Next Article in Journal
Insights from Farmers’ Rice Culture Practices under Integrated Rice–Crayfish Farming System in the Hongze Lake District of China
Previous Article in Journal
Organic Amendment for Disinfecting Soil Alters the Metabolites in Spinacia oleracea
 
 
Search for Articles:
Title / Keyword
Author / Affiliation / Email
Journal
Article Type
 
 
Advanced Search
 
Section
Special Issue
Volume
Issue
Number
Page
 
Journals
Agriculture
Volume 13
Issue 12
10.3390/agriculture13122228
agriculture-logo
Submit to this Journal Review for this Journal Propose a Special Issue
► Article Menu

Article Menu

share Share announcement Help format_quote Cite question_answer Discuss in SciProfiles
first_page
settings
Order Article Reprints
Font Type:
Arial Georgia Verdana
Font Size:
Aa Aa Aa
Line Spacing:
Column Width:
Background:
Article

Classification of Plenodomus lingam and Plenodomus biglobosus in Co-Occurring Samples Using Reflectance Spectroscopy

by
Andrzej Wójtowicz
1,
Jan Piekarczyk
2,*,
Marek Wójtowicz
3,*,
Jarosław Jasiewicz
2,
Sławomir Królewicz
2 and
Elżbieta Starzycka-Korbas
3
1
Institute of Plant Protection-National Research Institute, 60-318 Poznan, Poland
2
Faculty of Geographic and Geological Sciences, Adam Mickiewicz University, 60-680 Poznan, Poland
3
Plant Breeding and Acclimatization Institute-National Research Institute in Radzików, 60-479 Poznan, Poland
*
Authors to whom correspondence should be addressed.
Agriculture 2023, 13(12), 2228; https://doi.org/10.3390/agriculture13122228
Submission received: 27 October 2023 / Revised: 20 November 2023 / Accepted: 28 November 2023 / Published: 30 November 2023
(This article belongs to the Section Crop Protection, Diseases, Pests and Weeds)
Browse Figures
Versions Notes

Abstract

:
Under natural conditions, mixed infections are often observed when two or more species of plant pathogens are present on one host. Thus, the detection and characterization of co-occurring pest species is a challenge of great importance. In this study, we focused on the development of a spectral unmixing method for the discrimination of two fungi species, Plenodomus lingam and Plenodomus biglobosus, the pathogens of oilseed rape. Over 24 days, spectral reflectance measurements from Petri dishes inoculated with fungi were conducted. Four experimental combinations were used: the first two were pure fungal samples, while the other two were co-occurring fungal samples. The results of the study show the possibility of distinguishing, based on spectral characteristics, between P. lingam and P. biglobosus not only in pure but also in co-occurring samples. We observed the changes in the reflectance of electromagnetic radiation from the tested fungi over time and a strong correlation between the reflectance and changes in the areas of the mycelia on the Petri dishes. Moreover, the wavelengths most useful for spectral classification of the tested fungal mycelia were selected. Finally, a spectral unmixing model was proposed, which enables the estimation of the areas of two pathogens in co-occurring samples based on the spectral characteristics of the entire plate with an error smaller than 0.2. To our knowledge, the present study is the first report examining the use of reflectance spectroscopy methods for classifying pathogens on the same Petri dish. The study results indicate the feasibility of reflectance spectroscopy as a nondestructive sampling method for plant pathogen detection and classification.

1. Introduction

Remote sensing techniques have recently been widely used to detect crop pathogens [1,2,3]. The successful application of these methods in field conditions was usually preceded by tests carried out under controlled conditions in the laboratory [4,5,6,7,8]. One of these methods is reflectance spectroscopy, in which the amount of reflected light is measured using hyperspectral cameras [9,10,11,12], multispectral cameras [13,14], RGB cameras [15,16], and spectrometers [17,18].
An example of reflectance spectroscopy conducted with the use of a spectrometer was the study of Piekarczyk et al. [19], who showed the possibility of identifying six species of entomopathogenic fungi (Beauveria bassiana, Isaria fumosorosea, I. farinosa, I. tenuipes, Lecanicillium lecanii, and L. muscarium). The separation accuracy of the analyzed pathogens, depending on the measurement methods used, was 84–100%. Applying machine learning with the Savitzky–Golay filter, two visible wavelengths (411 and 520 nm) were enough to distinguish the six species successfully. Meanwhile, Aboelghar and Wahab [20] demonstrated the effectiveness of short-wave infrared (SWIR) waves (2055–2315 nm) to distinguish B. cinerea from Aspergillus sp., Rhizopus sp., Penicillium italicum, Penicillium digitatum, Fusarium sp., Alternaria sp., and Rhizoctonia sp. SWIR was also helpful in distinguishing B. cinerea from R. solani and S. sclerotiorum, and for distinguishing R. solani from S. sclerotiorum; the best results were obtained using VIS wavelengths. The effectiveness of identifying B. cinerea, R. solani, and S. sclerotiorum in these studies was 91, 94, and 95%, respectively [21]. The authors of [22] used VNIR waves to group B. cinerea isolates, and VNIR (743, 458, and 541 nm) also allowed us to distinguish five species of fungi: Penicillium chrysogenum, Fusarium moniliforme (verticillioides), Aspergillus parasiticus, Trichoderma viride, and Aspergillus flavus, with an accuracy of 97.7%
The studies presented above focused on distinguishing pathogens that occur alone. However, under natural conditions, mixed infections are often observed when two or more species of pathogens are present on one host. An illustration of such a situation is the development of Phoma stem canker (blackleg) caused by two Plenodomus species on rapeseed. Plenodomus lingam (formerly Leptosphaeria maculans) and Plenodomus biglobosus (formerly Leptosphaeria biglobosa) [23] coexist on oilseed rape in North America [24], Argentina [25], and several European countries, including the UK [26], France [27], Germany [28], Scandinavia [29], Hungary [30], the Czech Republic [31], and Lithuania [32]. Phoma stem canker is a disease of major economic importance in the main oilseed rape-growing areas of Australia, Canada, and Europe. Yield losses caused by these pathogens usually do not exceed 10%. However, for susceptible cultivars in meteorological conditions favoring the pathogen development, they can be as high as 30–50% [27].
Molecular methods, such as PCR (polymerase chain reaction) and restriction fragment length polymorphism (RFLP), and sequencing strategies, such as multilocus sequence typing (MLST), are commonly used to detect several pathogens simultaneously. These methods are considered the most sensitive and specific but are time-consuming and expensive [33]. For this reason, there is a need to look for new tools to solve the problem that are cheaper and faster. A literature review that focused on the discrimination of fungi using spectroscopic methods revealed some reports describing successful results obtained with FTIR-ATR spectroscopy [34,35]. Although these methods successfully discriminate pathogens in mixtures, they do not offer a solution for the immediate identification of fungi in co-occurring samples on the same Petri dish.
The study aimed to develop a method for discriminating two fungi co-occurring on one Petri dish. Three specific objectives were identified to achieve the above goal: (1) to estimate the changes in the reflectance of electromagnetic radiation from the tested fungi over time and measure the correlation between the reflectance and changes in the area of the mycelia on the Petri dishes used, (2) to select the wavelengths most useful for spectral classification of the tested fungal mycelia, and (3) to propose a new machine learning model for distinguishing two species of co-occurring fungi and measuring their areas.

