ArticlePDF Available

Non-invasive Presymptomatic Detection of Cercospora beticola Infection and Identification of Early Metabolic Responses in Sugar Beet

Authors:

Abstract and Figures

Cercospora beticola is an economically significant fungal pathogen of sugar beet, and is the causative pathogen of Cercospora leaf spot. Selected host genotypes with contrasting degree of susceptibility to the disease have been exploited to characterize the patterns of metabolite responses to fungal infection, and to devise a pre-symptomatic, non-invasive method of detecting the presence of the pathogen. Sugar beet genotypes were analyzed for metabolite profiles and hyperspectral signatures. Correlation of data matrices from both approaches facilitated identification of candidates for metabolic markers. Hyperspectral imaging was highly predictive with a classification accuracy of 98.5–99.9% in detecting C. beticola. Metabolite analysis revealed metabolites altered by the host as part of a successful defense response: these were L-DOPA, 12-hydroxyjasmonic acid 12-O-β-D-glucoside, pantothenic acid, and 5-O-feruloylquinic acid. The accumulation of glucosylvitexin in the resistant cultivar suggests it acts as a constitutively produced protectant. The study establishes a proof-of-concept for an unbiased, presymptomatic and non-invasive detection system for the presence of C. beticola. The test needs to be validated with a larger set of genotypes, to be scalable to the level of a crop improvement program, aiming to speed up the selection for resistant cultivars of sugar beet. Untargeted metabolic profiling is a valuable tool to identify metabolites which correlate with hyperspectral data.
Content may be subject to copyright.
fpls-07-01377 September 19, 2016 Time: 12:36 # 1
ORIGINAL RESEARCH
published: 22 September 2016
doi: 10.3389/fpls.2016.01377
Edited by:
Choong-Min Ryu,
Korea Research Institute
of Bioscience and Biotechnology,
South Korea
Reviewed by:
Vijai Kumar Gupta,
National University of Ireland, Galway,
Ireland
Margaret Daub,
North Carolina State University, USA
*Correspondence:
Hans-Peter Mock
mock@ipk-gatersleben.de
Specialty section:
This article was submitted to
Plant Biotic Interactions,
a section of the journal
Frontiers in Plant Science
Received: 27 May 2016
Accepted: 29 August 2016
Published: 22 September 2016
Citation:
Arens N, Backhaus A, Döll S,
Fischer S, Seiffert U and Mock H-P
(2016) Non-invasive Presymptomatic
Detection of Cercospora beticola
Infection and Identification of Early
Metabolic Responses in Sugar Beet.
Front. Plant Sci. 7:1377.
doi: 10.3389/fpls.2016.01377
Non-invasive Presymptomatic
Detection of Cercospora beticola
Infection and Identification of Early
Metabolic Responses in Sugar Beet
Nadja Arens1, Andreas Backhaus2, Stefanie Döll1, Sandra Fischer3, Udo Seiffert2and
Hans-Peter Mock1*
1Applied Biochemistry, Department of Physiology and Cell Biology, Leibniz Institute of Plant Genetics and Crop Plant
Research, Gatersleben, Germany, 2Biosystems Engineering, Fraunhofer Institute for Factory Operation and Automation,
Magdeburg, Germany, 3Strube Research GmbH & Co. KG, Söllingen, Germany
Cercospora beticola is an economically significant fungal pathogen of sugar beet,
and is the causative pathogen of Cercospora leaf spot. Selected host genotypes
with contrasting degree of susceptibility to the disease have been exploited to
characterize the patterns of metabolite responses to fungal infection, and to devise
a pre-symptomatic, non-invasive method of detecting the presence of the pathogen.
Sugar beet genotypes were analyzed for metabolite profiles and hyperspectral
signatures. Correlation of data matrices from both approaches facilitated identification
of candidates for metabolic markers. Hyperspectral imaging was highly predictive with
a classification accuracy of 98.5–99.9% in detecting C. beticola. Metabolite analysis
revealed metabolites altered by the host as part of a successful defense response:
these were L-DOPA, 12-hydroxyjasmonic acid 12-O-β-D-glucoside, pantothenic acid,
and 5-O-feruloylquinic acid. The accumulation of glucosylvitexin in the resistant cultivar
suggests it acts as a constitutively produced protectant. The study establishes a proof-
of-concept for an unbiased, presymptomatic and non-invasive detection system for the
presence of C. beticola. The test needs to be validated with a larger set of genotypes, to
be scalable to the level of a crop improvement program, aiming to speed up the selection
for resistant cultivars of sugar beet. Untargeted metabolic profiling is a valuable tool to
identify metabolites which correlate with hyperspectral data.
Keywords: phenotyping, hyperspectral imaging, metabolic profiling, metabolomics, Beta vulgaris, LC–MS
INTRODUCTION
The fungal pathogen Cercospora beticola, the causative organism of the sugar beet leaf disease
Cercospora leaf spot, imposes a major constraint over the crop’s yield worldwide. The pathogen
employs the photosensitizer cercosporin to induce necrotic lesions on the leaves which impair the
photosynthetic performance and therefore negatively affect taproot yield (Staerkel et al., 2013).
Host variations of resistance are known, but its physiological basis is not fully understood. The
genetic basis of this resistance is very often complex, although there is one example known of a race-
specific, monogenic resistance (Lewellen and Whitney, 1976;Whitney and Lewellen, 1976;Koch
et al., 2000). Based on the observation that effectors of C. beticola suppress the transcription of the
Frontiers in Plant Science | www.frontiersin.org 1September 2016 | Volume 7 | Article 1377
fpls-07-01377 September 19, 2016 Time: 12:36 # 2
Arens et al. Presymtomatic Cercospora Detection in Sugar Beet
host’s gene encoding phenylalanine ammonia lyase, it has
been proposed that phenylpropanoids are involved in the
defense response (Schmidt et al., 2004, 2008). Thus adopting
a metabolomic approach to characterize the resistance reaction
may well-prove to be highly informative.
Two generalized strategies are followed in metabolomic
profiling: the first is non-targeted in the sense that an attempt
is made to identify as many metabolites as possible, thereby
enabling the comparison of metabolite profiles and identification
of disease resistance marker; in contrast the targeted approach
focuses on quantifying a set of pre-determined compounds.
The physiological status of the leaf in the course of a
compatible (susceptible) and incompatible (resistant) host–
pathogen interaction has been successfully monitored in some
detail using the former approach (Fiehn, 2002;Viant and
Sommer, 2013). Metabolomic profiling using reverse phase liquid
chromatography coupled with high resolution mass spectrometry
is commonly applied for the detection of semi-polar compounds.
Complex matrices (e.g., plant extracts) are separated by polarity
and mass/charge (m/z) values are determined. Software for
automated peak detection, deconvolution and alignment is used
to set up a data matrix for multivariate statistic analysis.
Traditionally, strategies to minimize yield loss comprise either
in breeding efforts for resistant cultivars or the application
of fungicides. Preventing disease outbreak is a more cost-
efficient and ecologically sustainable option. Fungicides and their
application are not only monetary factors, but may also be
overcome by resistant fungal strains and have less of a negative
impact on the environment which can lead to even higher
consequential expenses (Bolton et al., 2013).
The hyperspectral signature of the leaf, which reflects its
physiological status, varies in response to a number of imposed
biotic or abiotic stresses (Blackburn, 1998, 2007). Hyperspectral
imaging has the advantage of being a non-invasive assay and
can be applied to follow a dynamic process. The method relies
on the capture of the reflectance spectrum containing specific
absorption properties of the subject in response to exposure to
broadband illumination, typically in the visible (400–700 nm),
near-infrared (700–1100 nm), and short-wave infrared range
(1100–2500 nm; Mutka and Bart, 2014). The hyperspectral
signatures are acquired in a pixelwise manner, whereby each
pixel recorded from a hyperspectral line camera is containing an
individual spectral profile. Artificial neural network computing
(ANN) is a machine learning technique that when applied to
hyperspectral signatures can be utilized to predict classes, e.g.,
healthy or diseased plant (Backhaus et al., 2011). The approach
has been used to evaluate the physiological and disease status
of plants both in a greenhouse setting and conducting airborne
sensing of field crops (Lelong et al., 1998;Asner et al., 2007;
Gowen et al., 2007;Backhaus et al., 2011;Carvalho et al., 2012).
Hyperspectral images combine three data dimensions whereby
two are of spatial nature and one of spectral, giving rise to a wealth
of information to investigate, e.g., plant phenotypes (Fiorani and
Schurr, 2013).
The availability of a non-invasive, highly predictive and
early detection of this disease is attractive, as it will allow
improvements in crop management and in the selection
efficiency toward breeding for disease resistance (Mutka and
Bart, 2014). Hyperspectral imaging has been demonstrated
to be feasible for the detection of developed symptoms and
identification of several sugar beet leaf diseases, including
Cercospora leaf spot (Mahlein et al., 2010, 2012). However, the
transition from simply detecting symptoms on the leaf surface
caused by the presence of the fungus to presymptomatic (without
visible symptoms) infection recognition requires both a more
advanced mathematical data analysis and the general existence
of a pathogen-specific defense reaction of the host plant. Here,
the aim was to develop the means to non-invasively detect
presymptomatic Cercospora leaf spot disease in selected sugar
beet cultivars which differed from one another with respect
to disease susceptibility. The intention was firstly to provide
a proof-of-concept for a screening method usable within the
context of a breeding program targeted at improving the level of
resistance to Cercospora leaf spot; and secondly, to characterize
which metabolic pathway(s) were affected early in the process
of infection. A focus was set on phenolic compounds, which are
known to contribute to defense either as preformed or as induced
by pathogen. Hyperspectral and metabolic data correlated well
and facilitated filtering for relevant metabolites.
MATERIALS AND METHODS
Plant Material
Three sugar beet cultivars (STR1, STR2, and STR3), differing
in their level of susceptibility to Cercospora leaf spot (STR1
resistant, STR2 tolerant, and STR3 susceptible), were grown in
pots in a greenhouse (Strube, Schlanstedt, Germany) held at 60%
relative humidity, 18/22C (day/night) and a 16 h photoperiod.
The cultivars are genetically diverse. The plants were kept fully
watered and fertilized with 0.1% N/P/K solution once a week. The
experiments were initiated after about 8 weeks, at which point the
plants had reached growth stage 16 (Meier et al., 1993).
Stable Isotope Labeling with 15N
Sugar beet seed was germinated on moist filter paper and kept
in the dark. After 7 days, seedlings were transferred to half
strength Hoagland’s media (Hoagland and Arnon, 1950) and
grown hydroponically for 6 weeks at 20/18C (day/night) under
a 16 h photoperiod provided by 300 µmol m2s1light. Half of
the seedlings were subjected to medium containing 15NH415 NO3
(atom % 98, Campro Scientific GmbH, Berlin, Germany). Leaves
of 6 weeks old plants were harvested.
