Predicting Ascospore Release of Monilinia vaccinii-corymbosi of Blueberry with Machine Learning

Abstract

Mummy berry, caused by Monilinia vaccinii-corymbosi, causes economic losses of highbush blueberry in the U.S. Pacific Northwest (PNW). Apothecia develop from mummified berries overwintering on soil surfaces and produce ascospores that infect tissue emerging from floral and vegetative buds. Disease control currently relies on fungicides applied on a calendar basis rather than inoculum availability. To establish a prediction model for ascospore release, apothecial development was tracked in three fields, one in western Oregon and two in northwestern Washington in 2015 and 2016. Air and soil temperature, precipitation, soil moisture, leaf wetness, relative humidity and solar radiation were monitored using in-field weather stations and Washington State University's AgWeatherNet stations. Four modeling approaches were compared: logistic regression, multivariate adaptive regression splines, artificial neural networks, and random forest. A supervised learning approach was used to train the models on two data sets: training (70%) and testing (30%). The importance of environmental factors was calculated for each model separately. Soil temperature, soil moisture, and solar radiation were identified as the most important factors influencing ascospore release. Random forest models, with 78% accuracy, showed the best performance compared with the other models. Results of this research helps PNW blueberry growers to optimize fungicide use and reduce production costs.

Development of a Decision Support System for the Management of Mummy Berry Disease in Northwestern Washington

“…Different patterns of duration and intensity of apothecial presence were observed in northwestern Washington field sites across growing seasons despite the similarity of synoptic weather conditions across sites within growing seasons. Similar to previous reports [ 7 , 12 , 34 , 35 , 36 , 37 ], apothecial presence was observed for three to four weeks at the low-management sites between February to May. While these periods have been consistently high and low for apothecial presence at the Whatcom and Snohomish sites, respectively, a trend of reduction was observed at the Island and Skagit sites.…”

“…For example, we had to extrapolate the available data for the expected disease levels at temperatures above 18 °C to allow and ensure the operational applicability of this sub-component of the DSS. Environmental factors such as temperature, relative humidity, light, and soil moisture influence apothecial development and longevity [ 3 , 7 , 10 , 11 , 12 ]. In particular, the hydrothermal state of the top layer of the soil appears to be a key driver of carpogenic development of apothecia [ 12 ], and these conditions are approximated using air measurements that served as proxy variables.…”

“…in Maine, USA required 800–3000 h [ 9 ]. Secondly, the environmental factors identified to play a significant role in apothecia development and longevity are temperature, light, relative humidity, and soil moisture [ 3 , 7 , 10 , 11 , 12 ]. Lastly, the timing of apothecial development appears adapted to the timing of growth stages of early or late-maturing genotypes, indicating coevolution of the fungus with its host [ 13 ].…”

“…A similar model was developed to estimate post-chill degree days needed for 25% of the maximum number of pseudosclerotia to produce apothecia in lowbush blueberry systems in Maine [ 9 ]. A preliminary model was developed to predict apothecial development for highbush blueberry in the PNW [ 12 ] but it has not been validated to date. The Mummy Berry Forecast System, developed by Delbridge and Hildebrand [ 14 ], is a method of applying curative fungicides within 72 h of infection periods, which are identified by air temperature and surface wetness.…”

Modeling risk of Sclerotinia sclerotiorum-induced disease development on canola and dry bean using machine learning algorithms

“…ANN could predict disease development with high accuracy in classification and regression analyses on both crops, whereas the accuracies of other models, including LGR, were affected by the crop, the type of analysis, and the predictors used. Nevertheless, other ML techniques, e.g., RF, have produced excellent models in other pathosystems 21 , 23 . Thus, it could be said that more modern ML techniques albeit more complex than LGR, may be described as the “next generation” tools for modeling the risk of plant disease development.…”