Predicting symptoms of downy mildew, powdery mildew, and gray mold diseases of grapevine through machine learning

Authors

  • Iride Volpi AEDIT s.r.l., Pontedera, Pisa, Italy
  • Diego Guidotti AEDIT s.r.l., Pontedera, Pisa, Italy
  • Michele Mammini AEDIT s.r.l., Pontedera, Pisa, Italy
  • Susanna Marchi AEDIT s.r.l., Pontedera, Pisa, Italy

DOI:

https://doi.org/10.36253/ijam-1131

Keywords:

agrometeorology, ERA5, grape, IPM models, monitoring networks

Abstract

Downy mildew, powdery mildew, and gray mold are major diseases of grapevine with a strong negative impact on fruit yield and fruit quality. These diseases are controlled by the application of chemicals, which may cause undesirable effects on the environment and on human health. Thus, monitoring and forecasting crop disease is essential to support integrated pest management (IPM) measures. In this study, two tree-based machine learning (ML) algorithms, random forest and C5.0, were compared to test their capability to predict the appearance of symptoms of grapevine diseases, considering meteorological conditions, spatial indices, the number of crop protection treatments and the frequency of monitoring days in which symptoms were recorded in the previous year. Data collected in Tuscany region (Italy), on the presence of symptoms on grapevine, from 2006 to 2017 were divided with an 80/20 proportion in training and test set, data collected in 2018 and 2019 were tested as independent years for downy mildew and powdery mildew. The frequency of symptoms in the previous year and the cumulative precipitation from April to seven days before the monitoring day were the most important variables among those considered in the analysis for predicting the occurrence of disease symptoms. The best performance in predicting the presence of symptoms of the three diseases was obtained with the algorithm C5.0 by applying (i) a technique to deal with imbalanced dataset (i.e., symptoms were detected in the minority of observations) and (ii) an optimized cut-off for predictions. The balanced accuracy achieved in the test set was 0.8 for downy mildew, 0.7 for powdery mildew and 0.9 for gray mold. The application of the models for downy mildew and powdery mildew in the two independent years (2018 and 2019) achieved a lower balanced accuracy, around 0.7 for both the diseases. Machine learning models were able to select the best predictors and to unravel the complex relationships among geographic indices, bioclimatic indices, protection treatments and the frequency of symptoms in the previous year. 

References

Allen C., 1976. A modified sine wave method for calculating degree days. Environmental Entomology, 5: 388–396.

Barka E. A., Gognies S., Nowak J., Audran J. C., Belarbi A. 2002. Inhibitory effect of endophyte bacteria on Botrytis cinerea and its influence to promote the grapevine growth. Biological Control, 24: 135-142.

Bove F., Savary S., Willocquet L., Rossi V., 2020. Designing a modelling structure for the grapevine downy mildew pathosystem. European Journal of Plant Pathology, 158: 599-614.

Buonassisi D., Colombo M., Migliaro D., Dolzani C., Peressotti E., Mizzotti C., Velasco R., Masiero S., Perazzolli M., Vezzulli S., 2017. Breeding for grapevine downy mildew resistance: a review of “omics” approaches. Euphytica, 213:1–21.

Caffi T., Rossi V., Bugiani R., 2010. Evaluation of a warning system for controlling primary infections of grapevine downy mildew. Plant Disease, 94: 709-716.

Caffi T., Rossi, V., Legler S. E., Bugiani R., 2011. A mechanistic model simulating ascosporic infections by Erysiphe necator, the powdery mildew fungus of grapevine. Plant Pathology, 60: 522-531.

Carmichael P. C., Siyoum N., Jongman M., Korsten L., 2018. Prevalence of Botrytis cinerea at different phenological stages of table grapes grown in the northern region of South Africa. Scientia Horticulturae, 239: 57-63.

Carisse O., Bacon R., Lefebvre A., Lessard K., 2009. A degree-day model to initiate fungicide spray programs for management of grape powdery mildew [Erysiphe necator]. Canadian Journal of Plant Pathology, 31: 186–194.