Inoculation with C. beticola
The isolate Holtensen 2011 (obtained from Mark Varrelmann, IfZ
Göttingen) of C. beticola was used for the experiments. C. beticola
isolate was multiplied by culturing them on agar-solidified V8
medium at 25C under natural daylight for 2 weeks. Mycelia and
spores were scraped off the plate and suspended in sterile water
to produce an inoculum containing 50,000 infectious units per
mL, which was sprayed uniformly over the test plants, and the
spraying was repeated 2 h later. After inoculation a foil tunnel
was established to obtain a humidity of 95%. The lights were
Frontiers in Plant Science | www.frontiersin.org 2September 2016 | Volume 7 | Article 1377
fpls-07-01377 September 19, 2016 Time: 12:36 # 3
Arens et al. Presymtomatic Cercospora Detection in Sugar Beet
switched off for the next 3 days, after which the foil covering was
removed and the growing conditions were set to deliver 28/20C
(day/night) and a 14 h photoperiod.
Quantification of C. beticola Biomass
Fungal biomass was quantified according to De Coninck et al.
(2012) with modifications. DNA was extracted from 100 mg
frozen leaf material or plate-grown fungal mycel using DNeasy
Plant Mini Kit (50) (Qiagen, Venlo, Netherlands) according
to the manufacturer’s instructions. Real time quantitative PCR
(qRT-PCR) was performed with a LightCycler R
480 (Roche
Diagnostics, Risch, Switzerland) using 40 cycles (95C 10 s, 60C
30 s). In contrast to De Coninck et al. (2012) CercoCal1 and
SbEc1 were amplified in separate reactions, each containing 5 µl
SsoAdvanced SYBR Green Supermix (Bio-Rad, Hercules, CA,
USA), 2 µl DNA extract (10 ng/µl), 1.5 µl primermix (2 pmol/µl)
and 1.5 µl water. C. beticola infestation was calculated by two
methods, the comparative method (1ct =ctCercoCal1 – ctSbEc1)
or by utilizing a calibration curve of fungal DNA (Supplementary
Figure S1). Statistics including Student’s t-test were performed
using Microsoft Excel1software.
Hyperspectral Image Acquisition and
Pre-processing
Young sugar beet leaves where placed on a low reflective base,
which was moved on a translation stage mounted 1 m below a
HySpex SWIR-320m-e line camera (Norsk Elektro Optikk A/S,
Skedsmokorset, Norway). To calibrate the set-up for relative
reflectance, a standard optical PTFE (polytetrafluoroethylene)
card was included in each image. The spectra captured in the
infra-red range (970–2,500 nm) had a spatial resolution of 6 nm
and yielded a 256 dimensional spectral vector per pixel. The
camera line’s spatial resolution was 320 pixel, and its maximum
frame rate was 100 fps. Radiometric calibration was performed
using the vendor’s software package. Image processing was
performed using a workflow system implemented in Matlab (The
MathWorks, Inc., Natick, MA, USA). The standard calibration
pad is automatically detected within the image and extracted.
Each spectrum was referenced against the calibration pad mean
spectra to obtain a set of relative reflectance spectra per pixel.
Next, the spectral signatures were grouped into five clusters on
the basis of their similarity, as measured by Euclidean distance.
Grouping was performed by Neural Gas Clustering (Martinetz
and Schulten, 1991). Each pixel belongs to one particular
cluster and these clusters were used to obtain proper image
segmentation. Clusters representing leaf material were selected
automatically by means of reference spectra manually selected in
one image. The pre-processing steps for extracting pixel spectra
representing plant material are illustrated in Supplementary
Figure S2.
Machine Learning for Spectral Data
Analysis
A machine learning approach was followed to generate a
classification system to enable the pre-symptomatic detection
1https://products.office.com/en-gb/excel
of Cercospora leaf spot. As a first step, the machine learning
approach was tested for its capability to differentiate the cultivars
and in the second step to differentiate the infection status per
cultivar. An ANN was trained to convert the input (spectral
information) into category (infection state, cultivar). During
the training phase, the parameters were adjusted to reduce the
prediction error as much as possible. LDA was performed in
order to search for a linear combination of features capable to
discriminate between two or more classes of objects or events.
Note that the number of features or dimensions resulting from an
LDA is N-1 where N is the number of classes. The classification
was based on both a single model Radial-Basis Function (RBF)
Network (Moody and Darken, 1989;Backhaus et al., 2012)
and on a multi-model approach; for the latter, the first layer
assumed an RBF model with various metrics to calculate spectral
similarities, after which a fusion RBF model was applied to
combine the output of all the single models (Knauer et al.,
2014). The models were validated using a standard five-fold cross
validation design, in which the data were assigned to five equally
sized partitions, four of which were used for training and the fifth
for validation. Mean classification rates across test samples are
reported.
Systematic dependencies of metabolite data and spectral data
were revealed by applying an RBF Network to map spectral
reflectance input data onto metabolite output data. Differentially
abundant metabolites were ordered on the basis of their observed
FCs. The RBF Network was trained following a “leave-one-
out” scheme, in which the data captured from one replicate
or leaf number was held in reserve until the model had been
derived from the remaining data. The predicted peak intensity
for the reserve sample was then used to calculate a coefficient of
determination (R2), which varies from 1 (perfect prediction) to 0
(zero prediction).
Extraction of Semi-Polar Compounds
Leaf material was harvested, analyzed by the hyperspectral
imaging system and snap- frozen in liquid nitrogen. The frozen
leaf was then lyophilized and milled to a powder, of which a 15 mg
aliquot was extracted in 900 µL aqueous methanol (75% v/v)
with formic acid (0.1% v/v). Cell disruption was promoted by the
inclusion of 1.0–1.2 mm diameter zirconium beads (58% ZrO2)
(Muehlmeier, Baernau, Germany). The samples were centrifuged
(27,500 ×g, 10 min, 4C), and 20 µL 0.1% formic acid was
added to an 80 µL aliquot of the resulting supernatant. After a
second centrifugation (27,500 ×g, 5 min, 4C), the supernatant
was subjected to liquid chromatography–mass spectrometry
(LC–MS).
Metabolite Profiling via
(U)HPLC-UV-ESI-MS
Chromatographic separation and UV detection were carried
out using a Dionex Ultimate 3000 RSLC system (Thermo
Fisher Scientific, Inc., Waltham, MA, USA). The LC system was
fitted with an Acquity UPLCR
BEH Phenyl column (130 Å,
1.7 µm 2.1 mm ×100 mm), in combination with an Acquity
UPLC BEH Phenyl VanGuard pre-Column (130 Å, 1.7 µm,
Frontiers in Plant Science | www.frontiersin.org 3September 2016 | Volume 7 | Article 1377
fpls-07-01377 September 19, 2016 Time: 12:36 # 4
Arens et al. Presymtomatic Cercospora Detection in Sugar Beet
2.1 mm ×5 mm). The temperature of the column was maintained
at 35C. The solvent was administered at a flow rate of
500 µL/min, according to the following protocol: 0–1 min: 5%
solvent B, 1–10 min: 5–40% solvent B, 10–11.5 min: 40–97%
solvent B, 11.5–13 min: 97% solvent B, 13–13.5 min: 97–5%
solvent B; 13.5–17 min: 5% solvent B. Solvent B was acetonitrile
(Chemsolute, Renningen, Germany)/0.1% (v/v) formic acid (J.
T. Baker, Deventer, Netherlands), and solvent A was 18 m
water (Merck, Darmstadt, Germany)/0.1% (v/v) formic acid. The
mass spectrometry data were acquired by using a Bruker maXis
Impact device (Bremen, Germany) coupled to the LC system.
Electrospray ionization (ESI) was implemented in positive mode
at 200C dry temperature, three bar nebulizer, 4000 V capillary
voltage and a dry gas flow of 8 L/min. The MS settings were
adjusted to cope with small molecules (50–1,000 m/z), a hexapole
radio frequency (RF) voltage of 40 V peak-to-peak (Vpp), a
collision energy of 10 V, a funnel 1 RF of 300 Vpp, a funnel 2
RF of 300 Vpp, a prepulse storage time of 5 µs, a transfer time of
50 µs and a collision cell RF of 500 Vpp. The routine was run in
MS2scan modes: MS/MS (auto) and MRM.
Identification of Metabolites
Liquid chromatography–mass spectrometry data analysis was
performed using Compass Data Analysis 4.1, Compass Profile
Analysis 2.1 and various Compass utility tools offered by
Bruker. Data pre-processing for statistical analysis started with
an internal calibration for mass accuracy using ‘quadratic +
High Precision Calibration (HPC)’ mode, which was performed
for each injection with calibration solution containing 10 mM
sodium formate. After calibration, the peak finding algorithm
‘Find Molecular Features’ (FMFs) was applied to extract relevant
m/z signals. The FMF data were normalized aligned and used
to calculate a ‘bucket table’ which plots the mass retention
time pairs against the intensities of the individual samples.
Principal component analyses (PCA) was utilized to reduce
the multivariate data set to its leading features. The Pareto
scaling algorithm was applied with a confidence level of 95% to
calculate the PCAs. Samples were grouped according to their
treatment condition, and compared using the Student’s t-test
to identify differentially abundant metabolites. The selection
criteria were a FC of at least 1.5 and a FWER P-value of less
than 0.05 (Bonferroni-corrected). Sum formulae were calculated
by SmartFormula software, based on mass and isotope pattern
information. Stable isotope labeling with 15N was used to
confirm the identification of N-containing molecules. Following
MS2analysis, fragmentation patterns were queried against the
massbank database2and in silico verification was performed
using MetFrag3and MetFusion4(Gerlich and Neumann, 2013)
routines. When available, authentic standards were used to
confirm the annotation. In-source fragmentation products were
sorted manually on the basis of MS2data and elution time.
(U)HPLC-UV data were processed and monitored at 280 nm,
2www.massbank.jp
3http://msbi.ipb-halle.de/MetFrag/
4http://msbi.ipb-halle.de/MetFusion/
and the outputs analyzed using Compass Data Analysis 4.1 and
Microsoft Excel5software.
RESULTS
Comparison of Cercospora beticola
Infestation in Sugar Beet Genotypes
At 10 dpi the selected sugar beet genotypes showed contrasting
numbers of C. beticola induced necrotic lesions (Figure 1).
The susceptible genotype displayed numerous large necrotic
spots, while the size of lesions was lower in the tolerant
genotype. The resistant genotype showed no lesions. The
susceptibility to Cercospora leaf spot was quantified by qRT-
PCR. Comparison of absolute quantification of fungal biomass
to relative quantification (1ct is determined by subtraction of ct
values of an endogenous plant target from ct values from fungal
calmodulin) showed high correlation R2=0.97 (Supplementary
Figure S3). Therefore, both methods were equally suitable for
quantification. Figure 2 shows the mean values for relative
quantification obtained from three independent experiments
including each six biological replicates per condition (n=6 per
experiment). The higher the 1ct value, the less fungal biomass
was detected compared to plant biomass (De Coninck et al.,
2012). Fungal biomass was significantly different between all
genotypes, whereby leaves of the susceptible genotype contained
the highest amount of fungal biomass.