2. Materials and Methods

2.1. Mycelial Growth

The fungal strains used in the study came from the collection of the Plant Breeding and Acclimatization Institute—National Research Institute. They were isolated from the stems of oilseed rape plants and classified as P. lingam and P. biglobosus based on their morphological and molecular characteristics (ITS1 region sequences). The experiment was carried out in Petri dishes with Potato Dextrose Agar medium (PDA) inoculated by placing a 5 mm square of fungus growth obtained from the edge of another culture in a Petri dish, which guaranteed vigorous growth of the mycelia. Four experimental combinations in 10 repetitions were used. The first two combinations were pure fungal samples (PFSs), while the other two were co-occurring fungal samples (CoFSs) (Table 1). In the PFSs, P. lingam and P. biglobosus were placed at the center of separate Petri dishes, while in the CoFSs both fungi were put together on the same plate. In the 3rd combination, the inocula of P. lingam and P. biglobosus were deposited next to each other, while in the 4th combination, they were placed at a distance of 4 cm from one another.

2.2. Measurement of Mycelial Areas

Photographs of the Petri dishes with the fungal mycelia were taken on eight dates (1, 3, 8, 10, 15, 17, 22, and 24 days after inoculation) with a Canon 80D SLR camera (Canon, London, Great Britain) with a varifocal lens F-S 18-135mm f/3.5–5.6 IS USM. Images were taken at 85 mm focal length, 5.6 aperture, ISO 1600, and variable auto-adjustable shutter speed (the only variable shooting parameter). Photos were saved in JPG format with the lowest degree of compression available in the settings (the so-called highest quality).
Geometric distortions of the lens, caused mainly by radial distortion and each shift of the image center relative to the center of the Petri dish glass, forced geometric correction of each photo. This allowed us to recreate and preserve the original circular shape of the Petri dish in all images. Image correction was performed using the Lens Correction module with Datum Geospatial 2023 software from the Landscan company. This module uses the Lensfun database distributed based on the General Public License (version 3), which contains data on the properties of over 1200 camera lenses from almost all commercial camera manufacturers [36].
In the next step, the classification of each image was performed, the aim of which was to obtain a percentage area of two classes: one was the extent of the mycelia, and the other covered the remaining part of the surface of the dish. The percentage was related to the area occupied by the mycelia. The classification was performed unsupervised using the k-means method, implemented in the Datum Geospatial software (Automatic Image Classification Feature process); as input data for classification, the spectral channels R, G, and B were introduced as three separate features.
Different numbers of classification repetitions were used. For some photos, the desired classification result was obtained on the first attempt (2nd class); for others, after several tries. In complex cases, the classification parameters were optimized using different numbers of y classes from 2 to 6; sometimes, other classification parameters were optimized, i.e., the number of iterations, the initial distance between classes, the maximum movement for steadiness, and the minimum cluster steady percentage. Finally, the number of two classes was obtained by manually combining classes. The results of classification were visually verified by superimposing the classification results on the image in the RGB color composition.
In the case of the CoFSs, the results of the raster classification were converted to vector form. In this form, the area occupied by each species of fungi was obtained. The results obtained enabled us to express graphically the mycelial growth and calculate the area under the mycelial growth curve (AUMGC) using the following formula [37]:
A U M G C = ∑ i = 1 n − 1 ( X i + 1 + X i / 2 ] [ t i + 1 − t i ]
where Xi = the % of the Petri dish covered with the mycelium at the ith observation, ti = time (days after inoculation) at the ith observation, and n = the total number of observations.
Based on the data collected when measuring the areas of the mycelia on Petri dishes, the growth inhibition index (I) was calculated as [38]:
I= [(C − T)/C] × 100
where C is the area of P. lingam in the PFS (2nd combination) and T is the area of P. lingam in the CoFSs (3rd and 4th combinations).

