You are on page 1of 3

Detection of the fungal infection in post-harvest

onions by an electronic nose


Malgorzata Labanska Sascha Jenkins Sarah van Amsterdam
The Plant Breeding and School of Life Sciences, School of Life Sciences,
Acclimatization Institute - National University of Warwick University of Warwick
Research Institute Warwick, UK Warwick, UK
Bonin, Poland sascha.jenkins.1@warwick.ac.uk sarah.van-amsterdam@warwick.ac.uk
m.labanska@ihar.edu.pl
John Clarkson James Covington
School of Life Sciences, School of Engineering,
University of Warwick University of Warwick
Warwick, UK Warwick, UK
john.clarkson@warwick.ac.uk j.a.covington@warwick.ac.uk

Among emerging novel trend streams in agriculture, sometimes be seen in the field, a major problem for the
2022 IEEE International Symposium on Olfaction and Electronic Nose (ISOEN) | 978-1-6654-5860-3/22/$31.00 ©2022 IEEE | DOI: 10.1109/ISOEN54820.2022.9789676

techniques for the analysis of the volatile organic compounds industry is the rotting of bulbs in store post-harvest. Here we
emitted from plants are gaining increasing interest. In this demonstrate that electronic nose can discriminate between
work, the electronic nose abilities to detect onions infected with samples with varying proportions of infected onion bulbs and
Fusarium oxysporum f. sp. cepae as well as to discriminate show that storage temperature affects the rate of
between samples with varying proportion of infected onion development of Fusarium basal rot.
bulbs were evaluated. It was demonstrated that changes in the
samples related to the development of the disease can be
observed on the PCA plot. It was also proved that storage of II. EXPERIMENTAL
the onions at a low temperature inhibits the development of the
infection. Presented findings of the preliminary studies A. Sample preparation
revealed that electronic nose is a promising analytical tool for Twenty onion bulbs were inoculated with Fusarium
the detection of the infection of the onion bulbs. oxysporum f. sp. cepae (FOC) isolate FUS2. The outer dry
scales of the bulbs were removed, and a 2-3 mm slice of the
Keywords—electronic nose, volatile organic compounds, gas basal plate was removed with a scalpel, before spraying
analysis, onions, plant disease detection
bulbs with 70% ethanol. Once dry, an 8 mm plug of potato
dextrose agar (PDA) containing the actively growing edge of
I. INTRODUCTION FOC isolate FUS2 was inverted and placed on the cut basal
Plant diseases caused by pathogens and pests lead to plate of each bulb. Non-inoculated bulbs were also prepared
numerous short- and long-term consequences, including in the same way, except no agar plug was placed on the basal
economic loss, reduced availability of food, decreased food plate. Bulbs were placed in damp boxes in sealed plastic bags
quality and possible accumulation of toxins during storage. and incubated at 20˚C.
Early plant disease detection enables a quick response to After initial incubation the non-inoculated and inoculated
these issues, by informing application of pesticides, different bulbs were placed in the sealed 3 L plastic boxes with inlet
management of the infected crops (e.g. elimination or selling and outlet fittings added at both ends (Fig 1.). Each box
into the animal feed market), or as in the case of stored crops contained healthy and inoculated bulbs (a total of five bulbs)
such as onion or potato, increasing ventilation or changing at varying proportions resulting in six samples, i.e. 0%, 20%,
storage conditions to minimise further disease development 40%, 60%, 80% and 100% infected bulbs. Samples were
or spread [1,2]. Among emerging novel trends in agriculture prepared and stored for 5 weeks at 25ºC. To maintain high
methods based on the analysis of the volatile organic humidity throughout the experiment, bulbs were suspended
compounds (VOCs) released by plants are gaining increasing on mesh over some water.
attention. The idea of analysing VOC profiles lies in the
ability of plants to release various compounds according to
their health status and the biochemical response to pests or
pathogens [3,4]. The electronic nose systems are potentially
useful tools for the plant disease detection and recognition of
infected plants, before occurring visual symptoms of the
disease. So far, the electronic nose systems have been
applied for the disease detection of a wide range of crops
including onions [5], potatoes [6], rice [7], apples [8] and
wheat [9].
The aim of this work was to evaluate the ability of
electronic nose technology to detect onion bulbs infected
with the soilborne fungal pathogen Fusarium oxysporum f.
sp. cepae, which causes Fusarium basal rot disease.
Although above-ground symptoms of this disease can
This work has been supported by the Polish National Agency for Academic Fig. 1. A measurement setup with PEN 3 electronic nose.
Exchange within a framework of the Bekker programme
(PPN/BEK/2020/1/00383).