Chen M., Brun F., Raynal M., Makowski D. 2020. Forecasting severe grape downy mildew attacks using machine learning. Plos One, 15(3), e0230254

Ciliberti N., Fermaud M., Languasco L., Rossi V., 2015 (a). Influence of fungal strain, temperature, and wetness duration on infection of grapevine inflorescences and young berry clusters by Botrytis cinerea. Phytopathology, 105: 325–333.

Ciliberti N., Fermaud M., Roudet J., Rossi V., 2015 (b). Environmental conditions affect Botrytis cinerea infection of mature grape berries more than the strain or transposon genotype. Phytopathology, 105: 1090–1096.

Copernicus Climate Change Service (C3S) (2017): ERA5: Fifth generation of ECMWF atmospheric reanalyses of the global climate. Copernicus Climate Change Service Climate Data Store (CDS). https://cds.climate.copernicus.eu/cdsapp#!/home.

Elmer P.A.G., Michailides T.J., 2007. Epidemiology of Botrytis cinerea in orchard and vine crops. In: Elad Y, Williamson B, Tudzynski P, Delen N, editors. Botrytis: Biology, Pathology and Control. Springer Neth- erlands, pp. 243–272.

Gadoury D.M. and Pearson R. C., 1988. Initiation, development, dispersal, and survival of cleistothecia of Uncinula necator in New York vineyards. Phytopathology, 78:1413-1421.

Gadoury D.M., Seem R.C., Pearson R.C., Wilcox W.F., 2001. Effects of powdery mildew on vine growth, yield, and quality of Concord grapes. Plant Disease, 85:137–140.

Gadoury D.M., Cadle-Davidson L., Wilcox W.F., Dry I.B., Seem R.C., Milgroom M.G., 2012. Grapevine powdery mildew (Erysiphe necator): A fascinating system for the study of the biology, ecology and epidemiology of an obligate biotroph. Molecular Plant Pathology, 13: 1–16.

Gessler C., Pertot I., Perazzolli M., 2011. Plasmopara viticola: a review of knowledge on downy mildew of grapevine and effective disease management. Phytopathologia Mediterranea, 50:3–44.

González-Domínguez E. Caffi T., Ciliberti N. Rossi V. 2015. A mechanistic model of Botrytis cinerea on grapevines that includes weather, vine growth stage, and the main infection pathways. PLoS One, 10(10), e0140444.

Harrell F.E. Jr, 2019. Hmisc: Harrell Miscellaneous. R package version 4.2-0. https://CRAN.R-project.org/package=Hmisc.

Hijmans R.J., 2019 a. raster: Geographic Data Analysis and Modeling. R package version 2.8-19. https://CRAN.R-project.org/package=raster.

Hijmans R.J., 2019 b. geosphere: Spherical Trigonometry. R package version 1.5-10. https://CRAN.R-project.org/package=geosphere.

Hill G. N., Beresford R. M., Evans K. J., 2019. Automated analysis of aggregated datasets to identify climatic predictors of botrytis bunch rot in wine grapes. Phytopathology, 109: 84-95.

Jaspers M. V., Seyb A.M., Trought M.C.T., Balasubramaniam R., 2013. Overwintering grapevine debris as an important source of Botrytis cinerea inoculum. Plant Pathology, 62: 130–138.

John C.R., 2020. MLeval: Machine Learning Model Evaluation. R package version 0.3. https://cran.r-project.org/package=MLeval.

Kaur H., Pannu H.S., Malhi A.K., 2019. A systematic review on imbalanced data challenges in machine learning: Applications and solutions. ACM Computing Surveys, 52. Article 79.

Kuhn M., 2020. caret: Classification and Regression Training. R package version 6.0-85. https://CRAN.R-project.org/package=caret.

Lafon R., Clerjeau M., 1988. Downy mildew. In: Pearson, R.C., Goheen, A.C. (Eds.), Compendium of Grape Diseases. APS Press, St. Paul, Minnesota, USA, pp. 11–13.

Lee, D.S., Bae, Y.S., Byun, B.K., Lee, S., Park, J.K., Park, Y.S., 2019. Occurrence prediction of the citrus flatid planthopper (Metcalfa pruinosa (Say, 1830)) in South Korea using a random forest model. Forests, 10, 583.