2.3. Spectral Measurements

The spectral characteristics of the fungal mycelia on Petri dishes ranging from 350 nm to 2500 nm were measured using an ASD spectrophotometer (FieldSpec Analytical Spectral Devices, Inc., Boulder, CO, USA). Spectral measurements were converted to reflectance by dividing them by measurements performed using a white reference panel (Spectralon: Diffuse Reflectance Target, LabSphere Inc., North Sutton, NH, USA). The pistol probe of the ASD spectrophotometer with a 10° field of view (FOV) for optics was held vertically 30 cm above the Petri dish. The diameter of the measurement area was 8 cm, which was 1 cm smaller than the diameter of the Petri dish. Each dish with a fungi culture was measured four times at four different positions on the plate, resulting in 40 measurements for each combination of the experiment. A 400 W halogen lamp mounted 90 cm from the sample illuminated the measured area at an angle of 45° to the zenith. A black diffusing plate was placed under the Petri dish holder to decrease the reflectance of the semitransparent PDA medium.

2.4. Statistical Analysis

The visual assessment of the obtained spectra showed that the differences between the tested species on subsequent measurement dates were evident and covered the entire spectral range. Since there were more than 2000 bands to analyze, showing the differences between the analyzed experimental combinations was straightforward, but indicating their most important wavelengths was not trivial. The developed methodology used VIS-NIR-SWIR spectra to identify features that distinguished P. lingam from P. biglobosus on subsequent measurement dates and to estimate the proportions of their areas in the CoFSs (unmixing) (Figure 1).

2.4.1. Qualitative Differentiation of Pure (PFS) and Co-Occurring (CoFS) Fungal Samples

Principal component analysis (PCA) was used as a standard tool for dimensionality reduction in situations with identification problems. Loadings represent significant features in the PCA space, and a biplot is the traditional method of presenting them [39]. It uses principal component spaces (usually the first and second), which are more easily separable, to illustrate the variations in the objects and loadings, including the direction of original features and their changes. The number of features (2150) was too large to plot, so we used a method to identify those representing more extensive spectrum ranges. The loadings represent covariances between the original variables and the unit-scaled components and are usually represented by arrows. The longest arrows represent variables that contribute most to the given components. Based on the loadings for the first and second components, we determined the directions and intensity for each wavelength:
α = arctan(LPC1, LPC2)
ᾱ = hypot(LPC1, LPC2)
where α-direction, ᾱ-intensity, LPC1, and LPC2 are loadings of the first and second principal components and hypot is the root of the sum of squares.
In this way, we obtained information about the intensity of the loadings for each wavelength, presented as an ᾱ curve. For these curves, we used the find_peaks algorithm from the scipy.signal library, and we determined local maxima on the ᾱ curves, which we considered to represent significant wavelength ranges.

2.4.2. Quantitative Assessment of Species in Co-Occurring Fungal Samples (Unmixing)

We applied machine learning algorithms to estimate the probability of predicting the proportions of the two species of fungi. Among several popular regressors, such as Random Forest [40], XGBoost [41], Elastic Net [42], and K-Nearest Neighborhoods [43], we selected Partial Least Squares (PLS), a method frequently used in chemometrics [44]. In order to avoid co-linearity, a PLS regressor [45] was trained and tested on a selected number of PCs instead of raw data.
The dependent variable (modeled) was the proportion of P. biglobosus and P. lingam species. Since, in all observed samples, the share of P. biglobosus was greater than or equal to the share of P. lingam, the value of this relationship was 0 (no P. lingam) or 1 (the share of P. lingam equaled that of P. biglobosus). For this experiment, the machine learning procedure was implemented non-standardly. Samples were not divided into training and testing sets; rather, all spectral measurements of the mixtures of both fungal species were used as the testing set, while the training set was created as a collection of PFS spectra artificially mixed in random proportions.
The spectra were mixed according to the formula below [34]:
(1-γ) × L1–10 + γ × B1–10
where γ is a random number in a [0,1] range, L1–10 are P. lingam spectra selected at random from a set of 10 spectra for a given development phase, and B1–10 are P. biglobosus spectra selected at random from a set of 10 spectra for a given development phase. In order to simulate the spectra of the training set, we used the same ranges of species shares as in the testing set, e.g., P. lingam shares did not exceed P. biglobosus shares. In the end, 500 spectra and their corresponding species proportions were obtained as a training set. Thus, the testing set was entirely independent of the training set in terms of the samples used to train the regressor and the method used to obtain it. Finally, we evaluated the model’s quality using the Root Mean Square Error method. All routines were implemented in Python using the scikit-learn, XGBoost, and scipy/nympy/matplotlib libraries.

3. Results

3.1. Changes in Mycelial Areas

The mycelia of P. biglobosus developed faster than those of P. lingam, both in the PFSs and CoFSs (Figure 2 and Figure 3). The fastest increase in the area of a P. biglobosus mycelium was recorded for the PFSs. After 24 days, it occupied 56.3 cm2 (89% of the dish surface).
In the CoFSs, in the third and fourth combinations, the areas occupied by P. biglobosus were smaller than in the PFSs and were 26.1 (41% of the area of the Petri dish) and 28.1 cm2 (44%), respectively. The maximum area occupied by P. lingam in the PFSs after 24 days was 34.1 cm2 (54%). In the CoFSs, in the third and fourth combinations, the fungus areas were 7.3 (11%) and 22.6 cm2 (36%), respectively. The ratio of AUMGCs for P. biglobosus to P. lingam (third combination) grew for 15 days and then stabilized (Table 2). However, for the PFSs and the fourth combination, stabilization of these ratios was registered after 8 days of the experiments, and the ratios were 1.9 and 1.3, respectively.
The results presented in this study agree with outcomes obtained by others—faster growth of P. biglobosus than P. lingam mycelia was reported by Newbery et al. [46]. Mycelial growth of both species on culture media was also described by Jędryczka et al. [47]. The authors, conducting observations at 10, 15, and 20 °C temperatures, showed that the P. biglobosus mycelia grew more than twice as fast as those of P. lingam. Also, studies performed before the taxonomic separation of P. lingam and P. biglobosus reported faster growth of ‘avirulent’, ‘B group’ isolates (i.e., P. biglobosus) than ‘virulent’, ‘A group’ isolates (i.e., P. lingam) [48,49]. The results obtained allowed the authors to estimate the antagonistic relationships between the tested fungi. The inhibition of P. lingam mycelial growth (I) by P. biglobosus in the third and fourth combinations was 78.6% and 33.6%, respectively (Figure 3).