978-1-6654-5860-3/22/$31.00 ©2022 IEEE


Authorized licensed use limited to: VIT University. Downloaded on August 03,2022 at 13:20:57 UTC from IEEE Xplore. Restrictions apply.
Additionally, two more samples with 0% and 100% infected W1C
100%
20
onion bulbs were stored at 4ºC to determine the influence of W3S W5S 80%
60%
temperature storage on infection development. 15 40%
20%
B. Electronic nose analysis 10 0%
The PEN 3 electronic nose system (Airsense Analytics W2W 5 W3C
GmbHm Germany) equipped with 10 MOS sensors was used
to analyse the volatiles released by the onion bulbs. Each 0
sensor exhibits sensitivity towards a different class of
volatile compounds (Tab. 1) [10]. A wire resistor generated
high temperature that induce interactions between volatiles W2S W6S
from the sample and the metal-oxide semiconductor surface
resulting in changes in the conductance of the sensor.
Recorded sensor’s signal is the ratio of the resistance of a
sensor exposed to a sample and resistance of the sensor W1W W5C
exposed to a clean air. The measurements were performed
twice every week, for 5 weeks starting from the week after W1S

inoculation. Prior to the measurements the boxes stored at Fig. 2. The gas sensors responses towards samples with varying disease
4ºC were placed at 25ºC to reduce influence of the rate stored at 25ºC. Measurements made two weeks after inoculation.Onion
temperature on the sensors signals. The headspace from the bulbs stored in the plastic boxes.
each box was pumped in the sensor chamber at a constant
These results are in good accordance with a GC-MS
rate of 400 mL/min and sensors signals were recorded for
analysis from the literature [4] which revealed that various
180 s with 2 s intervals. Before each measurement, a 60 s
VOCs are released during development of the Fusarium
cleaning phase was applied, and the baseline was set for 10 s.
infection in onions including sulphuric organic compounds
The average values of the differential sensors’ responses
such as methyl propyl sulphide, and methyl propyl
recorded between 170-174 s formed the datasets that were
disulphide, and dimethyl sulphide, alcohols and ketones. It
subjected to analysis using Principal Component Analysis
can be seen that several sensors contribute to the volatile
(PCA). The raw sensors signals were autoscaled in columns.
patterns of the samples, and therefore multivariate data
Data analysis was conducted employing MATLAB software
approach is required to discriminate between samples with
(Math Works Inc., Natick, MA).
varying proportion of the infected and non-infected bulbs.
Over time, visual symptoms of the disease were
III. RESULTS AND DISCUSSION observed. Red-brown discoloration and rot were developed
First, responses of the sensors were evaluated based on on the base of the bulbs, and the tissue became soft.
the radar plots. A radar chart of the sensors responses Significantly earlier symptoms have been observed in the
towards samples with varying proportion of the infected case of bulbs stored at 25 C than 4C.
onion bulbs recorded two weeks after inoculation is The responses of the sensors towards samples consisting
presented on Fig. 2. It was found that patterns of the sensors of 100% and 0% infected bulbs kept under different storage
responses vary depend on the proportion of the infected conditions were processed by PCA to determine the
onion bulbs. The greatest contribution to the differences in influence of storage temperature on the disease development.
those patterns was the sensor selective towards sulphuric - The resulting PCA score plot is presented in Fig 3. Each
organic substances (i.e. sulphides). Furthermore, sensors point represents a different storage duration (the lighter
generally described as selective towards alcohols, methane colour the longer storage). The first two principal
and broad range of organic substances demonstrated components capture 90% of the variance of the dataset. Two
significant changes in responses. Generally, the responses of clusters of the samples are visible on the plot. The first one
the sensors increased with an increase in the proportion of consisting of the non-infected samples and samples at the
the infected onion bulbs. The smallest responses were early stage of infection (low PC1 values and high PC 2
recorded for samples consisting of non-inoculated onions values), and the second one with infected samples.
(0%), whereas the highest signals were observed for samples
with 80% and 100% infected onions.
3

TABLE I. DESCRIPTION OF THE PEN 3 ELECTRONIC NOSE SYSTEM


SENSORS 2
PC 2 (13.59%)

Sensors General description 1


W1C aromatic compounds
W5S broad range of compounds 0

W3C ammonia, aromatic compounds


-1
W6S hydrogen 100%, 25°C
W5C alkens, aromatic compounds 0%, 25°C
-2
W1S methane, broad range of compounds 100%, 4°C
0%, 4°C
W1W sulphur - organic compounds -3
-4 -2 0 2 4 6 8 10
W2S alcohols, broad organic compounds
PC 1(76.71%)
W2W aromatic compounds, sulphur organic compounds
W3S alkens, aliphatic organic compounds Fig. 3. The PCA score plots of the samples stored at different
temperatures

Authorized licensed use limited to: VIT University. Downloaded on August 03,2022 at 13:20:57 UTC from IEEE Xplore. Restrictions apply.
The greatest differences in the location of the samples on
the PCA plot were observed in the case of the sample
1
consisting of 100% infected onion bulbs stored at 25ºC. The
correlation between location on the PCA (especially towards
PC1) and the development of the infection is noticeable. The time

