1. Introduction
The irregular distribution of precipitation ascribed to climate change influences sweet corn production. This results in the need for irrigation. The water demands of corn plants differ depending on their stages of development. During the early stage of development of maize, water scarcity does not decrease the yield [
1], but the plants require the most water (80–100 mm) from the tasseling to grain filling periods. The amount of precipitation has a significant influence on the yield [
2]. Ear differentiation begins at the six and eight leaf stages of development of sweet corn. A water deficit at this stage results in a reduction in ear length and the number of kernel rows [
3]. During tasseling, water deficiency can decrease the yield by 40 to 50% [
4]. However, the length of the anthesis-silking interval also has a great effect on the grain yield under drought stress [
5]. Traditional sweet corn comprising homozygous gene (
sh1) has higher sugar content in the endosperm of the kernel than that of normal field corn. Nevertheless, the kernel quality of these sweet corn varieties declines rapidly after harvest due to the conversion of sugars into starch and moisture loss [
6]. The super-sweet varieties (
sh2) contain two to three times more sucrose concentration in the fresh kernel and retain higher sugar and moisture concentrations for longer postharvest periods than the traditional sweet corn variety (
sh1) [
7]. However, these are sensitive to the temperature and water content of the soil that can result in low field emergence and seedling vigour [
8].
Precision agriculture requires the use of non-destructive, cost-efficient methods that help farmers to make rapid management decisions, including irrigation for a large yield [
9]. Recently, the spread of remote sensing techniques has made it possible to assess the water status of plants during their growth [
10]. The leaf photosynthetic capacity of the varieties can be monitored by measuring chlorophyll content and net photosynthetic rate under different environmental and stress conditions [
11]. Photosynthetic pigments in the leaves absorb red and blue light in the visible waveband when the reflectance is minimal. As a result of stress factors, the chlorophyll content of leaves changes; therefore, the light absorption of the leaf decreases [
12] which results in an increase in light reflectance and a decrease in the photosynthetic activity [
13] that finally leads to yield decrease. A portable chlorophyll meter measures the light absorbance or reflectance given by SPAD values (relative chlorophyll content) on the leaf which is used to measure the greenness degree of leaf [
14] and detect the drought tolerance of some species [
15,
16,
17].
The degree of spectral reflectance on photosynthetic crop canopy has been detected by vegetation indices using remote sensing techniques. The normalized difference vegetation index (NDVI) is one of the most frequently used vegetative indices for the monitoring of vegetation. NDVI is suited to monitoring plant growth [
18] and to detect the presence of abiotic stress [
19,
20], and it could be used for the selection of genotypes with drought tolerance [
21]. A positive correlation was found between the NDVI and the chlorophyll concentration of turf grass [
22] and the leaf area index (LAI) of bean [
23]. However, the trend of the relationship between NDVI and LAI was determined by water supply conditions [
24]. The measurement of SPAD is considered suitable for predicting the expected yield of some plant species, such as maize and rice [
25,
26], while the relationship between the spectral vegetation index NDVI and yield can be the basis of yield prediction for wheat [
27] or beans [
24].
The improvement of drought tolerance is an objective for most plant species. The changes in physiological processes through the responses to drought stress can be tracked by the use of remote sensing methods. Understanding their relationship to yield would be advantageous for early selection among genotypes. Changes in the pigment concentration of plants using spectral reflectance measurements have been studied under different degrees of environmental stress. However, their effects on yield and suitability for yield prediction of sweet corn have not been investigated yet.
The objective of this study was to evaluate the change in spectral reflectance at leaf and canopy level in the stages of development of sweet corn and to reveal their relationship to yield and nutritional quality under dry conditions. In addition, the objective was to determine the appropriate stage of development for the prediction of yield.
2. Materials and Methods
2.1. Experimental Conditions
From 2011 to 2013, the productivity and yield quality of three super-sweet corn hybrids were examined under different water supply conditions in field experiments. The experiments were conducted on sandy loam soil at the Experimental Station of Centre of Agricultural Sciences, the University of Debrecen, Hungary. The super-sweet corn used in the experiments had different ripening times; the GSS 1477 hybrid was middle–early-ripening (74 days), Overland (84 days), and GSS 2259 (87days) represented middle–late and late-ripening hybrids, respectively. The sowing of the hybrids was carried out in two factorial experiments where the main factor was the water supply and the sweet corn hybrids were the second. Each hybrid was arranged in randomized complete block designs with three repetitions. The size of single plots was 8.4 m2 with 60 plant density. The plots comprised four rows each, 3 m long and 2.8 m wide, where the row spacing was 70 cm, and the plant spacing was 20 cm.
Three water supplies were used: I
1.0 represented the optimum water supply plots where the loss of evapotranspiration (ET 100%) was replenished, I
0.5 treatment where water deficit irrigation was used (i.e., irrigation 50% of I
1.0). In unirrigated plots (I
0), the plants were grown under rain-fed conditions. The loss of evapotranspiration was determined by the daily weather data from a weather station near the experiments using the method described by Shuttleworth and Wallace [
28]. The calculation of the required irrigation water was carried out on the basis of actual evapotranspiration, as described by Allen et al. [
29]. Drip irrigation was applied. When the plant produced 6 or 8 leaves, then 10 plants were selected and tagged to measure the SPAD in each plot.
