2.1.

Materials

Four kinds of tropical fruits were selected for the experimental measurements: nectarine (NE), mango (MG), papaya (PA), and guava (GU). All these fruits belong to a class referred to as climacteric fruits.¹⁸ The fruits were acquired from a local market in South China, and visually nonripe samples were chosen. Each type of fruit was represented by a group of six fruits from the same batch. Representative fruits are depicted in Fig. 1(a). Each fruit was marked with a prefix and a number (e.g., NE1, NE2 etc.). This identification mark was written on a small piece of paper, which was attached to the fruit far away from the measurement location, to avoid interaction with the skin surface to be investigated. In the fluorescence monitoring, possible photobleaching had to be considered. If photobleaching occurs, the intensity of the fluorescence will be gradually reduced.¹⁹ In order to fully eliminate residual effects of photobleaching, reflectance and fluorescence measurements were performed on the fruit, successively employing new locations as marked in the left side of Fig. 1(b). A $15\times 15\text{-}{\mathrm{mm}}^2$ area on the “equator” was selected, which appeared completely uniform with little structure on visual inspection. Within the area, nine individual measurement spots were selected. The locations for reflectance and fluorescence measurements were shifted every day to avoid “memory” from earlier measurements. The right-hand part of Fig. 1(b) shows the GASMAS injection and detection locations. It was ascertained that the distance from the injection fiber to the center of the detection was kept constant. The six fruits were divided into two groups, among which one group (such as, NE1, NE2, and NE3) was placed in an ambient temperature of 20°C and another group (such as, NE4, NE5, and NE6) at a temperature of 30°C.

Fig. 1

The four kinds of fruits studied in the experiments (a) and a schematic drawing of the measurement locations on the fruit skin for reflectance and fluorescence measurements, and for the gas in scattering media absorption spectroscopy (GASMAS) probing. (NE1 is short for nectarine #1) (b).

2.2.

Methods

2.2.1.

Broadband light reflectance spectroscopy

Schematic drawings of the experimental setups used in the study are shown in Fig. 2. The setup illustrated in Fig. 2(a) was used for measurements of the reflectance spectra. Broadband light from a tungsten lamp with 130 mW output was guided to the sample with a fiber. Reflected light was captured with a reflectance probe and then transmitted to the slit of a portable miniature spectrometer (USB4000, Ocean Optics, Dunedin, Florida). Before the measurements, the system was calibrated with regard to intensity using a calibrated lamp (IES 1000, Labsphere, North Sutton, New Hampshire) and regarding wavelength using a low-pressure mercury lamp.

Fig. 2

Experimental arrangement for the measurements of reflectance spectra as well as the fluorescence spectra (a) and for the monitoring of the second-harmonic wavelength modulation spectroscopy (2f WMS) signal of oxygen using GASMAS (b), TIA is a transimpedance amplifier.

3.1.

Reflectance and Fluorescence Measurements

Before the spectroscopic measurements, the system was calibrated for intensity using a calibration lamp, as mentioned above. Spectral reflectance curves obtained from different stages of ripeness for each kind of fruit (NE1, MG6, GU2, and PA6) are shown in Fig. 3 (top). We observed a strong absorption at 670 to 690 nm as evidenced from the spectral shape. This is the absorption of chlorophyll which changes with the stage of ripeness. At wavelengths higher than 720 nm, the spectral shapes are almost structure-less. The chlorophyll fluorescence seen with high intensity in the red/NIR spectral region with peaks at about 685 and 740 nm was excited with UV light at 385 nm (Fig. 3, bottom). With increasing ripeness, these two fluorescence peaks change in absolute intensities, and the ratio of peak intensities was also influenced. This is related to the fact that while the 690 nm peak is reabsorbed by the chlorophyll pigment itself, this is not the case for the 740 nm peak. Thus, the latter peak grows more than the first one when the pigment concentration increases.

Fig. 3

Reflectance spectra (top) and fluorescence spectra (bottom) of each kind of fruit during measurement day 1, 3, 5, 7, and 9.

From the facts above, for all four kinds of fruits, the reflectance and fluorescence intensities and spectral shapes are clearly sensitive to chlorophyll content, so the reflectance and fluorescence spectra can be used as indicators to estimate the changes of chlorophyll in fruit ripening.