3.2. Temporal Changes in Mycelial Spectral Characteristics

Due to the minimal area occupied by the mycelia of both P. biglobosus and P. lingam in all the experimental combinations, the spectral characteristics obtained till eight days after determined the PDA medium inoculation. The spectral reflectance from the PFSs increased in all spectral ranges, reaching its maximum on the 22nd (P. biglobosus) and 24th days (P. lingam) after deposition on the dish (Figure 4). The most significant changes in the spectra over time were observed in the range of 450–1000 nm (VNIR). Up to the 8th day after the deposition of the mycelia on the dish, the reflectance tended to decrease with wavelength, whereas after the 17th and 15th days, respectively, for P. biglobosus and P. lingam, the slopes of the spectra in that range increased rapidly.
The most significant spectral differences between P. biglobosus and P. lingam were registered from the 15th to the 24th day in the VNIR range. The spectra of the CoFSs were similar in shape to those of P. biglobosus due to the greater area occupied by that fungus compared to P. lingam. In contrast, the reflectance intensity was smaller than those registered for the PFSs. For all the experimental combinations, spectral reflectance decreased after reaching its maximum despite the continuous growth of mycelia. The latter result is consistent with the outcomes of Chu et al. [50], who monitored Colletotrichum truncatum and Colletotrichum gloeosporioides development on Petri dishes with hyperspectral images. These fungi species developed faster than the species in our experiment and reached maximum reflectance on the 3rd and 4th days after deposition on the dish.

3.3. Spectral Classification of Fungal Mycelia

The ability to distinguish all the experimental combinations changed with fungi development. Until the third day after the deposition of the mycelia on the dish (first and second dates of measurement), when the area occupied by the mycelia was minimal, discrimination was impossible. On the third and fourth dates of measurement, both the PFSs and CoFSs were successfully distinguished (Figure 5). PC1 discriminated P. biglobosus from co-occurring fungal samples placed next to each other (third combination). In contrast, PC2 divided pure P. lingam samples from the CoFSs placed at a distance of 4 cm from one another (fourth combination). Successful discrimination between all combinations was possible when two PCs were used.
On the fifth and sixth dates of measurement, pure P. lingam samples were well distinguished from the other combinations, while the third and fourth combinations became alike due to the growth of the mycelia. On the seventh and eighth measurement dates, PC1 discriminated successfully pure P. lingam from pure P. biglobosus samples and PFSs from CoFSs. As the mycelia developed, it became easier to distinguish the pure P. lingam from the pure P. biglobosus samples. Due to the small size of the mycelia, it was not possible to distinguish between the third and fourth combinations (CoFSs) on the first two dates. For the third and fourth measurement dates, the best discrimination of all combinations was registered. However, from the fifth date of measurement, the effectiveness of this discrimination decreased due to the similar mycelial areas of the co-occurring fungal samples.
From 15 days after inoculation, three bands in the 420–424 nm, 899–913 nm, and 1074–1095 nm ranges were the most useful for distinguishing all the experimental combinations. The 420–424 nm wavelength band separated the P. lingam PFS from the other combinations, while the wavelength bands of 899–913 nm and 1074–1095 nm discriminated the P. biglobosus PFS from the CoFSs. On the earlier dates (8 and 10 days after inoculation), seven and nine wavelengths helped discriminate all the experimental combinations, respectively.

3.4. Spectral Unmixing of Fungal Mycelia

In our research, we checked whether it is possible to determine the areas of two mycelia placed on one dish based on the reflectance characteristics of co-occurring fungal samples. Since the obtained spectrum represented the averaged reflection from the entire surface of the plate on which both fungi were located, there was a need to develop a new classification method. The first step of the method proposed consisted of testing five regressors (Random Forest, XGBoost, Elastic Net, K-Nearest Neighborhoods, and PLS). They were compared based on the values of the error rates. Only PLS reached a satisfactory value of this parameter (RMSE < 0.2) for the last two development phases (Table 3). An RMSE above 0.3 means, in practice, that the regressor predicts a similar level of mixing for all mixtures of species at the equilibrium level (0.5).
Figure 6 shows the relationship between the area ratios of P. biglobosus and P. lingam on the training (blue) and testing (orange) sets using the PLSR model. The accuracy levels of this method, expressed as RMSEs, differed during the experiment. Until the 8th day from the start of the experiment, when the average mycelial area was about 10% of the plate surface, the RMSE was greater than 1.0. From day 10 to day 17, as the fungi area increased to 27%, the RMSE decreased to 0.49. The best predictions were obtained on the last two dates of measurement. In these two cases, the average area occupied by the fungi on the plate was above 30%, and the RMSE was smaller than 0.2.
To our knowledge, the present study is the first attempt to explore reflectance spectroscopy methods for classifying pathogens on the same Petri dish. However, some examples exist where FTIR-ATR spectroscopy has been used to distinguish pathogens in mixtures. In such cases, the fungi samples before spectroscopic measurement were purified from the medium and suspended in distilled water. Such methods were used by Huleihel et al. [34] in a study focused on distinguishing four different fungi genera: Colletotrichum, Verticillium, Rhizoctonia, and Fusarium. The classification rate in this study was 85–100%. The classification success rate obtained with different PCs using the infrared spectra in the 625–11,111 nm region depended on the analyzed fungi’s mixing ranges. The best results, with an almost 100% success rate, were achieved when the mixing ratio was 50:50. Interesting results were presented by Salman et al. [35], who also used the FTIR method to distinguish Fusarium oxysporum and Fusarium solani from their mixed type with a success rate of 94%.