Legler S.E., Caffi T., Rossi V., 2011. A non linear model for temperature-dependent development of Erysiphe necator chasmothecia on grapevine leaves. Plant Pathology, 61: 96-105.

Lu W., Newlands N. K., Carisse O., Atkinson D. E., Cannon A. J. 2020. Disease Risk Forecasting with Bayesian Learning Networks: Application to Grape Powdery Mildew (Erysiphe necator) in Vineyards. Agronomy, 10(5), 622.

Marchi S., Guidotti D., Ricciolini M., Petacchi R., 2016. Towards understanding temporal and spatial dynamics of Bactrocera oleae (Rossi) infestations using decade-long agrometeorological time series. International Journal of Biometeorology, 60: 1681-1694.

Molitor D., Baus O., Hoffmann L., Beyer M., 2016. Meteorological conditions determine the thermal-temporal position of the annual Botrytis bunch rot epidemic on Vitis vinifera L. cv. Riesling grapes. Oeno One, 50: 231–244.

Orlandini S., Gozzini B., Rosa M., Egger E., Storchi P., Maracchi G., Miglietta F., 1993. PLASMO: a simulation model for control of Plasmopara viticola on grapevine 1. EPPO Bulletin, 23: 619-626.

Pellegrini A., Prodorutti D., Frizzi A., Gessler C., Pertot I., 2010. Development and evaluation of a warning model for the optimal use of copper in organic viticulture. Journal of Plant Pathology, 43-55.

Pertot I., Caffi T., Rossi V., Mugnai L., Hoffmann C., Grando M. S., Gary C., Lafond D., Duso C., Thiery D., Mazzoni V., Anfora G., 2017. A critical review of plant protection tools for reducing pesticide use on grapevine and new perspectives for the implementation of IPM in viticulture. Crop Protection, 97: 70-84.

R Core Team, 2020. R: A language and environment for statistical computing. R Foundation for Statistical Computing, Vienna, Austria. URL https://www.R-project.org/.

Rodríguez-Rajo F. J., Jato V., Fernández-González M., Aira M. J. 2010. The use of aerobiological methods for forecasting Botrytis spore concentrations in a vineyard. Grana, 49: 56-65.

Rossi V., Caffi T., Giosuè S., Bugiani R., 2008. A mechanistic model simulating primary infections of downy mildew in grapevine. Ecological Modelling, 212: 480–491.

Rossi V., Salinari F., Poni S., Caffi T., Bettati T. 2014. Addressing the implementation problem in agricultural decision support systems: the example of vite. net®. Computers and Electronics in Agriculture, 100: 88-99.

Rügner A., Rumbolz J., Huber B., Bleyer G., Gisi U., Kassemeyer H.H., Guggenheim R., 2002. Formation of overwintering structures of Uncinula necator and colonization of grapevine under field conditions. Plant Pathology, 51: 322–330.

Sperschneider J., 2019. Machine learning in plant–pathogen interactions: empowering biological predictions from field scale to genome scale. New Phytologist, 228: 35–41.

Valdés‐Gómez H., Araya‐Alman M., Pañitrur‐De la Fuente C., Verdugo‐Vásquez N., Lolas M., Acevedo‐Opazo C., Gray C., Calonnec A., 2017. Evaluation of a decision support strategy for the control of powdery mildew, Erysiphe necator (Schw.) Burr., in grapevine in the central region of Chile. Pest Management Science, 73: 1813-1821.

Volpi I., Guidotti D., Mammini M., Petacchi R., Marchi S., 2021. Managing complex datasets to predict Bactrocera oleae infestation at the regional scale. Computers and Electronics in Agriculture, in press.

Zou Q., Xie S., Lin Z., Wu M., Ju Y., 2016. Finding the Best Classification Threshold in Imbalanced Classification. Big Data Research, 5: 2–8.

Downloads

Published

2021-12-27

How to Cite

Volpi, I., Guidotti, D., Mammini, M., & Marchi, S. (2021). Predicting symptoms of downy mildew, powdery mildew, and gray mold diseases of grapevine through machine learning. Italian Journal of Agrometeorology, (2), 57-69. https://doi.org/10.36253/ijam-1131

Issue

Section

RESEARCH ARTICLES