Using Fluorescence Spectroscopy to Detect Rot in Fruit and Vegetable Crops

Abstract

The potential of the method of fluorescence spectroscopy for the detection of damage and diseases of fruits and vegetables was studied. For this purpose, the spectra of fluorescence of healthy and rotten apples and potatoes have been investigated. Excitation of samples was carried out using a continuous semiconductor laser with a wavelength of 405 nm and a pulsed solid-state laser with a wavelength of 527 nm. Peaks in the region of 600–700 nm in rotten samples were shifted towards shorter wavelengths for most samples in both modes of spectroscopy. The differences in the fluorescence spectra of a healthy and rotten apple surface have been revealed to be in the spectral range of 550–650 nm for 405 nm continuous excitation. When exposed to a laser in a pulsed mode (527 nm), the contribution of the 630 nm peak in the spectrum increases in rotten samples. The observed differences make it possible to use this method for separating samples of healthy and rotten fruits and vegetables. The article paid attention to the influence of many factors such as sample thickness, time after excitation, contamination by soil and dust, cultivar, and location of the probing on fluorescence spectra.

1. Introduction

Population growth opens new challenges for improving efficiency and productivity of the agriculture and food production industry, which can be solved with the development of new sensing technologies capable of providing in vivo or/and onsite monitoring. One of the high-tech solutions in this field is fluorescence spectroscopy. Fluorescence is used to test authenticity of different categories of food products [1,2,3], analyze the quality of fruits [4,5], diagnose of plant diseases [6,7], and obtain information about the influence of environmental stress factors on agricultural crops [6,8,9,10,11].

The kinetics of the development of apple disease at the early stage of fungal infection is studied under laboratory conditions using chlorophyll fluorescence [12]. In addition, fluorescence under laboratory conditions was used to detect patulin toxin in apples [13] and diagnose fungal diseases or substances harmful to humans in potato tubers [14,15].

On the other hand, the development of inexpensive, sensitive methods for detection of fluorescence leads to the development of portable devices that can diagnose the content of chlorophyll and some other compounds in fruits and leaves with accuracy comparable to laboratory studies [16].

The study of fluorescence spectra of the main diseases of fruit and vegetable crops are of interest not only from a scientific point of view, but also for solving many applied problems in agriculture. Therefore, the main goal of the work is the application of inexpensive methods of fluorescence spectroscopy for rot disease diagnosis and separation/quantification of healthy and diseased fruits and vegetables.

2. Materials and Methods

Fluorescence spectroscopy using laser sources was used for the detection of apple and potato rot infection. Two radiation sources and various spectrometers, as well as various samples, were used for the study.

2.1. Experiments with a 405 nm Continuous Laser

The first experimental setup consisted of a tripod with a vertical screen for fixing the test sample, a Laserland 1668-405D-20-5V laser (405 nm wavelength, 20 mW output), a L20 light filter (3 mm thickness, 350–450 nm bandwidth), a collimator with an optical fiber, and an Ocean Optics USB4000 spectrometer connected to a PC (Figure 1). The laser spot size was 4 mm. Integration time was 1000 ms.

2.2. Preparation and Measurement of Vegetables and Fruits in Experiments with a Continuous Laser

The apple cultivar “President” and the potato cultivar “Kumach” were used in experiments with the continuous laser mode (Figure 2). Skin surface samples of apples and potatoes ~0.5–0.8 mm thick and ~1 × 1 cm size were cut from whole fruits and vegetables. The pulp samples were prepared in the same way. The apple or potato sample was placed on the surface of the vertical screen in the absence of external illumination (number 4, Figure 1). The fluorescence spectrum was measured three times at different points on the samples. The results on the graphs are presented as an average of two apples/potatoes, each of which has been measured three times. A standard deviation was drawn as an error on the figures. The spectrum kinetics was recorded for 10 min with rate of 1 frame/s.

2.3. Experiments with a 527 nm Pulsed Laser

A compact Raman lidar was used to record the fluorescence spectra in the second part of experiment. Lidar is based on a diode-pumped pulsed Nd:YLiF4 solid-state laser (Laser Compact, DTL-319QT: 527 nm, 5 ns, 1 kHz, 200 µJ/pulse) [17]. The laser beam was directed to the sample using two rotating prisms. The laser spot diameter was 0.5 mm (measured by burn paper). The radiation scattered from the object was collected using a quartz lens. The signal recording system consists of the compact spectrograph (Spectra Physics, MS127i) (Andor Technology Ltd., Belfast, Northern Ireland) (with an optical resolution of 0.22 nm) equipped with the intensified CCD (ICCD) camera (Andor iStar) (Andor Technology Ltd., Belfast, Northern Ireland). To increase the signal-to-noise ratio, a diffraction grating with a small dispersion (400 lines/mm) and a 250 µm micron wide input slit were used. In order to avoid damage of the detector by scattered laser radiation, a band-pass light filter OS-13 was used for signal registration (Figure 3). The use of Lidar does not require sample darkening from external illumination since a pulsed source and a gated detector are used; however, this is necessary for a continuous laser. All gating was performed with the help of a built-in pulse generator in an ICCD camera. Chlorophyll fluorescence delay depends on different environments, but is generally less than 3 ns for in situ measurements [18]. Consequently, the detection gate was set to 10 ns in order to fit the pumping laser pulse duration, fluorescence delay, and jitter of gating (3 ns). The delay between the laser pulse and detection gate was varied in order to maximize backscattered elastic scattering; fluorescence spectrum intensity was also maximized (due to chosen gate duration).