Cultivar-Specific Differences in the
Hyperspectral Signatures and Metabolite
Profiles
The linear discriminant analysis (LDA) of the hyperspectral
reflectance signature was not capable of discriminating between
the host cultivars when a single model approach was applied
(Figure 3A). A better level of discrimination, which is also
reflected in the higher prediction accuracy, was achieved by
applying the multi model and visualizing its first layer in an LDA
plot (Figure 3B). The reflectance signature of the cultivars varied
mainly between 1,600–1,850 and 2,150–2,400 nm (Figure 3C),
and there was a high level of dissimilarity between the signatures
of the susceptible and the resistant cultivar. The LC–MS analysis
of the semi-polar metabolites revealed a cultivar-specific pattern
in non-inoculated plants (Figure 4). PCA were calculated using
the intensities of mass retention time pairs (from here on
named ‘features’). The PCAs generated a distinct cultivar-specific
clustering, in which the first principal component accounted
for 52.7% of the explained variance and the second component
for 25.5%. The (U)HPLC-UV-chromatograms revealed a higher
level of similarity between the tolerant and the susceptible
cultivar than between either of them and the resistant one
(Figure 5A). The most prominent difference between the three
genotypes was the peak associated with an m/z of 595.17,
identified by MS/MS2analysis as glucosylvitexin (Supplementary
Figure S4), a compound known to be present in the leaves of
5https://products.office.com/en-gb/excel
Frontiers in Plant Science | www.frontiersin.org 4September 2016 | Volume 7 | Article 1377
fpls-07-01377 September 19, 2016 Time: 12:36 # 5
Arens et al. Presymtomatic Cercospora Detection in Sugar Beet
FIGURE 1 | The leaves of non-infected and infected plants. Neither type
shows any visible disease symptoms when observed 4 days post-infection
(dpi). At 10 dpi leaves of the susceptible genotype showed numerous large
necrotic lesions. The tolerant genotype exhibits smaller lesions. No necrosis
was detected on leaves of the resistant genotype.
beet (Isayenkova et al., 2006). The intensity of this peak could
not be used for quantification as the concentration differences
between the cultivars were outside the dynamic range of the
mass spectrometry device. Instead, its UV signal was used for
quantification. The intensity differences were estimated at 15
fold between the resistant cultivar and the susceptible one, and
seven fold between the resistant cultivar and the tolerant one
(Figure 5B).
Hyperspectral Imaging Enables the
Detection of Presymptomatic
Cercospora Leaf Spot and Is Predictive
of Metabolite Status
At the early stage of infection [4 days post-inoculation
(dpi)], the hyperspectral signatures of the leaves differed only
slightly from those of the non-infected leaves (Figure 6, left
side). However, these subtle differences were still sufficient to
perform a robust classification utilizing the advanced concept
of multi-models. The outcomes of the single and multi-
model analyses are shown in Table 1. Mean accuracy values
and its standard deviation across cross-validations for the
test data are reported. The LDA of the first layer of the
multi-model approach depicted in Figure 6 (right hand side)
shows a clear separation between the non-infected and infected
FIGURE 2 | Comparative analysis of the fungal biomass in three
genotypes. Values for plant DNA (ct of SbEc1) were subtracted from values
for Cercospora beticola DNA (ct of Cercospora calmodulin) to determine a
relative value for fungal biomass (1ct). The lower the 1ct value the higher the
amount of fungal biomass. Mean values were calculated from three
independent experiments containing at least six biological replicates per
condition (n=18). ∗∗∗P<0.001.
plants for each of the three cultivars. No visible symptoms
of infection were apparent at 4 dpi, but the hyperspectral
signatures allowed an unequivocal classification of their infection
status.
The coefficients of determination (R2) for the set of selected
metabolites are shown in Table 2. Based on the reflection
spectrum, the neural network was able to predict the below-
peak areas with a high level of precision (R2values up to
0.94), demonstrating a systematic dependency between the
hyperspectral signature and the metabolite status of the plant.
The single model approach was moderately accurate, with the
rate of correct classification lying from 70 to 80% (Table 1).
Implementing the multi model raised this rate to >98.5%. The
resistant cultivar was classified with 99.9% accuracy.
Metabolite Profiles Respond Rapidly to
Pathogen Infection
Principal component analysis was employed to differentiate
between the metabolic profiles of control and infected plants.
The PCAs scores plots demonstrate that a clear discrimination
could be drawn between the metabolite profiles produced by
the non-infected and infected plants for each cultivar already
at just 4 dpi (Figure 7 left hand side). The higher the levels
of resistance, the more distinct were the profiles. For the
susceptible cultivar the first principal component (PC) accounted
for 25.5% of the variance and the second PC for 20.4%. The
equivalents for the contrast between non-infected and infected
in the resistant and the tolerant cultivar were, PC1 36.6%,
PC2 14.0% and PC1 26.8% PC2 21.9% explained variance,
respectively.
Frontiers in Plant Science | www.frontiersin.org 5September 2016 | Volume 7 | Article 1377
fpls-07-01377 September 19, 2016 Time: 12:36 # 6
Arens et al. Presymtomatic Cercospora Detection in Sugar Beet
FIGURE 3 | Distinguishing the three cultivars by hyperspectral imaging. (A) The LDA of reflectance spectra (single model), (B) the LDA of reflectance spectra
(first layer of multi-model classification neural network). (C) The mean reflectance spectra for each cultivar (plants not exposed to C. beticola).
The Identification of Disease-Responsive
Metabolites
Based on the chosen significance criteria [family wise error rate
(FWER) P0.05 and fold change (FC) | 1.5| ], inoculation with
the pathogen altered 105 features in the resistant cultivar, 84 in
the tolerant cultivar and 77 in the susceptible cultivar (Figure 7,
right hand side). The features showing the highest FC were
largely, but not exclusively, cultivar-specific. The tolerant and
the resistant cultivars exhibited the highest degree of similarity
(36 features), and the susceptible and resistant ones the lowest
(22 features; Supplementary Figure S5). Features associated with
an R2value 0.7 were considered to be highly correlated
to the hyperspectral signatures and were annotated (Table 2).
Accurate mass and isotopic pattern led in most cases to a
molecular formula for the compounds. A full list including in-
source products is provided in Supplementary Table S1. In the
compatible (susceptible) interaction, the abundance of citric acid
was decreased (FC =3.45) and that of two compounds with
the molecular formula C10H8O3was increased (1.63, 2.10).
In both the tolerant and resistant cultivar, the concentration of
pantothenic acid was boosted by the presence of the fungus,
while that of 12-hydroxyjasmonic acid 12-O-β-D-glucoside (12-
O-Glc-JA) was reduced (1.68, 1.97). In the infected leaf of
the tolerant (but not the resistant) cultivar, the level of DOPA
was significantly raised; while in the resistant (but not the
tolerant) cultivar, the level of 5-O-feruloylquinic acid was reduced
(3.34).
FIGURE 4 | Liquid chromatography–mass spectrometry (LC–MS)
based metabolite profiling of the three cultivars (plants not exposed to
C. beticola) as analyzed by PCA. The metabolite profiles of all genotypes
show high separation. Each symbol represents one analysis (two technical
replicates of 6–7 probes from individual plants).
DISCUSSION
In the present study, selected sugar beet genotypes differing
in their degrees of susceptibility to Cercospora leaf spot could
clearly be distinguished by metabolite profiling (invasively) and
Frontiers in Plant Science | www.frontiersin.org 6September 2016 | Volume 7 | Article 1377
fpls-07-01377 September 19, 2016 Time: 12:36 # 7
Arens et al. Presymtomatic Cercospora Detection in Sugar Beet
FIGURE 5 | UV profiling of leaf extracts. (A) Overlay of typical (U)HPLC-UV chromatograms obtained from healthy leaves. The profiles of the tolerant and
susceptible cultivar are highly similar. The most prominent peak (peak A) is most strongly represented in the resistant cultivar. (B) The quantification of
o-glucosylvitexin (corresponding to peak A), based on the UV-peak area. Each symbol represents one analysis (two technical replicates of 6–7 probes from individual
plants).
by hyperspectral imaging (non-invasively). Furthermore the
feasibility of non-invasive highly accurate detection of C. beticola
infection of sugar beet plants using hyperspectral imaging
was shown. As discussed in detail below, the nature of the
changes to the leaf metabolome induced presymptomatically by
C. beticola infection may be informative regarding the molecular
difference between a compatible and an incompatible host–
pathogen interaction. Finally, the experiments set out to identify
candidate metabolites associated with either constitutive or
induced resistance to Cercospora leaf spot.
The Hyperspectral Signatures
Differentiate Genotypes and Suggests
the Existence of Pre-formed Defense
Compounds
In the absence of the pathogen, the three cultivars differed
both with respect to their metabolome and their hyperspectral
signature, which shows constitutive differently abundant
compounds, some of which are likely to be involved in defense.
In particular, glucosylvitexin was highly abundant in the resistant
cultivar. The glycosylated form of vitexin may be more readily
stored in the cell than the active form and hydrolyzed as a
reaction to environmental cues. Vitexin has been implicated in
the resistance of cucumber against powdery mildew (McNally
et al., 2003), and has also been associated with the biotic stress
response in certain cereal species (Balmer et al., 2013). The
photosensitizing properties of cercosporin, the toxin responsible
for the pathogenicity of Cercospora spp. (Upchurch et al.,
1991;Daub and Ehrenshaft, 2000;Staerkel et al., 2013), lead
to the formation of toxic singlet oxygen molecules (1O2) and
superoxides (·O2) (Daub and Hangarter, 1983). Thus, the
ROS scavenging properties of vitexin could be advantageous
for preventing cercosporin induced cell damage (Shibano et al.,
2008). Constitutive defense offers several advantages to the host:
it serves as a means to circumvent reduction of defense responses
due to fungal effectors, and helps to limit the growth of the
fungus by avoiding the time lag involved between the initial
infection and the metabolic reprogramming required to mount
an induced defense response. On the negative side, in the absence
of pathogen pressure, it imposes a metabolic cost on the host,
which is reflected in a reduced yield potential.
Quantification of Cercospora Leaf Spot
Differences in susceptibility to Cercospora leaf spot were visually
detected and quantified by qPCR in three selected genotypes.