2.2. Measurement of Traits
Measurement on the leaf of selected plants was carried out by SPAD-502 (Minolta UK) portable chlorophyll meter. This equipment measures the ration of the light absorbance in the red and near-infrared region and the given SPAD value. The measurements were taken on both sides of the midrib on the upper third part of the developed leaf for each selected. This means 60 measurements on 10 plants per each plot. The measurements were performed between 10:00 and 14:00 hours at three times; at the 8-leaf stage (ST1), during tasseling (ST2), and during 50% of silking (ST3) in 2012 and 2013.
The green canopy of plants absorbs the large part of the incident visible light and reflects those in the near infrared region. Normalized difference vegetation index (NDVI) expresses the ratio of reflectance of active photosynthetic radiation in visible and near infrared regions of the spectrum [
30]. NDVI was measured by GreenSeeker 505 (Hand-Held, Manuel NTech Industries Inc., Ukiah, CA, USA) portable model equipment. This measures the reflected light on the canopy crops at 660 nm (red) and 770 nm (near-infrared) bands. Ten measurements were carried out covering approx. 0.8 m
2 in each plot (average of approx. 15–20 samples each) and these were performed between 10:00 and 14:00 hours. The measurements of NDVI were performed at the same stage of development as described for that of SPAD from 2011 to 2013.
The leaf area index (LAI) expresses the growth intensity of crop canopy. The measurements of LAI were performed by a LAI-2000 Plant Canopy Analyzer (LI-COR Inc., Lincoln, Nebraska, USA), and their date was the same as that of NDVI.
2.3. Measurement of Yield
In each plot, the height of 10 selected plants was measured then they were harvested individually. The ears of these plants were classified into two groups: I, the marketable group comprised well-developed and healthy ears with filled kernels along the rows and II, the non-marketable group involved short, abnormal formed ears with unfilled kernels of rows. The ears without husk of marketable group (I) were further evaluated, i.e., the weight, length, and diameter of the ears were measured. The number of rows per ear and number of kernels per row were also counted. Other plants of the plots were harvested and separated to ears and green biomass and weighed.
2.4. Chemical Analysis
Dry matter content of the kernels was determined as described by Shreve et al. [
31] and expressed by percentage to fresh weight. The sugar content of kernels was determined by the Luff–Schoorl method [
32] given in terms of fresh weight %. The total carotenoids content of the sample was determined with chromatography and was given mg kg
−1 as described by Nemeskéri [
33].
2.5. Statistical Analysis
Data were evaluated with two-way analysis of variance (ANOVA) (water supply × variety) using SPSS 20.0 for Windows (IBM, Armok, New York, USA). The means of each treatment were compared by Duncan’s Multiple Range Test at p < 0.05 level. Regression analysis was carried out to reveal the relationships among the SPAD, NDVI, LAI, and yield components and nutritional quality of kernel, respectively. Linear, logarithmic, hyperbolic, power, and exponential regression analysis were carried out. The most appropriate regression equation based on the highest correlation coefficient (r) and significant p level was interpreted.
4. Discussion
Remote sensing techniques are useful tools to monitor the growth and response of the crop to water stress or evaluate the yield [
18,
25]. Water-stressed sweet corn absorbs less light in the visible region and more light in the near-infrared region of the spectrum than irrigated, i.e., unstressed plants [
19]. Others [
24] established that the high SPAD values of bean leaves were visually associated with dark-green coloured leaves under severe water deficiency, which could be due to the change of the water and chlorophyll content in the leaves accompanied with decreasing light absorbance in the blue and red region of the spectrum and increasing reflectance for the water-stressed plants.
Maize is considered to be relatively tolerant to water stress in the vegetative stage and very sensitive during the tasseling, silking, and pollination periods [
34]. The degree of water stress tolerance of sweet corn hybrids is detected by the change in spectral reflectance on leaf and canopy measured during generative stages of development. It has been reported that the spectral reflectance on the leaf was higher for maize [
25], snap beans [
24], and green peas [
16] under drought stress than under optimal water supply conditions. Others [
35] recorded lower chlorophyll content (SPAD unit) for most inbred maize lines under drought stress condition than in the well-watered plants. On the contrary to these statements the results in this study were different; the leaf and canopy spectral reflectance of sweet corn hybrids measured under low rain-fed (I
0) condition were lower than under optimal water supply condition (
Table 1). However, the responses of hybrids to water deficiency were different during their generative stages of development.