2.4. Preparation and Measurement of Vegetables and Fruits in Experiments Pulsed Laser Mode

Two apple cultivars and potato cultivars “Udacha” (healthy) and “Zhukovsky” (rotten) were used in the experiments with the pulsed laser. All apple and potato samples were provided by the Federal Scientific Agroengineering Center VIM (Moscow) (Figure 4).

Initially, the fluorescence spectra of unwashed potatoes were recorded. The spectrum was averaged from the entire potato surface. Next, the potatoes were thoroughly washed from dust and soil residues without damaging the peel. The fluorescence spectra of washed and unwashed potatoes were compared to assess the contribution of dust and soil to the overall spectrum. Moreover, the fluorescence spectra for different zones of the rotten potatoes were recorded. Samples of potatoes were cut in half to examine the internal parts of the potatoes. Spectra were recorded in various tuber regions, the area in the center of the potato (clearly visible to the eye), and the main volume of the potato.

The fluorescence spectra of outer parts of healthy and rotting apples were registered. Apples were cut and the spectra of the spoiled pulp and the healthy samples were compared.

3. Results and Discussion

3.1. Experiments with a 405 nm Continuous Laser

Figure 5 shows the spectrum of fluorescence of the surface (left) and pulp (right) of a healthy and rotten apple. Two peaks can be identified on the spectra: in the green (520–530 nm) and in the red (680 nm) regions. Peaks in the blue-green fluorescence region (440–540 nm) are mainly associated with cinnamic and coumarin acids, some alkaloids and flavonoids [19]. The latter are usually located in the cell wall [19] and perform mainly antioxidant functions and protect against fungal and viral diseases [20].

The peak in the red area is well described in many studies; it is the peak of chlorophyll. Previously, for green leaves, it was shown that the ratio of the intensities of the peaks 680/740 depends on the concentration of chlorophyll [21]. The existence of a peak at 680 nm and the absence of a peak at an excitation of 405 nm at 740 nm confirms the presence of a small amount of chlorophyll in apple samples.

In Figure 5, the spectral differences between healthy and rot-affected fruits are visible. In samples affected by rotting, the peak of the maximum of chlorophyll is shifted from 680 nm to 675.5 nm, indicating a decrease in chlorophyll concentration. The second difference in the surface is the “plateau” between peaks in green (520 nm) and red (680 nm) regions. Healthy samples do not show a significant decrease in intensity between peaks. This plateau in the yellow-orange region (550–650 nm) is most likely due to the presence of riboflavin, carotenoids, and flavonoids [19]. The differences are even more pronounced for the apple pulp (Figure 5b). The fluorescence amplitude of the apple pulp affected by rot is reliably much smaller than the healthy specimens in the whole investigated region.

Healthy and rotten potato surface is less different in fluorescence spectra in contrast to apple. The peak of chlorophyll (680 nm) for healthy potatoes is less pronounced than for the rotten samples, and the peaks in the green region (530 nm) are the same (Figure 6). However, there are reliable differences in the pulp. The spectrum of the rotten sample goes beyond the sensitivity of the spectrometer in the green area (Figure 6b). This can be explained by the assumption that rotten potato pulp absorbs laser radiation much more efficiently, which leads to increased fluorescence. The efficiency of fluorescence quenching is rather high for the healthy plant, so any damage of the plant material should result in a change of its chemical composition; thus, this will decrease the quenching efficiency and hence the induced fluorescence intensity should be increased. Additionally, bacteria and fungi in rotten potato had high fluorescence, so with the increase of such organisms’ concentration, the fluorescence will be higher.

An increase in fluorescence occurs when the thickness of healthy potato pulp changes (Figure 6b, top right). The amplitude of the fluorescence spectrum increases 2.5–3 times with an increase in the slice thickness from 0.5 to 2 mm. Intensity of fluorescence is not linearly proportional to the sample thickness.

Figure 7 shows the kinetics of the decrease in fluorescence over time. The observed effect is similar, with the well-known Kautsky effect observed for the chlorophyll peak in green leaves [22]. However, there are differences; the Kautsky effect occurs in leaves previously placed in the dark, and times of growth of the fluorescence are ~300–500 ms, and decay times are ~1–10 s, although in some cases the fluorescence decay takes ~1–10 min [21,23]. Such kinetics (Figure 7b), where chlorophyll fluorescence decreases with a time of 1–10 min, has been described by many authors [1,24], and may be associated with several processes: changes in the uptake of photosynthetic complex II during long illumination, changes in the rate of carbon metabolism and the release of oxygen by cells, and due to the fact that oxygen in air can probably receive electrons from the photosynthetic complexes I and II [25]. The presence of kinetic dependencies and the dependence of fluorescence on the lighting regime impose some restrictions on the use of this method in the detection of fruit and vegetable diseases.