Necrotic lesions were observed on leaves of the susceptible
(numerous, large partly merging) and tolerant (smaller, separate)
Frontiers in Plant Science | www.frontiersin.org 7September 2016 | Volume 7 | Article 1377
fpls-07-01377 September 19, 2016 Time: 12:36 # 8
Arens et al. Presymtomatic Cercospora Detection in Sugar Beet
FIGURE 6 | Distinction between cultivars obtained by hyperspectral imaging of leaves sampled at a presymptomatic stage (4 dpi) of a Cercospora
leaf spot disease. Comparison of the normalized relative reflectance spectra between non-infected and infected plants, the profiles differ only marginally from one
another (A,C,E). The mapping of the spectral information onto one dimension resulting from a multi-model LDA depicts the distribution of pixels belonging to
“non-infected” or “infected”: the less the extent of overlap, the more well-separated are the groups (B,D,F).
TABLE 1 | The classification accuracy achieved by models derived from hyperspectral data collected from a presymptomatic sugar beet infected with
C. beticola.
Method Resistant Tolerant Susceptible
Mean accuracy Standard accuracy Mean accuracy Standard accuracy Mean accuracy Standard accuracy
Single model 78.4% 1.4% 74.9% 2.50% 69.4% 2.9%
Multi model 99.9% 0.1% 99.6% 0.2% 98.5% 0.2%
The mean accuracy and associated standard deviation obtained from a fivefold cross validation of infected vs. non-infected leaf were based on single pixel spectra.
(A 100% accuracy score is realized when each pixel is assigned to the correct class.) Two different methods have been used, a single model approach with one Neural
Network and a multi-model approach, in which an ensemble of classifiers is used to predict the respective class for each pixels. The rate of successful classification was
increased by using a multi-model approach.
genotype. Despite the detection of fungal biomass in all genotypes
there were no spots visible on leaves belonging to the resistant
genotype. Feindt et al. (1981) described equal sporulation
behavior and epidermal growth of C. beticola on susceptible and
resistant cultivars. Therefore, the fungal DNA detected in the
resistant genotype could be originated from initial epidermal
hyphal growth. Possibly, the lesion development and the collapse
of plant cells is delayed or inhibited by a strong defense response
Frontiers in Plant Science | www.frontiersin.org 8September 2016 | Volume 7 | Article 1377
fpls-07-01377 September 19, 2016 Time: 12:36 # 9
Arens et al. Presymtomatic Cercospora Detection in Sugar Beet
FIGURE 7 | Metabolite profiling of cultivars: leaves sampled at a presymptomatic stage of a Cercospora leaf spot disease. The LC–MS data-based PCA
reveals some clustering of metabolite profiles. Each symbol represents one analysis (two technical replicates of 6–7 probes from individual plants). The most
well-separated host was the resistant cultivar, suggesting that it experiences the most profound reprogramming as a result of the fungal infection. The volcano plots
depict significantly different features in all three cultivars in the contrast “non-infected” vs. “infected” (light gray). The thresholds were indicated with a red dashed line
for the corrected (FWER; P0.05), and a black dashed line for FC 1.5.
in the resistant genotype. When plant cells are intact the nutrient
availability in the intercellular space is poor, which might lead
to termination of hyphal growth due to starvation. With this
qPCR method presence but not viability of fungal DNA could be
assessed.
Hyperspectral Imaging Enables the Early
Diagnosis of Cercospora Leaf Spot
Infection
The hyperspectral imaging technique has demonstrated its ability
to diagnose Cercospora leaf spot disease presymptomatically
with the highest reported classification accuracy (98.5–99.9%) by
hyperspectral imaging (Rumpf et al., 2010;Mahlein et al., 2012),
representing thereby a significant advance toward the automation
of early generation screening in a resistance breeding program.
It is apparent that hyperspectral signatures are genotypically
variable, so an important validation step will be to trial the
method on a segregating population.
In other contexts, hyperspectral imaging has been
demonstrated to be suitable for the estimation of metabolites,
for instance the aflatoxin concentration on corn kernels was
successfully associated with hyperspectral imaging data (Yao
et al., 2010). In addition, predictive models based on reflectance
Frontiers in Plant Science | www.frontiersin.org 9September 2016 | Volume 7 | Article 1377
fpls-07-01377 September 19, 2016 Time: 12:36 # 10
Arens et al. Presymtomatic Cercospora Detection in Sugar Beet
TABLE 2 | Differentially abundant metabolites in young sugar beet leaves sampled 4 days after infection with C. beticola.
RT(min): observed
mass (m/z)
P<0.05
(FWER)
Genotype Fold change
‘control/infected’
Molecular
formula (M)
Adduct MS/MS fragments
(m/z)
Metabolite name 15N labeling Hyperspectral
R2
3.67 : 177.055 3.5E-03 Susceptible 1.63 C10H8O3M+H eV 35 (89.04; 134.04; N00.80
17.04; 105.03)
7.49 : 177.055 2.5E-04 Susceptible 2.10 C10H8O3M+H eV 35 (134.04; 89.04) N00.71
0.77 : 215.016 4.6E-03 Susceptible 3.45 C6H8O7M+Na Citric acidSN00.74
0.63 : 198.076 3.8E-04 Tolerant 3.60 C9H11NO4M+H eV 35 (107.05; 123.04; L-DOPASN10.80
135.04; 152.07)
1.32 : 220.118 4.3E-06 Tolerant 1.78 C9H17NO5M+H eV 10 (202.11; 184.1) Pantothenic acidSN10.91
0.76 : 198.076 1.4E-03 Tolerant 2.99 C9H12NO4M+H eV 35 (107.05; 123.04; DOPADN10.94
135.04; 152.07)
0.78 : 215.016 3.7E-05 Tolerant 3.25 C6H8O7M+Na Citric acidSN00.87
0.67 : 193.035 5.5E-06 Tolerant 2.94 C6H8O7M+H Isocitric acid SN00.71
3.51 : 163.075 1.5E-05 Tolerant 1.55 C10H10 O2M+H eV 15 (103.05; 131.05) N00.89
4.03 : 411.162 1.5E-05 Tolerant 1.68 C18H28 O9M+Na eV 10 (227.13; 12-Hydroxyjasmonic acid N00.90
(249 M+Na) 12-O-beta-D-glucosideD
3.14 : 177.055 8.0E-05 Resistant 1.76 C10H8O3M+H eV 35 (89.039; N00.76
134.036)
3.65 : 177.055 4.4E-03 Resistant 1.67 C10H8O3M+H eV 35 (89.04; 134.04; N00.76
117.04; 105.03)
1.32 : 220.118 8.8E-05 Resistant 2.05 C9H17NO5M+H eV 10 (202.11; 184.1) Pantothenic acidSN10.88
4.03 : 411.162 9.0E-04 Resistant 1.97 C18 H28O9M+Na eV 10( 227.13; 12-Hydroxyjasmonic N00.79
(249 M+Na) acid 12-O-beta-D-glucosideD
7.58 : 369.119 3.1E-03 Resistant 3.34 C17 H20O9M+H eV 35 (207.07; 175.04; 5-O-Feruloylquinic acidDN00.76
147.04; 119.05; 91.06
The metabolites were selected on the basis of their showing a high correlation (R20.7) with the hyperspectral data. Stable isotope labeling with 15N was used to confirm the molecular formulae of nitrogenous
compounds. Annotation was based on comparison with authentic reference standards (level 1) or databases (level 2) (Sumner et al., 2007). More details are given in Supplementary Table S1. SIdentified based on
authentic reference standard; Dannotation based on database search.
Frontiers in Plant Science | www.frontiersin.org 10 September 2016 | Volume 7 | Article 1377
fpls-07-01377 September 19, 2016 Time: 12:36 # 11
Arens et al. Presymtomatic Cercospora Detection in Sugar Beet
signatures could predict cardenolide concentration in milkweed
upon wounding (Couture et al., 2013). On a wider level, the
technology has been applied to assess foliar polyphenol and
nitrogen content across entire landscapes (Skidmore et al., 2010).
Presymptomatic Metabolic Defense
Response Was Detected
This is the first study performing metabolite profiling on the early
response reaction of sugar beet cultivars with differing degree of
susceptibility. Some candidate metabolites correlated positively
with hyperspectral signatures and possibly contribute to the
spectral reflectance. These more relevant candidates were further
annotated. A major challenge in the field of metabolomics is
the annotation of compounds of secondary metabolism because
authentic standards are not always available and there is limited
information in databases concerning MS and MS2spectra. In-
source fragmentation is another major issue which can lead to
fragments that are mistaken as cellular metabolites (Xu et al.,
2015). Time consuming manual data inspection is often required,
thus filtering candidates for relevance is helpful.
Metabolic Response to Cercospora Leaf
Spot during Incompatible Interaction
The metabolomics data has provided novel insights to
incompatible plant–pathogen interaction and is being
preliminary discussed here but functional analysis has to
follow regardless. The metabolites were identified with authentic
reference substances, if available, or annotated with MetFusion
based on exact mass and fragmentation pattern. The leaf
content of pantothenic acid was markedly increased in both
the tolerant and the resistant cultivar following their infection
by C. beticola. This compound, also referred to as vitamin
B5, functions as a precursor of coenzyme A (CoA; Smith
et al., 2007), which is involved in a wide range of biological
processes, including the TCA cycle and both fatty acid and
phenylpropanoid metabolism. A scan of the literature suggests
that this is the first documented instance of pantothenic
acid being associated with the biotic stress response. Many
of the (pro-) vitamins (A, B1, B6, B9, C, E, K1) have been
associated with antioxidative potential (Asensi-Fabado and
Munné-Bosch, 2010), and the structure of pantothenic acid
includes three free hydroxyl groups with potential antioxidative
activity; thus it may be that by accumulating pantothenic
acid, the sugar beet plant gives itself a measure of protection
against cercosporin-induced oxidative stress. Vitamin B6 has
demonstrated some capacity within Cercospora to inhibit
cercosporin autotoxicity (Bilski et al., 2008), and is known as a
protectant against photo-oxidative stress (Chen and Xiong, 2005;
Titiz et al., 2006;Havaux et al., 2009). Further experiments are
of need to elucidate the role of vitamin B5 in stress response
whether it is playing an active part or it is an accumulating
intermediate.
The molecule 12-O-Glc-JA, initially termed tuberonic acid
glucoside (Yoshihara et al., 1989), was less abundant in the
infected leaf of both the tolerant and the resistant cultivar than
in that of the susceptible one.
