Living organisms have evolved complex signaling networks to drive appropriate physiological processes in response to changing environmental conditions. Amongst them, electric signals are a universal method to rapidly transmit information. In animals, bioelectrical activity measurements in the heart or the brain provide information about health status. In plants, practical measurements of bioelectrical activity are in their infancy and transposition of technology used in human medicine could therefore, by analogy provide insight about the physiological status of plants. This paper reports on the development and testing of an innovative electrophysiological sensor that can be used in greenhouse production conditions, without a Faraday cage, enabling real-time electric signal measurements. The bioelectrical activity is modified in response to water stress conditions or to nycthemeral rhythm. Furthermore, the automatic classification of plant status using supervised machine learning allows detection of these physiological modifications. This sensor represents an efficient alternative agronomic tool at the service of producers for decision support or for taking preventive measures before initial visual symptoms of plant stress appear.

Herbivorous arthropods, such as spider mites, are one of the major causes of annual crop losses. They are usually hard to spot before a severe infestation takes place. When feeding, these insects cause external perturbation that triggers changes in the underlying physiological process of a plant, which are expressed by a generation of distinct variations of electrical potential. Therefore, plant electrophysiology data portray information of the plant state. Analyses involving machine learning techniques applied to plant electrical response triggered by spider mite infestation have not been previously reported. This study investigates plant electrophysiological signals recorded from 12 commercial tomatoes plants contaminated with spider mites and proposes a workflow based on Gradient Boosted Tree algorithm for an automated differentiation of the plant’s normal state from the stressed state caused by infestation. The classification model built using the signal samples recorded during daylight and employing a reduced feature subset performs with an accuracy of 80% in identifying the plant’s stressed state. Furthermore, the Hjorth complexity encloses the most relevant information for discrimination of the plant status. The obtained findings open novel access towards automated detection of insect infestation in greenhouse crops and, consequently, more optimal prevention and treatment approaches.

Automated monitoring of plant health is becoming a crucial component for optimizing agricultural production. Recently, several studies have shown that plant electrophysiology could be used as a tool to determine plant status related to applied stressors. However, to the best of our knowledge, there have been no studies relating electrical plant response to general stress responses as a proxy for plant health. This study models general stress of plants exposed to either biotic or abiotic stressors, namely drought, nutrient deficiencies or infestation with spider mites, using electrophysiological signals acquired from 36 plants. Moreover, in the signal processing procedure, the proposed workflow reuses information from the previous steps, therefore considerably reducing computation time regarding recent related approaches in the literature. Careful choice of the principal parameters leads to a classification of the general stress in plants with more than 80% accuracy. The main descriptive statistics measured together with the Hjorth complexity provide the most discriminative information for such classification. The presented findings open new paths to explore for improved monitoring of plant health.

Improved non-destructive instrumental approaches for grading fruit during post-harvest could be an efficient way to monitor stock in the apple industry. The objective of this study was to evaluate the ability of visible-near infrared (vis-NIR) spectroscopy in reflectance mode for classifying apples left on the shelf or stored in a cooled room. The ability of NIR spectroscopy to classify the duration of storage of three apple cultivars in two storage modalities was evaluated. A total of 450 fruit, sampled after 7, 14, 28, 60, 90 and 120 days of storage in a cooled room (CR) and 7, 14 and 28 days in shelflife (SL), has been studied. The classification of these modalities was analysed by factorial discriminant analysis (FDA) pooling the spectral data of all cultivars (global models) into a common data set. Then, the cultivar effect on the classification of the same modalities was analysed by processing data from each cultivar in separate factorial descriminant analyses. A preliminary analysis showed the genetic variability of spectral data due to the three apple cultivars. We show that vis-NIR spectroscopy allowed the correct classification of the fruits of each cultivar by more than 95%. The classification relied on both vis and NIR absorption bands: 500, 680, 1400 to 1700, 1850, 1950, 2200 and 2300 nm. We show that storage modalities of global models can be classified by more than 75% and 83% for fruits stored in a cooled room and shelf, respectively. Classification of the same storage modalities was improved by cultivar models with percentage of individuals correctly classified of 86% (Gala), 89% (Elstar) and 85% (Smoothee) for fruits stored in a cooled room and 95% (Gala), 98% (Elstar) and 95% (Smoothee) for fruits left in shelflife. We conclude that despite the slight increase of efficiency of the models when we considered each apple cultivar separately, global models applicable to a set of different cultivars presents a correct level of classification and could be usefull for some commercial applications.