For getting further spectral information, more than one spectral band was studied. Thus, contrast functions were constructed to compress the multispectral information. By choosing the contrast functions to be dimensionless quantities, intensity fluctuations and other variations in the spectral measurements can be eliminated.²¹ Here, we define a function $R$ to show the changes of reflectance spectral shape with $R=(R_{720}-R_{670})/R_{720}$, where $R_{670}$ and $R_{720}$ are the reflectance at 670 and 720 nm, respectively. For nectarines (NE) in the two different temperature groups, the changes of $R$ with time are shown in Fig. 4. Similarly, a function $F$ was formed to express the changes of the fluorescence spectral shape. $F$ is calculated as: $F=I_{740}/I_{685}$, where $I_{685}$ and $I_{740}$ are the fluorescence intensity values at 685 and 740 nm, respectively.²² As discussed earlier, $F$ is quite sensitive to the chlorophyll concentration. The time evolutions of $F$ for NE are shown in Fig. 5.

Fig. 4

The reflectance function $R$ for nectarines formed from the measured data (a) and the average for different temperatures (b).

Fig. 5

The fluorescence function $F$ for nectarines formed from the measured data (a) and the average for different temperatures (b).

It can be seen in Figs. 4 and 5 that the changes of $R$ and $F$ for NE have a similar tendency for the two temperatures (20°C and 30°C). Both functions, $R$ and $F$, decrease rapidly to very low values and then stabilize. It can also be noted that the average value of $R$ and $F$ for the 20°C and 30°C groups have very small differences for reflectance and fluorescence, respectively [Figs. 4(b) and 5(b)]. Considering the error bars in these two figures, for most of the values of $R$ and $F$, the temporal displacement of the fall-off for the two temperatures is not more than 1 day, meaning that the time for passing a chosen threshold, related to ripeness, is not more than 1 day. Although small differences are observed, we combine both temperature groups to gain higher statistical significance. The observations made for NE are also made for the three other types of fruits. The average values of $R$ and $F$ for the four fruit types are shown in Fig. 6. We note that while $F$ approaches zero (the disappearance of the fluorescence peak at 740 nm), the minimum of $R$ has a finite value since the reduction in reflectance at 670 nm due to chlorophyll [Fig. 3(a)] never becomes negligible.

Fig. 6

The time evolution of the average $R$ and $F$ values for the fruits with standard deviation error bars indicated. Threshold values indicating the onset of full ripeness are given.

The time evolution of $R$ and $F$ is seen to be decreasing and then arriving at a basically constant value for all the kinds of fruits, which indicates that the chlorophyll content gradually decreases to a minimum (Fig. 6). As discussed, the influence of temperature was small, but it still has some effects, which is reflected in the error bars seen in Fig. 6. Moreover, the measurement results for each fruit show some spread which is included in the error bars. The four kinds of fruits in this study are climacteric fruits. Such fruits continue to ripen after harvesting, accompanied by a color change. The skin color of all kinds of fruits in this study does change (e.g., mango skin will change from green to yellow) when the fruits ripen gradually, which is induced by the breakdown of chlorophyll and the formation of additional pigments. With the visual appearance of the fruits and our experience, the fruits are ripe when $R$ and $F$ begin stabilizing, which means that the fruit quality will be acceptable to the consumer at this point. Here, we set a threshold value for each kind of fruit, which is indicated by the dashed lines in Fig. 6. It is clear that the change of skin color in fruit ripening must be reflected in a change in the values of $R$ and $F$. The transition process combined with the threshold value is used to determine the time when the fruits are ripe. Due to some uncertainty and the statistical spread, there is clearly an uncertainty in the determination of the full ripeness day for different fruits. The time of ripeness can be inferred from the curves and the selected threshold values for all the fruits, as shown in Table 1. It should be noted that the absolute number of days does not have any significance here, since the prehistory of the fruits in the fruit dealers shop was unknown. The important aspect is that the ripeness of each kind of fruit can be assessed objectively with a quite small uncertainty using the function values and the set individual thresholds.

Table 1

Day of onset of full ripeness for four kinds of fruits assessed with reflectance and fluorescence spectroscopies.

3.2.

Gas in Scattering Media Absorption Spectroscopy Measurements

Since the injected laser power from the fiber was quite low in this study, a limited useful light propagation path resulted considering the reduction of light intensity due to absorption and scattering. Thus, rather than using transmission, the measurements were performed in a backscattering-like geometry with an injection-detection angular separation of 90°C, as can be seen in Fig. 1(b). Typical examples of acquired 2f signals of the four kinds of fruits are given in Fig. 7. It can be seen that the 2f signals for the fruits are quite good, so it is feasible to study changes of gas content inside the intact fruits during the fruit ripening.

Fig. 7

Measured 2f signals (gray) and the fitted curve (black) for four kinds of fruits. The shapes used for fitting were attained by recording the high-quality 2f WMS oxygen signal for 1500 mm of free air.