Glycosylation is discussed to be a modification of the
bioactive aglycon that enables transport and/or storage (Jones
and Vogt, 2001). Plant hormones are also subjected to it,
for instance increased glycosylation was shown to affect
abscisic acid (ABA) homeostasis (Priest et al., 2006). As a
consequence to environmental stimulus enzymatic hydrolysis
of storage compounds releases the physiologically active
form. The observed reduction in 12-O-Glc-JA content upon
infection suggests its conversion by deglycosylation. In rice,
an enzyme (OsTAGG1) has been purified which is capable
of deglycosylating 12-O-Glc-JA to form the physiologically
active form 12-hydroxyjasmonic acid (12-OH-JA; Wakuta
et al., 2010). 12-O-Glc-JA is reportedly synthesized from
the phytohormone jasmonic acid (JA), which has variety of
functions in biotic stress defense; for example, it regulates
the production of phenylpropanoids (lignins, flavonoids, and
other antioxidants; Gundlach et al., 1992). Storage forms might
provide a safety net for plants when de novo synthesis is
inhibited due to fungal effector molecules disturbing pathogen
defense. 12-O-Glc-JAs can be transported to underground
parts and act as a signaling molecule that may mediate
changes of source-sink relationship upon pathogen attack
(Yoshihara et al., 1996;Seto et al., 2009). JAs have various
functions in defense and development whereby defense is
prioritized over growth, for instance root growth inhibition
(Yang et al., 2012;Wasternack and Hause, 2013). Activation of
JA signaling cascade in the tolerant and resistant sugar beet
genotypes suggest a more efficient recognition of C. beticola
by MAMPs/PAMPs and subsequent triggering of defense
responses.
The leaf content of 5-O-feruloylquinic acid was lowered in
the resistant cultivar upon infection. This compound belongs to
the chlorogenic acids (Clifford, 1999), which act as important
intermediates in lignin synthesis (Vanholme et al., 2010). So a
reduced level of 5-O-feruloylquinic acid implies an increased
rate of lignin synthesis. Lignin is of prime importance for the
physical strength of the cell wall; in cotton, it has been shown that
resistance to the pathogen Verticilium dahliae is associated with
an increased level of cell wall lignification (Xu et al., 2011).
The tolerant sugar beet cultivar responded to the presence of
C. beticola by accumulating the molecule L-DOPA, a precursor
of dopamine, thought to play a role in the host’s resistance to
Cercospora leaf spot (Harrison et al., 1970). In addition, L-DOPA
and dopamine are both strong antioxidants, so could help in
the scavenging of the reactive oxygen species induced by the
action of cercosporin (Kanazawa and Sakakibara, 2000;Gülçin,
2007).
In summary, genotypes could be distinguished based on
their hyperspectral signature and on their metabolic profiles.
Whether metabolites contribute to the hyperspectral signature
or act as separate markers is unclear. Further experiments
to investigate the contribution of phenolic compounds to the
hyperspectral signature in sugar beet leaves subjected to modified
light conditions are underway.
The study represents a successful proof-of-concept for an
effective and efficient screening system for presymptomatic
identification of Cercospora leaf spot. Compared to the
Frontiers in Plant Science | www.frontiersin.org 11 September 2016 | Volume 7 | Article 1377
fpls-07-01377 September 19, 2016 Time: 12:36 # 12
Arens et al. Presymtomatic Cercospora Detection in Sugar Beet
widely used visual disease assessment, quantification of disease
severity by hyperspectral imaging provides the advantage of
standardized and objective measurement. Further, phenotyping
by hyperspectral imaging is amenable to automation, and high-
throughput analysis is possible after the initial establishment
is realized. As a result, the need for greenhouse space, labor
and time can be reduced. Hand-held devices capable of
hyperspectral phenotyping under field conditions are currently
under development and could potentially improve effective
fungicide application. The characterization of the metabolomes
of the contrasting cultivars in response to C. beticola infection
has provided interesting new candidates for the components of
the defense response. The biological role of these compounds
in the context of the host–pathogen interaction remains to be
characterized; however, in the meantime, they can be used as
informative candidates associated with resistance in beet.
AUTHOR CONTRIBUTIONS
Contributions to conception and design of this study: NA, AB,
SF, US, and H-PM; Participation in drafting and revising of the
manuscript: NA, AB, SD, SF, US, and H-PM; Experiments and
data acquisition: plant cultivation and pathogen inoculation: SF;
Hyperspectral imaging: AB; Metabolomics study: NA.
FUNDING
This study was financially supported by a grant (FKZ 22010512)
from FNR (Fachagentur Nachwachsende Rohstoffe e.V.).
ACKNOWLEDGMENT
We thank Katrin Harbordt, Strube Research (Schlanstedt) for
providing the sugar beet cultivars, and Martin Lietz and Felix
Rose for their technical assistance.
SUPPLEMENTARY MATERIAL
The Supplementary Material for this article can be found online
at: http://journal.frontiersin.org/article/10.3389/fpls.2016.01377
REFERENCES
Asensi-Fabado, M. A., and Munné-Bosch, S. (2010). Vitamins in plants:
occurrence, biosynthesis and antioxidant function. Trends Plant Sci. 15, 582–
592. doi: 10.1016/j.tplants.2010.07.003
Asner, G. P., Knapp, D. E., Kennedy-Bowdoin, T., Jones, M. O., Martin, R. E.,
Boardman, J., et al. (2007). Carnegie airborne observatory: in-flight fusion
of hyperspectral imaging and waveform light detection and ranging for
three-dimensional studies of ecosystems. J. Appl. Remote Sens. 1:013536. doi:
10.1117/1.2794018
Backhaus, A., Bollenbeck, F., and Seiffert, U. (2011)). “High-throughput quality
control of coffee varieties and blends by artificial neural networks and
hyperspectral imaging,” in Proceedings of the 1st International Congress on
Cocoa, Coffee and Tea, CoCoTea, Novara.
Backhaus, A., Lachmair, J., Rückert, U., and Seiffert, U. (2012). “Hardware
accelerated real time classification of hyperspectral imaging data for coffee
sorting,” in Proceedings of the European Symposium on Artificial Neural
Networks, Computational Intelligence and Machine Learning, Bruges.
Balmer, D., Flors, V., Glauser, G., and Mauch-Mani, B. (2013). Metabolomics of
cereals under biotic stress: current knowledge and techniques. F. Plant Sci. 4:82.
doi: 10.3389/fpls.2013.00082
Bilski, P., Obryk, B., Olko, P., Mandowska, E., Mandowski, A., and Kim, J. (2008).
Characteristics of LiF: Mg, Cu, P thermoluminescence at ultra-high dose range.
Radiat. Meas. 43, 315–318. doi: 10.1016/j.radmeas.2007.10.015
Blackburn, G. A. (1998). Quantifying chlorophylls and caroteniods at leaf
and canopy scales: an evaluation of some hyperspectral approaches.
Remote Sens. Environ. 66, 273–285. doi: 10.1016/S0034-4257(98)
00059-5
Blackburn, G. A. (2007). Hyperspectral remote sensing of plant pigments. J. Exp.
Bot. 58, 855–867. doi: 10.1093/jxb/erl123
Bolton, M. D., Rivera, V., and Secor, G. (2013). Identification of the G143A
mutation associated with QoI resistance in Cercospora beticola field isolates
from Michigan, United States. Pest. Manag. Sci. 69, 35–39. doi: 10.1002/
ps.3358
Carvalho, S., Macel, M., Schlerf, M., Skidmore, A. K., and Putten, W. H. (2012). Soil
biotic impact on plant species shoot chemistry and hyperspectral reflectance
patterns. New Phytol. 196, 1133–1144. doi: 10.1111/j.1469-8137.2012.
04338.x
Chen, H., and Xiong, L. (2005). Pyridoxine is required for post-embryonic root
development and tolerance to osmotic and oxidative stresses. Plant J. 44,
396–408. doi: 10.1111/j.1365-313X.2005.02538.x
Clifford, M. N. (1999). Chlorogenic acids and other cinnamates–nature, occurrence
and dietary burden. J. Sci. Food Agric. 79, 362–372. doi: 10.1002/(SICI)1097-
0010(19990301)79:3<362::AID-JSFA256>3.0.CO;2-D
Couture, J. J., Serbin, S. P., and Townsend, P. A. (2013). Spectroscopic sensitivity of
real-time, rapidly induced phytochemical change in response to damage. New
Phytol. 198, 311–319. doi: 10.1111/nph.12159
Daub, M. E., and Ehrenshaft, M. (2000). The photoactivated Cercospora
toxin cercosporin: contributions to plant disease and fundamental
biology. Annu. Rev. Phytopathol. 38, 461–490. doi: 10.1146/annurev.phyto.
38.1.461
Daub, M. E., and Hangarter, R. P. (1983). Light-Induced Production of singlet
oxygen and superoxide by the fungal toxin. Cercosporin. Plant Physiol. 73,
855–857. doi: 10.1104/pp.73.3.855
De Coninck, B., Amand, O., Delauré, S., Lucas, S., Hias, N., Weyens, G.,
et al. (2012). The use of digital image analysis and real-time PCR
fine-tunes bioassays for quantification of Cercospora leaf spot disease in
sugar beet breeding. Plant pathol. 61, 76–84. doi: 10.1111/j.1365-3059.2011.
02497.x
Feindt, F., Mendgen, K., and Heitefuß, R. (1981). Der Einfluss der
Spaltöffnungsweite und des Blattalters auf den Infektionserfolg von Cercospora
beticola bei Zuckerrüben (Beta vulgaris L.) Unterschiedlicher Anfälligkeit.
Konstanz: Bibliothek der Universität Konstanz.
Fiehn, O. (2002). Metabolomics - the link between genotypes and
phenotypes. Plant Mol. Biol. 48, 155–171. doi: 10.1023/A:1013713
905833
Fiorani, F., and Schurr, U. (2013). Future scenarios for plant phenotyping.
Annu. Rev. Plant Biol. 64, 267–291. doi: 10.1146/annurev-arplant-050312-
120137
Gerlich, M., and Neumann, S. (2013). MetFusion: integration of compound
identification strategies. J. Mass Spectrom. 48, 291–298. doi: 10.1002/
jms.3123
Gowen, A., O’Donnell, C., Cullen, P., Downey, G., and Frias, J. (2007).
Hyperspectral imaging–an emerging process analytical tool for food
quality and safety control. Trends Food Sci. Technol. 18, 590–598. doi:
10.1016/j.tifs.2007.06.001
Gülçin, I. (2007). Comparison of in vitro antioxidant and antiradical activities of
L-tyrosine and L-Dopa. Amino acids 32, 431–438. doi: 10.1007/s00726-006-
0379-x
Gundlach, H., Müller, M. J., Kutchan, T. M., and Zenk, M. H. (1992). Jasmonic acid
is a signal transducer in elicitor-induced plant cell cultures. Proc. Natl. Acad. Sci.
U.S.A. 89, 2389–2393. doi: 10.1073/pnas.89.6.2389
Frontiers in Plant Science | www.frontiersin.org 12 September 2016 | Volume 7 | Article 1377
fpls-07-01377 September 19, 2016 Time: 12:36 # 13
Arens et al. Presymtomatic Cercospora Detection in Sugar Beet
Harrison, M., Maag, G., Hecker, R., and Payne, M. (1970). Some speculations on the
role of dopamine in the resistance of sugarbeets to Cercospora leaf spot. J. Am.