4. Conclusions

The growth of P. biglobosus was faster than that of P. lingam in pure fungal samples (PFSs) and co-occurring fungal samples (CoFSs). The fastest development of mycelia was observed for the PFSs, while the slowest was observed for the CoFS combination with the mycelia placed next to each other. The spectral characteristics changed during the experiment. From the fifth date of measurement, the spectral properties of P. biglobosus determined the shape of the spectra of the CoFSs. The study showed the possibility of distinguishing between the PFSs and between the PFSs and CoFSs. The accuracy of classification was determined by the date of the measurement. The conducted experiments showed the usefulness of the PLS method for estimating the areas of two pathogens in CoFSs based on the spectral characteristics of the entire plate. The developed model accurately predicted the mycelial areas of pathogens growing in CoFS samples with an error smaller than 0.2. The effectiveness of the method proposed depended on the degree of pathogen development. The best results were obtained 22 days from the start of the experiment, when the sums of the mycelial areas of the CoFSs placed next to each other and at a distance of 4 cm were 46% and 77% of the plate area, respectively.
The positive results of the study encourage the continuation of research focused on the discrimination of more than two co-occurring fungi using reflectance spectroscopy because this is a fast, cheap, and robust method.

Author Contributions

J.P., A.W. and M.W.: Conceptualization, Investigation, Writing—Original Draft Preparation, Writing—Review and Editing; S.K. and E.S.-K.: Data Curation, Resources, Investigation; J.J.: Methodology, Visualization, Writing—Original Draft Preparation. All authors have read and agreed to the published version of the manuscript.

Funding

This research received no external funding.

Institutional Review Board Statement

Not applicable.

Informed Consent Statement

Not applicable.

Data Availability Statement

The data presented in this study are available on request from the corresponding author.

Conflicts of Interest

The authors declare no conflict of interest.

