete Erysiphe necator (Schw.) Burr., is a polycyclic disease with primary infections caused by ascospores produced in the overwintered fruiting bodies (called chasmothecia), and secondary infections caused by asexual conidia produced in powdery mildew colonies during the grapevine growing season. Despite an initial controversy about the importance of the ascospores as a relevant source of inoculum for the disease, ascospores are now considered an important form of primary inoculum in many grape-growing areas. Sanitation, i.e. the interventions aimed at reducing the overwintering inoculum that triggers the disease epidemic in the following year, has been already tested in vineyards and is now considered a key part of the disease control strategy in the vineyards with severe attach of powdery mildew after harvest. Sanitation can be performed using different kinds of fungicides, natural compounds or the biocontrol agent Ampelomyces spp., which are all more effective when applied against developing chasmothecia, than on mature or dispersed ones. A model was therefore developed, based on the principles of the systems analysis, to predict the development and maturation of the E. necator chasmothecia in the vineyard. The model considers four state variables, which represent the four developmental stages of the chasmothecia: from the white ascocarp primordia to the yellow, brown and at the end black mature chasmothecia. Chasmothecia advance from one stage of development to the next at specific rates that depend on air temperature and are calculated based on the equations developed by Legler et al. (2012). Mature black chasmothecia are finally dispersed by rain splashes whenever there is a measurable rainfall. To validate the model, a total of 38 commercial vineyards were considered Northern Italy, Baden-W\ufcrttemberg (Germany), and Riverland region (South Australia). In each vineyard, a plot was maintained untreated with fungicides against powdery mildew and disease severity was periodically assessed until leaf fall was complete. In 31 vineyards, the number of chasmothecia dispersed from the powdery mildew colonies were periodically trapped by using plastic trap funnels. The numbers of chasmothecia trapped per cm2 of trap surface was enumerated and expressed as a cumulative proportion of the total chasmothecia trapped in each vineyard; these data formed the first dataset used for model validation. In 8 vineyards, the number of yellow, brown and black chasmothecia produced by powdery mildew colonies on vine leaves was enumerated as numbers/cm2 leaf, and expressed as a proportion of the total number of mature chasmothecia; these data were used as the second dataset for model validation. Model predictions provided a good fit of the actual data collected in different vineyards and years: i) for the first dataset, the R2 for the aggregate data of all the vineyard was 0.95, the s.e. of estimate, which measures the accuracy of predictions, was 0.097, the indexes of goodness-of-fit were close to 1 (NS=0.94,W=0.98,EF=0.94) and MAE was 0.12; ii) for the second dataset, the R2 for the aggregate data was 0.91, the s.e. of estimate was 0.091, the indexes of goodness-of-fit were all close to 1 (NS=0.90,W=0.98,EF=0.90) and MAE was 0.18