Soc. Sugar Beet Technol. 16, 34–40.
Havaux, M., Ksas, B., Szewczyk, A., Rumeau, D., Franck, F., Caffarri, S., et al.
(2009). Vitamin B6 deficient plants display increased sensitivity to high light
and photo-oxidative stress. BMC Plant Biol. 9:130. doi: 10.1186/1471-2229-
9-130
Hoagland, D. R., and Arnon, D. I. (1950). The water-culture method for growing
plants without soil. Circ. Calif. Agric. Exp. Stn. 347, 1–32.
Isayenkova, J., Wray, V., Nimtz, M., Strack, D., and Vogt, T. (2006).
Cloning and functional characterisation of two regioselective flavonoid
glucosyltransferases from Beta vulgaris.Phytochemistry 67, 1598–1612. doi:
10.1016/j.phytochem.2006.06.026
Jones, P., and Vogt, T. (2001). Glycosyltransferases in secondary plant
metabolism: tranquilizers and stimulant controllers. Planta 213, 164–174. doi:
10.1007/s004250000492
Kanazawa, K., and Sakakibara, H. (2000). High content of dopamine, a strong
antioxidant, in cavendish banana. J. Agric. Food Chem. 48, 844–848. doi:
10.1021/jf9909860
Knauer, U., Backhaus, A., and Seiffert, U. (2014). Fusion trees for fast and accurate
classification of hyperspectral data with ensembles of Gamma-divergence-
based RBF networks. Neural Comput. Appl. 25, 1–10.
Koch, G., Jung, C., Asher, M., Holtschulte, B., Molard, M., Rosso, F., et al. (2000).
“Genetic localization of Cercospora resistance genes,” in Cercospora Beticola
Sacc. Biology, Agronomic Influence and Control Measures in Sugar Beet, eds
M. J. C. Asher, B. Holtschulte, M. M. Richard, F. Rosso, G. Steinruecken, and
R. Beckers (Bruxelles: IIRB-Eigenverlag), 197–209.
Lelong, C. C., Pinet, P. C., and Poilvé, H. (1998). Hyperspectral imaging
and stress mapping in agriculture: a case study on wheat in Beauce
(France). Remote Sens. Environ. 66, 179–191. doi: 10.1016/S0034-4257(98)
00049-2
Lewellen, R., and Whitney, E. (1976). Inheritance of resistance to race
C2 of Cercospora beticola in sugarbeet. Crop Sci. 16, 558–561. doi:
10.2135/cropsci1976.0011183X001600040032x
Mahlein, A.-K., Steiner, U., Dehne, H.-W., and Oerke, E.-C. (2010). Spectral
signatures of sugar beet leaves for the detection and differentiation of diseases.
Precis. Agric. 11, 413–431. doi: 10.1186/1746-4811-8-3
Mahlein, A.-K., Steiner, U., Hillnhütter, C., Dehne, H.-W., and Oerke, E.-C. (2012).
Hyperspectral imaging for small-scale analysis of symptoms caused by different
sugar beet diseases. Plant Methods 8:3. doi: 10.1186/1746-4811-8-3
Martinetz, T., and Schulten, K. (1991). A“Neural-Gas” Network Learns Topologies.
Champaign, IL: University of Illinois at Urbana-Champaign.
McNally, D. J., Wurms, K. V., Labbé, C., Quideau, S., and Bélanger, R. R. (2003).
Complex C-Glycosyl flavonoid phytoalexins from Cucumis s ativus. J. Nat. Prod.
66, 1280–1283. doi: 10.1021/np030150y
Meier, U., Bachmann, L., Buhtz, H., Hack, H., Klose, R., Märländer, B., et al.
(1993). Phänologische Entwick-lungsstadien der Beta-Rüben (Beta vulgaris L.
ssp.). Codierung und beschreibung nach der erweiterten BBCH-Skala (mit
abbildungen). Nachrichtenbl. Deut. Pflanzenschutzd 45, 37–41.
Moody, J., and Darken, C. J. (1989). Fast learning in networks of locally-tuned
processing units. Neural Comput. 1, 281–294. doi: 10.1162/neco.1989.1.2.281
Mutka, A. M., and Bart, R. S. (2014). Image-based phenotyping of plant disease
symptoms. Front. Plant Sci. 5:734. doi: 10.3389/fpls.2014.00734
Priest, D. M., Ambrose, S. J., Vaistij, F. E., Elias, L., Higgins, G. S., Ross, A. R.,
et al. (2006). Use of the glucosyltransferase UGT71B6 to disturb abscisic acid
homeostasis in Arabidopsis thaliana.Plant J. 46, 492–502. doi: 10.1111/j.1365-
313X.2006.02701.x
Rumpf, T., Mahlein, A.-K., Steiner, U., Oerke, E.-C., Dehne, H.-W., and Plümer, L.
(2010). Early detection and classification of plant diseases with support vector
machines based on hyperspectral reflectance. Comput. Electron. Agric. 74,
91–99. doi: 10.1016/j.compag.2010.06.009
Schmidt, K., Heberle, B., Kurrasch, J., Nehls, R., and Stahl, D. J. (2004). Suppression
of phenylalanine ammonia lyase expression in sugar beet by the fungal pathogen
Cercospora beticola is mediated at the core promoter of the gene. Plant Mol. Biol.
55, 835–852. doi: 10.1007/s11103-005-2141-2
Schmidt, K., Pflugmacher, M., Klages, S., Maeser, A., Mock, A., and Stahl, D. J.
(2008). Accumulation of the hormone abscisic acid (ABA) at the infection site
of the fungus Cercospora beticola supports the role of ABA as a repressor of
plant defence in sugar beet. Mol. Plant Pathol. 9, 661–673. doi: 10.1111/j.1364-
3703.2008.00491.x
Seto, Y., Hamada, S., Matsuura, H., Matsushige, M., Satou, C., Takahashi, K.,
et al. (2009). Purification and cDNA cloning of a wound inducible
glucosyltransferase active toward 12-hydroxy jasmonic acid. Phytochemistry 70,
370–379. doi: 10.1016/j.phytochem.2009.01.004
Shibano, M., Kakutani, K., Taniguchi, M., Yasuda, M., and Baba, K. (2008).
Antioxidant constituents in the dayflower (Commelina communis L.) and their
α-glucosidase-inhibitory activity. J. Nat. Med. 62, 349–353. doi: 10.1007/s11418-
008-0244-1
Skidmore, A. K., Ferwerda, J. G., Mutanga, O., Van Wieren, S. E., Peel, M.,
Grant, R. C., et al. (2010). Forage quality of savannas—simultaneously mapping
foliar protein and polyphenols for trees and grass using hyperspectral imagery.
Remote Sens. Environ. 114, 64–72. doi: 10.1016/j.rse.2009.08.010
Smith, A. G., Croft, M. T., Moulin, M., and Webb, M. E. (2007). Plants need their
vitamins too. Curr. Opin. Plant Biol. 10, 266–275. doi: 10.1016/j.pbi.2007.04.009
Staerkel, C., Boenisch, M. J., Kroger, C., Bormann, J., Schafer, W., and Stahl, D.
(2013). CbCTB2, an O-methyltransferase is essential for biosynthesis of the
phytotoxin cercosporin and infection of sugar beet by Cercospora beticola.BMC
Plant Biol. 13:50. doi: 10.1186/1471-2229-13-50
Sumner, L. W., Amberg, A., Barrett, D., Beale, M. H., Beger, R., Daykin, C. A.,
et al. (2007). Proposed minimum reporting standards for chemical analysis.
Metabolomics 3, 211–221. doi: 10.1007/s11306-007-0082-2
Titiz, O., Tambasco-Studart, M., Warzych, E., Apel, K., Amrhein, N., Laloi, C.,
et al. (2006). PDX1 is essential for vitamin B6 biosynthesis, development
and stress tolerance in Arabidopsis.Plant J. 48, 933–946. doi: 10.1111/j.1365-
313X.2006.02928.x
Upchurch, R., Walker,D., Rollins, J., Ehrenshaft, M., and Daub, M. (1991). Mutants
of Cercospora kikuchii altered in cercosporin synthesis and pathogenicity. Appl.
Environ. Microbiol. 57, 2940–2945.
Vanholme, R., Demedts, B., Morreel, K., Ralph, J., and Boerjan, W. (2010).
Lignin biosynthesis and structure. Plant Physiol. 153, 895–905. doi:
10.1104/pp.110.155119
Viant, M. R., and Sommer, U. (2013). Mass spectrometry based environmental
metabolomics: a primer and review. Metabolomics 9, 144–158. doi:
10.1007/s11306-012-0412-x
Wakuta, S., Hamada, S., Ito, H., Matsuura, H., Nabeta, K., and Matsui, H.
(2010). Identification of a β-glucosidase hydrolyzing tuberonic acid
glucoside in rice (Oryza sativa L.). Phytochemistry 71, 1280–1288. doi:
10.1016/j.phytochem.2010.04.025
Wasternack, C., and Hause, B. (2013). Jasmonates: biosynthesis, perception, signal
transduction and action in plant stress response, growth and development. An
update to the 2007 review in annals of botany. Ann. Bot. 111, 1021–1058. doi:
10.1093/aob/mct067
Whitney, E., and Lewellen, R. (1976). Identification and distribution of races C1
and C2 of Cercospora beticola from sugarbeet. Phytopathology 66, 1158–1160.
doi: 10.1094/Phyto-66-1158
Xu, L., Zhu, L., Tu, L., Liu, L., Yuan, D., Jin, L., et al. (2011). Lignin metabolism has
a central role in the resistance of cotton to the wilt fungus Verticillium dahliae
as revealed by RNA-Seq-dependent transcriptional analysis and histochemistry.
J. Exp. Bot. 62, 5607–5621. doi: 10.1093/jxb/err245
Xu, Y.-F., Lu, W., and Rabinowitz, J. D. (2015). Avoiding Misannotation
of in-source fragmentation products as cellular metabolites in liquid
chromatography–mass spectrometry-based metabolomics. Anal. Chem. 87,
2273–2281. doi: 10.1021/ac504118y
Yang, D.-H., Hettenhausen, C., Baldwin, I. T., and Wu, J. (2012). Silencing
Nicotiana attenuata calcium-dependent protein kinases, CDPK4 and
CDPK5, strongly up-regulates wound-and herbivory-induced jasmonic
acid accumulations. Plant Physiol. 159, 1591–1607. doi: 10.1104/pp.112.
199018
Yao, H., Hruska, Z., Kincaid, R., Brown, R., Cleveland, T., and Bhatnagar, D. (2010).