References

  1. Yu, H.; Cheng, X.; Chen, C.; Heidari, A.A.; Liu, J.; Cai, Z.; Chen, H. Apple leaf disease recognition method with improved residual network. Multimed. Tools. Appl. 2022, 81, 7759–7782. [Google Scholar] [CrossRef]
  2. Camino, C.; Araño, K.; Berni, J.A.; Dierkes, H.; Trapero-Casas, J.L.; León-Ropero, G.; Montes-Borrego, M.; Roman-Écija, M.; Velasco-Amo, M.P.; Landa, B.B.; et al. Detecting Xylella fastidiosa in a machine learning framework using Vcmax and leaf biochemistry quantified with airborne hyperspectral imagery. Remote Sens. Environ. 2022, 282, 113281. [Google Scholar] [CrossRef]
  3. Moskovsky, M.N.; Shogenov, Y.H.; Lavrov, A.V.; Gulyaev, A.A.; Efremenkov, I.Y.; Pyatchenkov, D.S.; Belyakov, M.V. Spectral Photoluminescent Parameters of Barley Seeds (Hordéum vulgare) Infected with Fusarium ssp. Photochem Photobiol. 2023, 99, 29–34. [Google Scholar] [CrossRef] [PubMed]
  4. Martinelli, F.; Scalenghe, R.; Davino, S.; Panno, S.; Scuderi, G.; Ruisi, P.; Villa, P.; Stroppiana, D.; Boschetti, M.; Goulart, L.; et al. Advanced methods of plant disease detection. A review. Agron. Sustain. Dev. 2014, 35, 1–25. [Google Scholar] [CrossRef]
  5. Gorretta, N.; Nouri, M.; Herrero, A.; Gowen, A.; Roger, J.-M. Early detection of the fungal disease “apple scab” using SWIR hyperspectral imaging. In Proceedings of the 2019 10th Workshop on Hyperspectral Imaging and Signal Processing: Evolution in Remote Sensing (WHISPERS), Amsterdam, The Netherlands, 24–26 September 2019; pp. 1–4. [Google Scholar]
  6. He, L.; Qi, S.-L.; Duan, J.-Z.; Guo, T.-C.; Feng, W.; He, D.-X. Monitoring of Wheat Powdery Mildew Disease Severity Using Multiangle Hyperspectral Remote Sensing. IEEE Trans. Geosci. Remote Sens. 2021, 59, 979–990. [Google Scholar] [CrossRef]
  7. Francesconi, S. High-throughput and point-of-care detection of wheat fungal diseases: Potentialities of molecular and phenomics techniques toward in-field applicability. Front. Agron. 2022, 4, 980083. [Google Scholar] [CrossRef]
  8. Kasalica, B.V.; Miletic, K.M.; Sabovljevic, A.D.; Vujicic, M.M.; Jeremic, D.A.; Belca, I.D.; Petkovic-Benazzouz, M.M. Nondestructive optical method for plant overall health evaluation. Acta Agric. Scand. B Soil Plant Sci. 2021, 71, 1017–1023. [Google Scholar] [CrossRef]
  9. Calderón Madrid, R.; Navas Cortés, J.A.; Lucena León, C.; Zarco-Tejada, P.J. High-resolution hyperspectral and thermal imagery acquired from UAV platforms for early detection of Verticillium wilt using fluorescence, temperature and narrow-band indices. In Proceedings of the Workshop on UAV-Basaed Remote Sensing Methods for Monitoring Vegetation, Cologne, Germany, 11–12 September 2013; pp. 7–14. [Google Scholar]
  10. Oerke, E.C.; Leucker, M.; Steiner, U. Sensory assessment of Cercospora beticola sporulation for phenotyping the partial disease resistance of sugar beet genotypes. Plant Methods 2019, 15, 133. [Google Scholar] [CrossRef]
  11. Guo, A.; Huang, W.; Ye, H.; Dong, Y.; Ma, H.; Ren, Y.; Ruan, C. Identification of wheat yellow rust using spectral and texture features of hyperspectral images. Remote Sens. 2020, 12, 1419. [Google Scholar] [CrossRef]
  12. Chen, N.; Liu, F.; Jiang, L.; Feng, L.; He, Y.; Bao, Y. Diagnosis of Sclerotinia Infected Oilseed Rape (Brassica napus L) Using Hyperspectral Imaging and Chemomtrics. 2022. Available online: https://www.ispag.org/proceedings/ (accessed on 27 October 2023).
  13. Pereira, F.V.; Martins, G.D.; Vieira, B.S.; de Assis, G.A.; Orlando, V.S.W. Multispectral images for monitoring the physiological parameters of coffee plants under different treatments against nematodes. Precis. Agric. 2022, 23, 2312–2344. [Google Scholar] [CrossRef]
  14. Cubero, S.; Marco-Noales, E.; Aleixos, N.; Barbé, S.; Blasco, J. Robhortic: A field robot to detect pests and diseases in horticultural crops by proximal sensing. Agriculture 2020, 10, 276. [Google Scholar] [CrossRef]
  15. Osuna-Caballero, S.; Olivoto, T.; Jiménez-Vaquero, M.A.; Rubiales, D.; Rispail, N. RGB image-based method for phenotyping rust disease progress in pea leaves using R. Plant Methods 2023, 19, 86. [Google Scholar] [CrossRef]
  16. Pavicic, M.; Overmyer, K.; Rehman, A.U.; Jones, P.; Jacobson, D.; Himanen, K. Image-based methods to score fungal pathogen symptom progression and severity in excised Arabidopsis leaves. Plants 2021, 10, 158. [Google Scholar] [CrossRef] [PubMed]
  17. Long, T.; Ziyi, W.; Bowen, X.; Dong, L.; Hengbiao, Z.; Xia, Y.; Yan, Z.; Weixing, C.; Tao, C. A disease-specific spectral index tracks Magnaporthe oryzae infection in paddy rice from ground to space. Remote Sens. Environ. 2023, 285, 113384. [Google Scholar]
  18. Croeser, L. The Role of Phytophthora in Predisposing Corymbia Calophylla (marri) to a Canker Disease. Ph.D. Thesis, Murdoch University, Perth, Australia, 2021. [Google Scholar]
  19. Piekarczyk, J.; Ratajkiewicz, H.; Jasiewicz, J.; Sosnowska, D.; Wójtowicz, A. An application of reflectance spectroscopy to differentiate of entomopathogenic fungi species. J. Photochem. Photobiol. B Biol. 2019, 190, 32–41. [Google Scholar] [CrossRef] [PubMed]
  20. Aboelghar, M.; Wahab, H.A. Spectral footprint of Botrytis cinerea, a novel way for fungal characterization. Adv. Biosci. Biotechnol. 2013, 4, 374–382. [Google Scholar] [CrossRef]
  21. Piekarczyk, J.; Wójtowicz, A.; Wójtowicz, M.; Jasiewicz, J.; Sadowska, K.; Łukaszewska-Skrzypniak, N.; Świerczyńska, I.; Pieczul, K. Machine Learning-Based Hyperspectral and RGB Discrimination of Three Polyphagous Fungi Species Grown on Culture Media. Agronomy 2022, 12, 1965. [Google Scholar] [CrossRef]
  22. Aboelghar, M.; Moustafa, M.S.; Ali, A.M.; Wahab, H.A. Hyperspectral analysis of Botrytis cinerea infected lettuce. EPH-Int. J. Agric. Environ. Res. 2019, 5, 26–42. [Google Scholar]
  23. Shoemaker, R.A.; Brun, H. The teleomorph of the weakly aggressive segregate of Leptosphaeria maculans. Canad. J. Bot. 2001, 79, 412–419. [Google Scholar] [CrossRef]
  24. Fitt, B.D.L.; Huang, Y.J.; van den Bosch, F.; West, J.S. Coexistence of related pathogen species on arable crops in space and time. Annu. Rev. Phytopathol. 2006, 44, 163–182. [Google Scholar] [CrossRef]
  25. Gaetan, S.A. First outbreak of blackleg caused by Phoma lingam in commercial canola fields in Argentina. Plant Dis. 2005, 89, 435. [Google Scholar] [CrossRef] [PubMed]
  26. Fitt, B.D.L.; Brun, H.; Barbetti, M.J.; Rimmer, S.R. World-wide importance of phoma stem canker (Leptosphaeria maculans and L. biglobosa) on oilseed rape (Brassica napus). Eur. J. Plant Pathol. 2006, 114, 3–15. [Google Scholar] [CrossRef]
  27. West, J.S.; Kharbanda, P.D.; Barbetti, M.J.; Fitt, B.D.L. Epidemiology and management of Leptosphaeria maculans (phoma stem canker) on oilseed rape in Australia, Canada and Europe. Plant Pathol. 2001, 50, 10–27. [Google Scholar] [CrossRef]
  28. Kuswinanti, T.; Koopmann, B.; Hoppe, H.H. Virulence pattern of aggressive isolates of Leptosphaeria maculans on an extended set of Brassica differentials. J. Plant Dis. Prot. 1999, 106, 12–20. [Google Scholar]
  29. Kuusk, A.K.; Happstadius, I.; Zhou, L.; Steventon, L.A.; Giese, H.; Dixelius, C. Presence of Leptosphaeria maculans group A and group B isolates in Sweden. J. Phytopathol. 2002, 150, 349–356. [Google Scholar] [CrossRef]
  30. Bagi, B.; Nagy, C.; Tóth, A.; Palkovics, L.; Petróczy, M. Plenodomus biglobosus on oilseed rape in Hungary. Phytopathol. Mediterr. 2020, 59, 345–351. [Google Scholar]
  31. Mazáková, J.; Urban, J.; Zouhar, M.; Rysanek, P. Analysis of Leptosphaeria species complex causing phoma leaf spot and stem canker of winter oilseed rape (Brassica napus) in the Czech Republic. Crop. Pasture Sci. 2017, 68, 254–264. [Google Scholar] [CrossRef]