As a result of the strong scattering of light in fruits, the optical path length that the light travelled through the fruits is unknown, so straight-forward quantification of the oxygen concentration using the Beer–Lambert law is not applicable.¹⁵ Instead, the equivalent mean path length, $L_{\mathrm{eq}}$, is used to express the gas content. The $L_{\mathrm{eq}}$ value is the distance light would need to travel in normal air (21% oxygen) to result in the same fractional absorptive imprint on the received light intensity. $L_{\mathrm{eq}}$ is proportional to the normalized WMS signal (i.e., the 2f WMS signal amplitude divided by the light intensity received by the detector). Thus, $L_{\mathrm{eq}}$ is used to estimate the changes of oxygen content in fruits ripening in this study. Each sample was monitored five times consecutively every day, and the average derived $L_{\mathrm{eq}}$ was presented. The $L_{\mathrm{eq}}$ was measured as a function of time for the different fruits; see Fig. 8.

Fig. 8

The measured time evolution of the oxygen signal for nectarine, mango, guava, and papaya.

The time evolution of the oxygen equivalent mean path length shows a similar tendency for all fruits as seen in the figure. During the first days, the value of $L_{\mathrm{eq}}$ is found to decrease gradually to a minimum value, and then rise to a second peak. Eventually, it decreases to a very low value. In this process, the fruits are subjected to a series of physiological and biochemical changes. It should be noted that the changes in $L_{\mathrm{eq}}$ can be attributed to changes in oxygen concentration or in the path length through the gas, or both phenomena. A reason for the initial reduction of $L_{\mathrm{eq}}$ could be attributed to the changes of respiratory rate during fruit ripening. Because the four kinds of fruits are climacteric, the respiratory rate rapidly increases to a peak value (i.e., the climacteric point which is a stage of fruit ripening associated with maximum respiration rate and ethylene production) and then returns to or below the point before the event.^23,24 For the climacteric fruits, the climacteric point can be seen as the end point of the fruit ripening period. Then, the fruits are more susceptible to fungal invasion and begin to degrade, ultimately leading to cell death. It can be seen (Fig. 8) that the $L_{\mathrm{eq}}$ rises to a secondary peak and then decreases to a very low value after the climacteric point. This can be interpreted as due to the fact that the membrane structure is destroyed and the increasing membrane permeability makes the ambient air penetrate inside the aging fruits. Eventually, the fruits are decaying seriously and are filled with liquid replacing the air.²⁵ We note that the characteristic temporal variation of $L_{\mathrm{eq}}$ is useful for assessment of ripeness, regardless of detailed exploration. The day of full ripeness can be inferred from the time evolutions of the oxygen $L_{\mathrm{eq}}$ for the fruits, and can be associated with the occurrence of the initial minimum in the GASMAS signal. Also, considering the standard deviation errors, the results are shown in Table 2. We should note that from a practical point of view, on a particular measurement occasion it is not possible to locate the minimum of the curves shown in Fig. 8. However, again, since the GASMAS signal is monotonously falling in the interesting phase of the ripening process, we can again define individual threshold values for each kind of fruit, defining the initiation of full ripeness.

Table 2

Day of onset of full ripening for four kinds of fruits as assessed from the minimum of the gas in scattering media absorption spectroscopy signal.

Combining Tables 1 and 2, the time duration during which the fruits can be considered to be progressing to full ripeness is estimated in Table 3 for each fruit type, using the combination of reflectance and fluorescence spectroscopies, as well as the GASMAS technique.

Table 3

The estimated number of days to the onset of full ripeness using a combination of the three spectroscopic techniques.

From Table 3, we note that the width in the period indicated becomes larger when combining the data. However, by setting thresholds more correctly based on experience of factual maturity these intervals could be narrowed down. What we have demonstrated is that the spectroscopic signals from all three methods are useful for estimating the degree of ripeness. Clearly, to get agreement with subjective maturity, the thresholds for individual kinds of fruits need to be firmly established. It seems that each of the spectroscopic techniques, or to gain more certainty, a combination thereof, could be quite useful to study fruit ripening and fruit quality. Useful shelf life might extend by a few days beyond the given time for the onset of full ripeness with basically nonimpaired quality.

Ethylene, a gaseous organic compound in fruits and widely known to be involved in the ripening process, is not considered in our study. Especially for the climacteric fruits, large quantities of ethylene are produced during fruit ripening with a sharp increase of ethylene concentration occurring in connection with the reaching of the climacteric point for respiration.^24,26 The increasing ethylene level causes an increase in the respiration rate, accompanied by biochemical and physiological transformations during the ripening process. Therefore, the ethylene content together with oxygen can also be used as an indicator for fruit ripening.