Correlation and classification of single kernel fluorescence hyperspectral data
with aflatoxin concentration in corn kernels inoculated with Aspergillus flavus
spores. Food Addit. Contam. 27, 701–709. doi: 10.1080/19440040903527368
Yoshihara, T., Amanuma, M., Tsutsumi, T., Okumura, Y., Matsuura, H.,
and Ichihara, A. (1996). Metabolism and transport of [2-14C]( ±)
jasmonic acid in the potato plant. Plant Cell Physiol. 37, 586–590. doi:
10.1093/oxfordjournals.pcp.a028985
Frontiers in Plant Science | www.frontiersin.org 13 September 2016 | Volume 7 | Article 1377
fpls-07-01377 September 19, 2016 Time: 12:36 # 14
Arens et al. Presymtomatic Cercospora Detection in Sugar Beet
Yoshihara, T., Omir, E.-S. A., Koshino, H., Sakamura, S., Kkuta, Y., and Koda, Y.
(1989). Structure of a tuber-inducing stimulus from potato leaves (Solanum
tuberosum L.). Agric. Biol. Chem. 53, 2835–2837. doi: 10.1271/bbb1961.53.2835
Conflict of Interest Statement: The authors declare that the research was
conducted in the absence of any commercial or financial relationships that could
be construed as a potential conflict of interest.
Copyright © 2016 Arens, Backhaus, Döll, Fischer, Seiffert and Mock. This is an
open-access article distributed under the terms of the Creative Commons Attribution
License (CC BY). The use, distribution or reproduction in other forums is permitted,
provided the original author(s) or licensor are credited and that the original
publication in this journal is cited, in accordance with accepted academic practice.
No use, distribution or reproduction is permitted which does not comply with these
terms.
Frontiers in Plant Science | www.frontiersin.org 14 September 2016 | Volume 7 | Article 1377
... Besides Trp, two compounds, C4 and Fer, were discovered to have increased levels for well storable varieties after 13 weeks of storage. There are indications that phenylpropanoids influence biotic and abiotic stress resistance in sugar beet, e.g. against Cercospora beticola or during salt stress [55,56]. Hence, an involvement during sugar beet storage may be possible. ...
Article
Full-text available
Background Sugar beet is an important crop for sugar production. Sugar beet roots are stored up to several weeks post-harvest waiting for processing in the sugar factories. During this time, sucrose loss and invert sugar accumulation decreases the final yield and processing quality. To improve storability, more information about post-harvest metabolism is required. We investigated primary and secondary metabolites of six sugar beet varieties during storage. Based on their variety-specific sucrose loss, three storage classes representing well, moderate, and bad storability were compared. Furthermore, metabolic data were visualized together with transcriptome data to identify potential mechanisms involved in the storage process. Results We found that sugar beet varieties that performed well during storage have higher pools of 15 free amino acids which were already observable at harvest. This storage class-specific feature is visible at harvest as well as after 13 weeks of storage. The profile of most of the detected organic acids and semi-polar metabolites changed during storage. Only pyroglutamic acid and two semi-polar metabolites, including ferulic acid, show higher levels in well storable varieties before and/or after 13 weeks of storage. The combinatorial OMICs approach revealed that well storable varieties had increased downregulation of genes involved in amino acid degradation before and after 13 weeks of storage. Furthermore, we found that most of the differentially genes involved in protein degradation were downregulated in well storable varieties at both timepoints, before and after 13 weeks of storage. Conclusions Our results indicate that increased levels of 15 free amino acids, pyroglutamic acid and two semi-polar compounds, including ferulic acid, were associated with a better storability of sugar beet taproots. Predictive metabolic patterns were already apparent at harvest. With respect to elongated storage, we highlighted the role of free amino acids in the taproot. Using complementary transcriptomic data, we could identify potential underlying mechanisms of sugar beet storability. These include the downregulation of genes for amino acid degradation and metabolism as well as a suppressed proteolysis in the well storable varieties.
... In agricultural plant-breeding systems, this strategy has proved beneficial in enhancing efficacy and precision (Ge et al. 2019;Meacham-Hensold et al. 2019). Therefore, the spectral identification of foliar functional characteristics enables in identifying, tracing, and simulating physiological and biochemical pathogenic processes that enable the employment of spectroscopy to determine plant-fungal interactions (Arens et al. 2016;Couture et al. 2018;Fallon et al. 2019;Meacham-Hensold et al. 2020;Gold et al. 2019Gold et al. , 2020. Since the 1980s, hyperspectral and wideband approaches based on visible and near-infrared reflectance indicators, such as the normalized difference index, have been employed to detect late-stage photo disease (Jackson 1986;Hatfield and Pinter 1993;Nilsson 1995). ...
... In agricultural plant-breeding systems, this strategy has proved beneficial in enhancing efficacy and precision (Ge et al. 2019;Meacham-Hensold et al. 2019). Therefore, the spectral identification of foliar functional characteristics enables in identifying, tracing, and simulating physiological and biochemical pathogenic processes that enable the employment of spectroscopy to determine plant-fungal interactions (Arens et al. 2016;Couture et al. 2018;Fallon et al. 2019;Meacham-Hensold et al. 2020;Gold et al. 2019Gold et al. , 2020. Since the 1980s, hyperspectral and wideband approaches based on visible and near-infrared reflectance indicators, such as the normalized difference index, have been employed to detect late-stage photo disease (Jackson 1986;Hatfield and Pinter 1993;Nilsson 1995). ...
... In agricultural plant-breeding systems, this strategy has proved beneficial in enhancing efficacy and precision (Ge et al. 2019;Meacham-Hensold et al. 2019). Therefore, the spectral identification of foliar functional characteristics enables in identifying, tracing, and simulating physiological and biochemical pathogenic processes that enable the employment of spectroscopy to determine plant-fungal interactions (Arens et al. 2016;Couture et al. 2018;Fallon et al. 2019;Meacham-Hensold et al. 2020;Gold et al. 2019Gold et al. , 2020. Since the 1980s, hyperspectral and wideband approaches based on visible and near-infrared reflectance indicators, such as the normalized difference index, have been employed to detect late-stage photo disease (Jackson 1986;Hatfield and Pinter 1993;Nilsson 1995). ...
... When applied to plant disease, spectranomics allows for accurate and non-destructive detection of direct and indirect changes to plant physiology, morphology, and biochemistry which induces the disease, both pre-and post-symptomatically (Arens et al. 2016;Couture et al. 2018;Fallon et al. 2020;Gold et al. 2020). Beneficial (Sousa et al. 2021) and parasitic (Zarco-Tejada et al. 2018) plant-microbe interactions impact a variety of plant traits that can be remotely sensed. ...
Article
Over the last 20 years, researchers in the field of digital plant pathology have chased the goal to implement sensors, machine learning and new technologies into knowledge-based methods for plant phenotyping and plant protection. However, the application of swiftly developing technologies has posed many challenges. Greenhouse and field applications are complex and differ in their study design requirements. Selecting a sensor type (e.g., thermography or hyperspectral imaging), sensor platform (e.g., rovers, unmanned aerial vehicles, or satellites), and the problem-specific spatial and temporal scale adds to the challenge as all pathosystems are unique and differ in their interactions and symptoms, or lack thereof. Adding host-pathogen-environment interactions across time and space increases the complexity even further. Large data sets are necessary to enable a deeper understanding of these interactions. Therefore, modern machine learning methods are developed to realize the fast data analysis of such complex data sets. This reduces not only human effort but also enables an objective data perusal. Especially deep learning approaches show a high potential to identify probable cohesive parameters during plant-pathogen-environment interactions. Unfortunately, the performance and reliability of developed methods are often doubted by the potential user. Gaining their trust is thus needed for real field applications. Linking biological causes to machine learning features and a clear communication, even for non-experts of such results, is a crucial task that will bridge the gap between theory and praxis of a newly developed application. Therefore, we suggest a global connection of experts and data as the basis for defining a common and goal-oriented research roadmap. Such high interconnectivity will likely increase the chances of swift, successful progress in research and practice. A coordination within international excellence clusters will be useful to reduce redundancy of research while supporting the creation and progress of complementary research. With this review, we would like to discuss past research, achievements, as well as recurring and new challenges. Having such a retrospect available, we will attempt to reveal future challenges and provide a possible direction elevating the next decade of research in digital plant pathology.
... But adverse interaction of the sensor with the required UV light source restrains interpretation of these results. Another study combined spectral imaging in the SWIR range (970-2500 nm) with untargeted metabolic fingerprinting of three different sugar beet genotypes infected with Cercospora beticola [82]. Although there were correlations between several secondary metabolites and spectral data, it remains unclear if this correlation is due to direct contribution of these metabolites to the reflectance spectrum. ...
Article
Full-text available
Plant pathogens cause yield losses in crops worldwide. Breeding for improved disease resistance and management by precision agriculture are two approaches to limit such yield losses. Both rely on detecting and quantifying signs and symptoms of plant disease. To achieve this, the field of plant phenotyping makes use of non-invasive sensor technology. Compared to invasive methods, this can offer improved throughput and allow for repeated measurements on living plants. Abiotic stress responses and yield components have been successfully measured with phenotyping technologies, whereas phenotyping methods for biotic stresses are less developed, despite the relevance of plant disease in crop production. The interactions between plants and pathogens can lead to a variety of signs (when the pathogen itself can be detected) and diverse symptoms (detectable responses of the plant). Here, we review the strengths and weaknesses of a broad range of sensor technologies that are being used for sensing of signs and symptoms on plant shoots, including monochrome, RGB, hyperspectral, fluorescence, chlorophyll fluorescence and thermal sensors, as well as Raman spectroscopy, X-ray computed tomography, and optical coherence tomography. We argue that choosing and combining appropriate sensors for each plant-pathosystem and measuring with sufficient spatial resolution can enable specific and accurate measurements of above-ground signs and symptoms of plant disease.
... Advancements have been made in assessing resistance to one of the most dangerous sugar beet pathogens -Cercospora beticola causing cercospora leaf spot (CLS). Hyperspectral imaging can detect symptoms of CLS early, accurately and in a non-destructive fashion (Arens et al., 2016;Leucker et al., 2016). PhenoTest is an automated high-throughput germination test for sugar beet seeds. ...