When both leaf and canopy reflectance is low within the photosynthetic active radiation (PAR) from 400 to 700 nm, then photosynthesis is undisturbed [
25]. A relatively low spectral reflectance on leaf shown by 40 to 45 SPAD values provided an undisturbed photosynthetic activity until the tasseling period. Therefore, the crop canopy (LAI) and spectral vegetation index (NDVI) increased intensively (
Figure 2). During the silking period, the difference in the SPAD between the hybrids was pronounced under different water supply conditions. The highest SPAD values were detected both during the tasseling and silking periods for the middle–late-ripening Overland sweet corn under severe drought (I
0). Nevertheless, under water deficiency (I
0.5) using deficit irrigation, a larger SAPD value was detected during the silking period than in the tasseling period. As a result of high SPAD, the photosynthetic activity decreased during silking which contributed to a significant decrease in LAI that was shown for the late-ripening GSS 2259 hybrid (
Figure 3b and
Figure 5b).
In drought, decreasing crop canopy results in a decrease in the water loss of plants, closing stomata and the high stomatal resistance result in the reduction of plant height and a decrease in the diameter and weight of the ears of sweet corn [
36]. It has been published that deficit irrigation decreases the length and diameter of ears significantly, the number of kernels per ear, and the final yield of sweet corns [
37,
38]. In contrast to these reports, our findings showed that although deficit irrigation decreased the height of plants under water scarcity, the size of ear and number of kernels per ear only decreased under severe drought, compared with the optimal water condition (
Table 1 and
Table 2). Under moderate water deficiency, a strong correlation (
r = 0.69) between stomatal resistance measured during tasseling and husked ear yield has been reported [
36], but the spectral traits (i.e., SPAD and NDVI) also have considerable influence on the development of individual husked ears. The finding revealed that during tasseling, the expected husked ear per plant could be predicted by both individual measurements on the leaf (SPAD) and on the canopy level with spectral vegetation index (NDVI) and could be used to select genotype with water stress tolerance.
A reliable prediction of yield and quality before harvest would be useful in precision agriculture [
9]. Nevertheless, the reliability of the yield prediction is determined by the measurements taken on the leaf or canopy level, the stages of development, and the growing conditions. Spectral vegetation index (NDVI) has been offered to predict the expected yield of some plant species. Strong correlation was found between NDVI and final yield at 15 days after flowering of maize [
20], at 54 days after sowing for Pinto bean [
10], and during early flowering of cotton [
39]. According to the others [
25], the relationship between the predicted and actual grain yield was higher when the measurements of spectral reflectance were taken on leaf level than the canopy level. This has been confirmed by our results, wherein the measurement on the leaf expressed by SPAD values was more suitable to predicting both individual ear yield and final yield than that on the canopy level presented by NDVI. However, it could be done only under moderate water shortage (I
0.5). Under this growing condition, as long as the SPAD value is low during tasseling (e.g., 46), then the difference between the final and predicted yield is approx. 5.7%, while this can be higher (7–14%) at a 49 SPAD value. Nevertheless, the difference between the final yield and expected yield could be either 6.2 or 25.8% in the range of 46 to 49 SPAD measured during the silking period, depending on the hybrids. This result points to the fact that the spectral reflectance measured on the individual leaf during tasseling is more suitable to predict the yield of sweet corn than those during the silking period under moderate water deficiency.
Water shortage has effects not only on the yield of plants but on their nutritional quality. Under mild and severe water deficit, higher protein and sugar content of kernels were observed than those of well-watered plants [
40]. Knowledge of the relationship between the spectral traits and nutritional quality of yield is rather lacking. In drought, a close correlation between the SPAD values measured during pod development of snap beans and dry matter content of the pods has been established but the relationship of SPAD with the protein and fibre content of pods being dependent on crop year [
24]. The SPAD measured during silking has been considered to be suitable for the prediction of protein content of maize yield [
41]. Nevertheless, the NDVI was not found to be appropriate for the reliable prediction of protein content of grain of barley and wheat [
9,
42].
According to the results, under water shortage, the nutritional quality of sweet corn kernel can be predicted by the measurement of spectral reflectance on both leaf and canopy level. On the basis of the correlation between the SPAD, NDVI, and the sugar content of kernels, the expected sugar content of yield of sweet corn grown under unirrigated dry condition (I
0) can be indicated with 58% and 80% reliability by the measurements during the tasseling period. Nevertheless, their realization will be determined by the LAI results obtained during silking period. Under moderate water deficiency (I
0.5), the expected sugar content of kernels of sweet corns can be predicted with 64 and 68% reliability by SPAD and NDVI, respectively, measured during tasseling. However, the LAI seemed to have larger role in the promotion of accumulation of carotene in the corn kernels than spectral traits during silking (
Table 5).
The 46 to 49 range of SPAD measured during tasseling provided a reliable prediction of yield, high sugar, and carotene content of kernels of sweet corn hybrids grown under moderate water deficiency which was confirmed by the data shown in
Table 6. High leaf spectral reflectance (SPAD) measured during generative stages of development resulted indirectly a decrease in the biomass and low yield and high dry matter content of kernels which was occurred for the Overland variety (
Figure 3,
Table 6). The late-ripening GSS 2259 hybrid was more sensitive to deficit irrigation than the middle–early GSS 1477 because its higher SPAD value measured during silking (
Figure 3) contributed to the decrease by 16.7% of yield, in comparison with the optimal water condition (
Table 6).