  32. Brazauskiene, I.; Piliponyte, A.; Petraitiene, E.; Brazauskas, G. Diversity of Leptosphaeria aculans/L. biglobosa species complex and epidemiology of phoma stem cancer on oilseed rape in Lithuania. Plant Pathol. J. 2011, 93, 577–585. [Google Scholar]
  33. Gil-Lamaignere, C.; Roilides, E.; Hacker, J.; Müller, F.-M.C. Molecular typing for fungi—A critical review of the possibilities and limitations of currently and future methods. Clin. Microbiol. Infect. 2003, 9, 172–185. [Google Scholar] [CrossRef]
  34. Huleihel, M.; Shufan, E.; Tsror, L.; Sharaha, U.; Lapidot, I.; Mordechai, S.; Salman, A. Differentiation of mixed soil-borne fungi in the genus level using infrared spectroscopy and multivariate analysis. J. Photochem. Photobiol. B Biol. 2018, 180, 155–165. [Google Scholar] [CrossRef]
  35. Salman, A.; Shufan, E.; Lapidot, I.; Tsror, L.; Mordechai, S.; Sharaha, U.; Huleihel, M. Infrared spectroscopy and multivariate analysis: Classification of mixed fusarium species solani and oxysporum isolates at the species level. In Proceedings of the 2016 IEEE International Conference on the Science of Electrical Engineering (ICSEE), Eilat, Israel, 16–18 November 2016; pp. 1–5. [Google Scholar]
  36. Xian, W.; Božič, A.; Snavely, N.; Lassner, C. Neural Lens Modeling. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, Vancouver, BC, Canada, 17–24 June 2023; pp. 8435–8445. [Google Scholar] [CrossRef]
  37. Mueller, D.S.; Dorrance, A.E.; Derksen, R.C.; Ozkan, E.; Kurle, J.E.; Grau, C.R.; Gaska, J.M.; Hartman, G.L.; Bradley, C.A.; Pedersen, W.L. Efficacy of fungicides on Sclerotinia sclerotiorum and their potential for control of Sclerotinia stem rot on soybean. Plant Dis. 2002, 86, 26–31. [Google Scholar] [CrossRef] [PubMed]
  38. Sivakumar, D.; Wijeratnam, W.R.S.; Wijesundera, R.L.C.; Marikar, F.M.T.; Abeyesekere, M. Antagonistic effect of Trichoderma harzianum on post-harvest pathogens of Rambutan (Naphelium lappaceum). Phytoparasitica 2000, 28, 240–247. [Google Scholar] [CrossRef]
  39. Gabriel, K.R. The biplot graphic display of matrices with application to principal component analysis. Biometrika 1971, 58, 453–467. [Google Scholar] [CrossRef]
  40. Breiman, L. Random Forests. Mach. Learn. 2001, 45, 5–32. [Google Scholar] [CrossRef]
  41. Chen, T.; Guestrin, C. Xgboost: A scalable tree boosting system. In Proceedings of the 22nd ACM Sigkdd International Conference on Knowledge Discovery and Data Mining, San Francisco, CA, USA, 13–17 August 2016; pp. 785–794. [Google Scholar]
  42. Zou, H.; Hastie, T. Regularization and variable selection via the elastic net. J. R. Stat. Soc. Ser. B Stat. Methodol. 2005, 67, 301–320. [Google Scholar] [CrossRef]
  43. Fix, E.; Hodges, J.L., Jr. Discriminatory Analysis—Nonparametric Discrimination: Small Sample Performance; University of California: Berkeley, CA, USA, 1952. [Google Scholar]
  44. Ludwig, B.; Murugan, R.; Parama, V.R.; Vohland, M. Use of different chemometric approaches for an estimation of soil properties at field scale with near infrared spectroscopy. J. Plant Nutr. Soil Sci. 2018, 181, 704–713. [Google Scholar] [CrossRef]
  45. Helland, I.S. PLS regression and statistical models. Scand. J. Stat. 1990, 17, 97–114. [Google Scholar]
  46. Newbery, F.; Ritchie, F.; Gladders, P.; Fitt, B.D.L.; Shaw, M.W. Inter-individual genetic variation in the temperature response of Leptosphaeria species pathogenic on oilseed rape. Plant Pathol. 2020, 69, 1469–1481. [Google Scholar] [CrossRef]
  47. Jędryczka, M. Epidemiologia i szkodliwość suchej zgnilizny kapustnych na rzepaku ozimym w Polsce. Rozpr. I Monogr. IGR PAN. 2006, 17, 1–150. (In Polish) [Google Scholar]
  48. McGee, D.; Petrie, G. Variability of Leptosphaeria maculans in relation to blackleg of oilseed rape. Phytopathology 1978, 68, 625–630. [Google Scholar] [CrossRef]
  49. Williams, R.H.; Fitt, B.D.L. Differentiating A and B groups of Leptosphaeria maculans, causal agent of stem canker (blackleg) of oilseed rape. Plant Pathol. 1999, 48, 161–175. [Google Scholar] [CrossRef]
  50. Chu, X.; Chen, J.; Tang, Y.; Gao, S.; Zhuang, J.; Luo, S. Evaluating Growth of Colletotrichum species by Near infrared (NIR) hyperspectral imaging. IFAC-Pap. 2019, 52, 257–262. [Google Scholar] [CrossRef]
Agriculture 13 02228 g001
Figure 1. Fungal mycelia classification and unmixing scheme.
Figure 1. Fungal mycelia classification and unmixing scheme.
Agriculture 13 02228 g001
Agriculture 13 02228 g002
Figure 2. Development of fungal mycelia in four experimental combinations.
Figure 2. Development of fungal mycelia in four experimental combinations.
Agriculture 13 02228 g002
Agriculture 13 02228 g003
Figure 3. Temporal and spatial changes in the mycelial areas of P. lingam and P. biglobosus in pure (PFS) and co-occurring (CoFS) fungal samples.
Figure 3. Temporal and spatial changes in the mycelial areas of P. lingam and P. biglobosus in pure (PFS) and co-occurring (CoFS) fungal samples.
Agriculture 13 02228 g003
Agriculture 13 02228 g004
Figure 4. Mean spectra of P. lingam and P. biglobosus in PFSs and CoFSs registered on eight dates.
Figure 4. Mean spectra of P. lingam and P. biglobosus in PFSs and CoFSs registered on eight dates.
Agriculture 13 02228 g004
Agriculture 13 02228 g005
Figure 5. The wavelengths most useful for distinguishing experimental combinations and results of spectral classification of fungal mycelia.
Figure 5. The wavelengths most useful for distinguishing experimental combinations and results of spectral classification of fungal mycelia.
Agriculture 13 02228 g005
Agriculture 13 02228 g006
Figure 6. Relationship between the area ratios of P. biglobosus and P. lingam on the training (blue) and testing (orange) sets using the PLSR model.
Figure 6. Relationship between the area ratios of P. biglobosus and P. lingam on the training (blue) and testing (orange) sets using the PLSR model.
Agriculture 13 02228 g006
Table 1. Experimental design.
Table 1. Experimental design.
NumberExperimental CombinationPure/Co-Occurring Samples
1P. lingamPure fungal samples
2P. biglobosusPure fungal samples
3P. lingam and P. biglobosus together on the same plate next to each otherCo-occurring fungal samples
4P. lingam and P. biglobosus together on the same plate at a distance of 4 cm from one anotherCo-occurring fungal samples
Table 2. Areas under mycelial growth curves (AUMGCs) for P. lingam and P. biglobosus and their ratios.
Table 2. Areas under mycelial growth curves (AUMGCs) for P. lingam and P. biglobosus and their ratios.
Days after InoculationPure Fungal Samples (PFSs)Co-Occurring Fungal Samples (CoFSs)
1st Comb.2nd Comb. 3rd Comb.4th Comb.
P. ling.P. biglo.Ratio
P. biglo./P. ling.		P. ling.P. biglo.Ratio
P. biglo./P. ling.		P. ling.P. biglo.Ratio
P. biglo./P. ling.
10.30.30.90.30.20.90.20.21.0
31.92.31.21.91.20.62.01.70.9
824.346.81.97.815.52.042.050.61.2
1045.090.42.010.929.82.779.994.11.2
15143.6303.32.127.6102.03.7217.2258.61.2
17205.6430.22.138.4149.43.9279.2342.51.2
22423.6814.71.977.6307.74.0437.3560.91.3
24529.2988.01.998.7384.73.9505.7649.21.3
Table 3. Root mean square errors (RMSEs) obtained in evaluation of five regressors used for spectral unmixing of fungal mycelia.
Table 3. Root mean square errors (RMSEs) obtained in evaluation of five regressors used for spectral unmixing of fungal mycelia.
Days after InoculationPLSRXGBRFENKNN
11.2340.6760.6870.670.769
31.8911.6511.6561.8231.634
81.0140.550.5250.4950.707
100.9730.7070.7180.5140.561
150.5090.3210.3570.3680.463
170.5020.2790.3180.3230.392
220.1740.2570.2560.3390.343
240.1990.270.260.3780.25
Disclaimer/Publisher’s Note: The statements, opinions and data contained in all publications are solely those of the individual author(s) and contributor(s) and not of MDPI and/or the editor(s). MDPI and/or the editor(s) disclaim responsibility for any injury to people or property resulting from any ideas, methods, instructions or products referred to in the content.