Thesis
Full-text available
Sugar beets (Beta vulgaris ssp. vulgaris L.) are important source of sucrose covering about 14% of global production. Increasing demand for caloric sweeteners and changing climate raises need for more robust and efficient varieties of this crop. There are many new technologies being developed which can aid sugar beet breeding programmes with faster and more informative phenotyping, however, optical coherence tomography (OCT) and microscopy (OCM) have not been examined yet. This study evaluates OCT and OCM as potential techniques for assessment of sugar beet leaf morphology traits and compares their usefulness in achieving this task. In previous research on OCM measurements of beet leaves it was hypothesized that the thickness of the cuticle could be measured. However, from OCT measurements it was concluded that the cuticle cannot be visualized with either setup, and that the ‘cuticle thickness’ measured before this study was in fact the thickness of guard cells. Therefore, this study focused on the thickness of the epidermal layer as a phenotypic trait. Leaves of plants representing four genotypes of sugar beets were tested with both techniques (5 spots per leaf, 2 leaves per plant, 4 plants per genotype, in total 160 samples). The acquired data was analysed to determine thickness of stomatal guard cells and epidermal cells. Both techniques allowed for assessment of guard cell thickness, although only OCT allowed for measurement of epidermal cell thickness. Alongside descriptive statistics, analysis of variance was performed to determine sources of variation in measured samples, which resulted in significant models explaining impact of genotype, plant, leaf and their interactions on the results. Significant differences in measured traits between genotypes were found, however, high variability within samples and significant interactions between factors were also observed in the data. Differences between the two optical imaging techniques and their respective advantages and downsides were described. This study found OCT and OCM useful for quantifying morphological properties of beet leaves, some of which can be linked to drought resistance. Both methods are worth further examination due to their ability to assess more anatomical characteristics of beet leaves linked with tolerance to water scarcity and possibly with pathogen resistance.
Article
Plant pathogens continuously impair agricultural yields and food security. Therefore, the dynamic characterization of early pathogen progression is crucial for disease monitoring and presymptomatic diagnosis. Hyperspectral imaging (HSI) has great potential for tracking the dynamics of initial infected sites for presymptomatic diagnosis; however, no related studies have extracted fingerprint spectral signatures (FSSs) that can capture diseased lesions on leaves during the early infection stage in vivo or investigated the detection mechanism of HSI relating to the host biochemical responses. The FSSs denote unique and representative spectral signatures that characterize a specific plant disease. In this study, the FSSs of spot blotch on barley leaves inoculated with Bipolaris sorokiniana were discovered to characterize symptom development for presymptomatic diagnosis based on time-series HSI data analysis. The early spectral and biochemical responses of barley leaves to spot blotch progression were also investigated. The full-spectrum FSSs were physically interpretable and could capture the unique characteristics of chlorotic and necrotic tissues along with lesion progression, enabling the in situ visualization of the spatiotemporal dynamics of early plant-pathogen interactions at the pixel level. Presymptomatic diagnosis of spot blotch was achieved 24 h after inoculation—12 h earlier than the traditional polymerase chain reaction (PCR) assay or biochemical measurements. To uncover the mechanism of HSI presymptomatic diagnosis, quantitative relationships between the mean spectral responses of leaves and their biochemical indicators (chlorophylls, carotenoids, malondialdehyde (MDA), ascorbic acid (AsA), and reduced glutathione (GSH)) were developed, achieving determination coefficient (R²) of prediction set (Rp²)> 0.84 for regression models. The overall results demonstrated that, based on the association between HSI and in vivo plant-trait alterations, the extracted FSSs successfully tracked the spatiotemporal dynamics of bipolaris spot blotch progression for presymptomatic diagnosis. Tests of this methodology on other plant diseases demonstrated its remarkable generalization potential for the early control of plant diseases.
Article
Full-text available
The ability to detect diseased trees before symptoms emerge is key in forest health management because it allows for more timely and targeted intervention. The objective of this study was to develop an in-field approach for early and rapid detection of beech leaf disease (BLD), an emerging disease of American beech trees, based on supervised classification models of leaf near-infrared (NIR) spectral profiles. To validate the effectiveness of the method we also utilized a qPCR-based protocol for the quantification of the newly identified foliar nematode identified as the putative causal agent of BLD, Litylenchus crenatae ssp. mccannii (LCM). NIR spectra were collected in May, July, and September of 2021 and analyzed using support vector machine and random forest algorithms. For the May and July datasets, the models accurately predicted pre-symptomatic leaves (highest testing accuracy = 100%), but also accurately discriminated the spectra based on geographic location (highest testing accuracy = 90%). Therefore, we could not conclude that spectral differences were due to pathogen presence alone. However, the September dataset removed location as a factor and the models accurately discriminated pre-symptomatic from naïve samples (highest testing accuracy = 95.9%). Five spectral bands (2,220, 2,400, 2,346, 1,750, and 1,424 nm), selected using variable selection models, were shared across all models, indicating consistency with respect to phytochemical induction by LCM infection of pre-symptomatic leaves. Our results demonstrate that this technique holds high promise as an in-field diagnostic tool for BLD.
Article
Powdery mildew caused by blumeria graminis is responsible for wheat yield losses in combination with a decline in quality. Hyperspectral imaging as a promising non-invasive sensor technique has potential for early diagnosis and pathogenesis monitoring of wheat powdery mildew, which is a practice that allows for precision crop protection. Hyperspectral images were first captured before inoculation as healthy samples and daily 2 to 5 days after inoculation (dai) as infected ones. Principal component analysis (PCA) was applied to observe the discrimination capability between samples at different infected stages, while a gray-level co-occurrence matrix (GLCM) was used to extract textural features from the first three principal component images. Then partial least squares discriminant analysis (PLS-DA) model was developed to evaluate the ability for early diagnosis of the disease using effective wavelengths, texture features and their fusion, respectively. Compared with the models using spectral or textural feature alone, PLS-DA model using the fused dataset obtained the best performances with classification accuracy of 91.4 % in validation sets. Furthermore, spectral angle mapping (SAM) was performed to identify the infected tissue in wheat leaves 2 dai, and to monitor the pathogenesis of powdery mildew over time. The results from this study could be used to develop a portable field monitoring sensor for wheat powdery mildew.
Article
Full-text available
There is a general consensus that supports the need for standardized reporting of metadata or information describing large-scale metabolomics and other functional genomics data sets. Reporting of standard metadata provides a biological and empirical context for the data, facilitates experimental replication, and enables the re-interrogation and comparison of data by others. Accordingly, the Metabolomics Standards Initiative is building a general consensus concerning the minimum reporting standards for metabolomics experiments of which the Chemical Analysis Working Group (CAWG) is a member of this community effort. This article proposes the minimum reporting standards related to the chemical analysis aspects of metabolomics experiments including: sample preparation, experimental analysis, quality control, metabolite identification, and data pre-processing. These minimum standards currently focus mostly upon mass spectrometry and nuclear magnetic resonance spectroscopy due to the popularity of these techniques in metabolomics. However, additional input concerning other techniques is welcomed and can be provided via the CAWG on-line discussion forum at http://msi-workgroups.sourceforge.net/ or http://Msi-workgroups-feedback@lists.sourceforge.net. Further, community input related to this document can also be provided via this electronic forum.
Article
Full-text available
Plant diseases cause significant reductions in agricultural productivity worldwide. Disease symptoms have deleterious effects on the growth and development of crop plants, limiting yields and making agricultural products unfit for consumption. For many plant-pathogen systems, we lack knowledge of the physiological mechanisms that link pathogen infection and the production of disease symptoms in the host. A variety of quantitative high-throughput image-based methods for phenotyping plant growth and development are currently being developed. These methods range from detailed analysis of a single plant over time to broad assessment of the crop canopy for thousands of plants in a field and employ a wide variety of imaging technologies. Application of these methods to the study of plant disease offers the ability to study quantitatively how host physiology is altered by pathogen infection. These approaches have the potential to provide insight into the physiological mechanisms underlying disease symptom development. Furthermore, imaging techniques that detect the electromagnetic spectrum outside of visible light allow us to quantify disease symptoms that are not visible by eye, increasing the range of symptoms we can observe and potentially allowing for earlier and more thorough symptom detection. In this review, we summarize current progress in plant disease phenotyping and suggest future directions that will accelerate the development of resistant crop varieties.
Article
Liquid chromatography-mass spectrometry (LC-MS) technology allows rapid quantitation of cellular metabolites, with metabolites identified by mass spectrometry and chromatographic retention time. Recently, with the development of rapid scanning high-resolution high accuracy mass spectrometers and the desire for high throughput screening, minimal or no chromatographic separation has become increasingly popular. When analyzing complex cellular extracts, however, the lack of chromatographic separation could potentially result in misannotation of structurally related metabolites. Here we show that, even using electrospray ionization, a soft ionization method, in-source fragmentation generates unwanted byproducts of identical mass to common metabolites. For example, nucleotide-triphosphates generate nucleotide-diphosphates and hexose-phosphates generate triose-phosphates. We evaluated yeast intracellular metabolite extracts and found more than 20 cases of in-source fragments that mimic common metabolites. Accordingly, chromatographic separation is required for accurate quantitation of many common cellular metabolites.
Article
In recent years there has been increased interest in the role of phenolic compounds in the physiology of disease resistance. We believe that in the resistance of sugarbeets to Cercospora leaf spot a specific phenolic compound, dopamine (3-hydroxy­ tyramine), has a prominent function. In this paper we present a few ideas as to the complex response that enables some varieties of sugarbeets to ma nifest resistance to attacks of Cercospora beticola (Sacc.). A correlation has been found between the amount of dopa­ mine in the leaves of sugarbeets and their resistance to attacks of Cercospora beticola (9,12) . Although there is a significant re­ lationship between dopamine content and resistance, a few susce ptible varieties are relatively high in dopamine. Hence the mechanism of resistance is not a simple phenomenon as in the case of resistance to onion smudge. If the role of dopamine in disease resistance were elucidated it might lead to understanding and control of leaf spot disease and might have implications in other fungal diseases. In previous papers we have discussed some of the aspects on infection of sugarbeets with Cercospora beticola (6,8,9, II ,12) and recent studies indicate that the resistance may be due to specific biochemical properties of the host. One of the best documented characteristics of resistance to fungal diseases in general is a sharp, post-infectional raise in phenol concentrations (5,17,20). Dopamine and other . phenolic compounds have been detected and identified in the leaves of sugarbeets (6,9). "There are considerable quantities of dopamine in the leaves of healthy, resistant sugarbeets and the amount of dopamine is further in­ creased upon mechanical injury to the leaves (8). When dopa­ mine is oxidized it inhibits the growth of the fungus (9) . The correlation between the dopamine content of sugarbeet leaves and their disease resistance has been reported (12).
Article
Ensembles of RBF networks trained with \(\gamma\)-divergence-based similarity measures can improve classification accuracy of hyperspectral imaging data significantly compared to any single RBF network as well as to RBF ensembles based on the Euclidian distance. So far, the drawback of using classifier ensembles is the need to compute the results of a typically large number of RBF networks prior to combination. In this paper, a modified approach to the fusion of classifier outputs is proposed which is based on decision trees. It is shown for several real-world datasets that a small subset of the RBF networks contributes to the decisions in the average case. Hence, for any decision, a conditional computation of required RBF network outputs yields a significant decrease in the computational costs. Additionally, a selection scheme for subsets of RBF classifiers based on their relevance in the fusion process is proposed. This alternative approach can be used, if analysis requires fixed settings, e.g., to meet time constraints.