PC 2(4.12%)
higher value of the PC1 the more severe disease. Minor
0
differences between samples kept at 4ºC suggest that little
disease development occurred at this storage temperature.
Low temperature is commonly used in commercial storages 100%
80%
to prevent development of disease. The least differences 60%
-1 time
between samples were noticed in the case of the non- 40%
20%
inoculated onion bulbs stored at 4ºC. 0%
Finally, sensors signals recorded during four weeks of the -4 -2 0 2 4 6
experiment were subjected to PCA processing score plot of PC 1(91.08%)
the samples consisted of varying ratios of inoculated to non-
inoculated onion bulbs stored at 25ºC is shown in Fig 4. In Fig. 4. PCA score plot of the samples with different disease rate and
this case, prior to the data processing, a feature extraction with varying storage time.
step was implemented that involved the elimination of the
insignificant sensors’ signals from the dataset. Therefore, the REFERENCES
dataset was formed only by sensors signals that differed
significantly depending on the sample tested. This led to [1] I. Buja, E. Sabella, A.G. Monteduro, M.S. Chiriacò, L. De Bellis, A.
Luvisi, G. Maruccio, “Advances in plant disease detection and
better discrimination between samples. The first two monitoring: From traditional assays to in-field diagnostics”, Sensors.
principal components captured 95% of the variance of the vol. 21, pp. 1–22, March 2021.
dataset. It was found that samples with lower proportion of [2] N. Dudareva, F. Negre, D.A. Nagegowda, I. Orlova, “Plant volatiles:
the infected onion bulbs are placed close to each other and Recent advances and future perspectives”, CRC. Crit. Rev. Plant Sci.
are characterized by low PC1 and PC2 values, whereas as the vol. 25, pp. 417–440, October 2006.
ratio of inoculated to non-inoculated onions increases, the [3] A. Cellini, S. Blasioli, E. Biondi, A. Bertaccini, I. Braschi, F. Spinelli,
samples are more dispersed in the plot and have significantly “Potential applications and limitations of electronic nose devices for
plant disease diagnosis”, Sensors. vol. 17, 2596., November 2017.
higher value of PC1. Analysis of the loading plot (data not
[4] S. Blasioli, E. Biondi, I. Braschi, U. Mazzucchi, C. Bazzi, C.E. Gessa,
shown) revealed that all six sensors significantly contribute “Electronic nose as an innovative tool for the diagnosis of grapevine
to presented discrimination. Furthermore, as the storage time crown gall”, Anal. Chim. Acta. vol. 672, pp. 20–24, July 2010.
increases, the samples are placed further from each other and [5] C. Li, N. E., Schmidt., R. Gitaitis, “Detection of onion postharvest
have higher values of PC1 indicating the development of the diseases by analyses of headspace volatiles using a gas sensor array
infection. On the other hand, samples slightly overlap each and GC-MS”, LWT. vol. 44, pp. 1019-1025, May 2011.
other suggesting that a more advanced pattern recognition [6] M. F. Rutolo, D. Iliescu, J. P. Clarkson, J. A. Covington, “Early
method might improve discrimination between samples. identification of potato storage disease using an array of metal-oxide
Nevertheless, PCA results have shown that the location of based gas sensors”, Postharvest Biol. Technol. vol. 116, pp. 50-58,
June 2016.
the sample on the plot is related to the disease severity.
[7] A. Jiarpinijnun, K. Osako, U. Siripatrawan, ”Visualization of
volatomic profiles for early detection of fungal infection on storage
IV. CONCLUSIONS Jasmine brown rice using electronic nose coupled with
In this work, the PEN 3 electronic nose system was chemometrics”, Measurement, vol. 157, pp. 107561, June 2020.
applied to discriminate between onions infected with [8] A. Cellini et al., “Early detection of bacterial diseases in apple plants
Fusarium oxysporum f. sp. cepae and those non-inoculated by analysis of volatile organic compounds profiles and use of
electronic nose”, Ann. Appl. Biol. vol. 168, pp. 409-420, May 2016.
as well as to differentiate between samples with varied
[9] J. Eifler, E. Martinelli, M. Santonico, R. Capuano, D. Schild , C. Di
proportion of infected onion bulbs based on the volatiles Natale, “Differential detection of potentially hazardous Fusarium
released by the onions. Additionally, the influence of species in wheat grains by an electronic nose”, PloS one, vol. 6, pp.
temperature storage on the development of Fusarium e21026, June 2011.
infection was also determined. The sensors responses [10] A. Wang, M.N. Islam, A. Johansen, M. Haapalainen, S. Latvala, M.
showed significant differences between volatiles released by Edelenbos, “Pathogenic Fusarium oxysporum f. sp. cepae growing
onions non-inoculated and those infected with Fusarium inside onion bulbs emits volatile organic compounds that correlate
with the extent of infection”, Postharvest Biol. Technol. vol. 152, pp.
species with varying disease ratios. Sensor sensitive towards 19–28, June 2019.
sulphuric organic compounds was found to have the highest
contribution to differentiating samples. It was shown that
PCA analysis of the sensors’ responses provided clear
discrimination between diseased samples and uninfected
controls. Moreover, changes in the samples related to the
development of the disease were observed on the PCA plot.
It was also shown that storage of the onions at low
temperature inhibits the development of the disease. The
findings of this preliminary study suggest that electronic
nose is a promising analytical tool for the detection Fusarium
basal rot in onion bulbs.

Authorized licensed use limited to: VIT University. Downloaded on August 03,2022 at 13:20:57 UTC from IEEE Xplore. Restrictions apply.

You might also like