3.2. Experiments with a 527 nm Pulsed Laser

In experiments with a 527 nm pulsed laser, fluorescence spectra were obtained in the 560–720 nm range. In this spectral range, chlorophyll peaks at 680 nm were clearly visible, especially in green apples.

The rotten apple is easily distinguishable by the fluorescence spectra from the healthy apple (Figure 8). The intensities of fluorescence are increased, and fluorescence peaks are shifted to shorter wavelengths for rotten apple. Interestingly, controversial influence of rot on laser-induced fluorescence spectra can be observed while comparing Figure 5b and Figure 8c. In the case of 527 nm pumping, the fluorescence signal is also increased for rot as expected, but for 405 nm, the excitation situation is the opposite. This can be explained by the difference in absorption coefficient for healthy and rotten apple surfaces that resulted in different sampling depths. For example, a penetration depth of 405 nm is smaller compared to 527 nm radiation, so fluorescence spectra will originate from different layer thicknesses. Typically, any damage of plant material will increase the intensity of fluorescence spectra due to the decreased efficiency of quenching. In the case of 527 nm pumping, the penetration depth is significantly larger, so apple rotting will result in an increase of fluorescence signal. Alternatively, the 405 nm wavelength had a small penetration depth that was further increased for the rotten apple surface due to the increased absorption coefficient of the decomposed organic material.

In the case of the green apples (Figure 9), the intensity of fluorescence is conversely dramatically decreased, although fluorescence peaks are shifted to shorter wavelengths in the same manner for rotten red apple. The maximal shift is observed for the rotten pulp. Intensity of fluorescence is decreased, and contribution of the 630 nm peak is increased in the following order: healthy surface, rotten surface, rotten pulp.

Significant differences between emission spectra of healthy and scab-affected apple samples were also found in the region of 680–750 nm in [26], where a conventional serial fluorescence spectrometer with a xenon lamp as the radiation source was used.

The fluorescence spectra of washed and unwashed potatoes were compared to assess the contribution of dust and soil to the overall spectrum (Figure 10a). The overall intensity is higher for washed potatoes, and the spectrum of the unwashed sample is noisier. However, the envelope of the spectrum is the same in both cases. The contamination does not distort the spectrum and it is possible to analyze directly unwashed potatoes in the future in field experiments.

The spectra of whole potatoes of different potato samples are quite different from each other (Figure 10b). There is a significant shift of chlorophyll peak to the region of shorter wavelengths for the rotten sample. The 680 nm peak is more pronounced than the 630 nm peak for healthy potato samples of the “Udacha” cultivar.

Pulp spectra also demonstrate differences depending on cultivar (Figure 11). The fluorescence spectra of the central “vein” (point 1) and the main volume (point 2) differ for “Udacha” potatoes. The spectra are almost identical for the “Zhukovsky” potato.

Since the rotten potatoes were damaged in various ways, the fluorescence spectra were recorded for each type of defect (Figure 12). Points 1 and 2 were taken at the places of rot on the peel. Point 3 corresponds to the thickened part of the peel, and point 4 corresponds to the thin part. Point 5 corresponds to the potato eye. The spectra vary for different defects.

The following pros and cons of using pulsed and continuous lasers for detecting the fluorescence spectra of vegetables and fruits can be noted. Fluorescence spectroscopy with continuous laser pumping has several positive features, including low-cost equipment, and low power density preventing possible sample damage. Among the negative aspects, a single one can be highlighted; over time, the sample “burns out”—the method is sensitive to external light and geometry, and as a result method is not very suitable for field conditions. Alternatively, fluorescence spectroscopy measurements with the pulsed laser have its own advantages, including that external light influence can be easily skipped by gated measurements including remote sensing with the appropriate gating so distant targets can be sensed. For example, in vivo pulsed fluorescence spectroscopy measurements of plants can be done onsite in the field during different daylight conditions; additionally, long time period measurements can be conducted. The negative aspects of a pulsed laser-induced fluorescence spectroscopy include the potential damage of biological structures induced by high peak power density, and the high cost of equipment.

4. Conclusions

The observed differences (Figure 5, Figure 6, Figure 7, Figure 8, Figure 9, Figure 10, Figure 11 and Figure 12) allow the use of fluorescent spectroscopy for the separation of healthy and rotting fruit and vegetable samples. Its application is based on changes in the fluorescence of plant fluorophores, mainly chlorophyll. During rotting, a shift of chlorophyll peaks to the region of shorter wavelengths and a decrease in the ratio of the intensities of the peaks 680/630 was observed.

Both pulse and continuous modes of light source operation have shown the potential for scanning rotten fruits and vegetables, thus opening up the possibility of using various fluorescence-based devices for the detection of plant diseases.

The revealed variety of spectra showed that it is necessary to consider the characteristics of the cultivars, as well as the probe areas when selecting the parameters of the fluorescence analysis for detecting rot.

The practical aspect of the approach is automation of the process of crop sorting. We hope it could be used in the future by implementation in real conditions, on the field or on sorting lines.