Share and Cite

MDPI and ACS Style

Wójtowicz, A.; Piekarczyk, J.; Wójtowicz, M.; Jasiewicz, J.; Królewicz, S.; Starzycka-Korbas, E. Classification of Plenodomus lingam and Plenodomus biglobosus in Co-Occurring Samples Using Reflectance Spectroscopy. Agriculture 2023, 13, 2228. https://doi.org/10.3390/agriculture13122228

AMA Style

Wójtowicz A, Piekarczyk J, Wójtowicz M, Jasiewicz J, Królewicz S, Starzycka-Korbas E. Classification of Plenodomus lingam and Plenodomus biglobosus in Co-Occurring Samples Using Reflectance Spectroscopy. Agriculture. 2023; 13(12):2228. https://doi.org/10.3390/agriculture13122228

Chicago/Turabian Style

Wójtowicz, Andrzej, Jan Piekarczyk, Marek Wójtowicz, Jarosław Jasiewicz, Sławomir Królewicz, and Elżbieta Starzycka-Korbas. 2023. "Classification of Plenodomus lingam and Plenodomus biglobosus in Co-Occurring Samples Using Reflectance Spectroscopy" Agriculture 13, no. 12: 2228. https://doi.org/10.3390/agriculture13122228

APA Style

Wójtowicz, A., Piekarczyk, J., Wójtowicz, M., Jasiewicz, J., Królewicz, S., & Starzycka-Korbas, E. (2023). Classification of Plenodomus lingam and Plenodomus biglobosus in Co-Occurring Samples Using Reflectance Spectroscopy. Agriculture, 13(12), 2228. https://doi.org/10.3390/agriculture13122228

Note that from the first issue of 2016, this journal uses article numbers instead of page numbers. See further details here.

Article Metrics

Back to TopTop