Salinity is a significant threat to wheat sustainability in arid and semiarid regions, where brackish water is commonly used for irrigation and the salinity-affected soil is expanding over time. Therefore, developing salt-tolerant wheat varieties is a cost-effective solution to ensure the sustainability of this crop in these regions. In the last two decades, there have been considerable efforts to improve the salt tolerance of wheat genotypes, but progress has been limited. Reviews of current research have pointed out challenges in creating salt-tolerant wheat genotypes through breeding programs. These obstacles include the lack of reliable screening criteria, limited genetic diversity for salt tolerance, the inadequate assessment of salt tolerance in real field conditions, and the inefficiency of traditional phenotyping methods due to time and cost limitations when assessing the salt tolerance of a large number of genotypes [
14,
15,
21,
45,
46]. In this study, we assessed 24 wheat genotypes, including salt-tolerant and salt-sensitive check genotypes, as well as sixteen F
8 RILs, to cover a broad spectrum of genetic diversity for salinity tolerance. The ANOVA supported this objective and revealed significant variations among genotypes for four morpho-physiological traits (PDW, LRWC, Chlt, and GY) in each year and when the data from two years were combined (
Table 1). These findings also highlight the importance of these four traits as screening criteria for evaluating the salt tolerance of wheat genotypes, which will be discussed further below. Importantly, the salt tolerance of these genotypes was assessed in real field conditions using both destructively measured morpho-physiological traits and high-throughput genotyping and phenotyping tools that are cost-effective, time-efficient, and non-destructive.
Salt stress components, such as osmotic stress, nutritional deficits for essential ions, and specific ion toxicities, interact to restrict several physiological and biochemical processes involved in plant growth and development. These negative impacts of salinity stress ultimately translate into significant decreases in plant biomass. Furthermore, plant biomass, which encompasses a range of physiological processes and integrates plant responses to salinity stress across different growth stages, is closely linked to radiation use efficiency, the conversion of light into biomass, and overall crop yield. Additionally, improving grain yield through genetic improvements may require a focus on increasing plant biomass rather than the harvest index. Importantly, when plants are exposed to high salinity levels, their energy and metabolic resources are diverted to activate stress-tolerance mechanisms instead of being used for growth and biomass production [
6,
40,
47,
48,
49]. These facts about plant biomass indicate that this plant trait could serve as a reliable screening criterion for distinguishing salt tolerance among various wheat genotypes. The study results confirmed that PDW measured at 90 days after sowing is a crucial screening criterion for assessing salt tolerance in the tested genotypes. Significant differences (
p < 0.001) were observed in PDW among the genotypes (
Table 1 and
Figure 1). The PDW also exhibited high broad-sense heritability (h
2 = 77.52%) and a moderate value for GG (7.8%), with the PCV value comparable to the GCV value (
Table 2). Notably, SSR markers showed a strong association with PDW (R
2 = 0.75), with markers Gwm55 and Wmc154 displaying a significant association (R
2 = 0.64) with this trait (
Table 3). These findings indicate that the PDW is under genetic control, emphasizing the importance of phenotyping this trait as an important screening criterion for evaluating salt tolerance in wheat genotypes under real field conditions.
Previous studies have demonstrated that the water status of plants is essential for efficient photosynthesis. When leaves have reduced turgor potential, it can affect their enlargement, stomata opening, and consequently, the overall efficiency of photosynthesis in plants [
6,
50]. High levels of NaCl in the soil can hinder a plant’s water absorption, leading to physiological drought stress. This affects the plant’s water balance, particularly its LRWC, which reflects the balance between water uptake and transpiration. Therefore, the LRWC can serve as a reliable indicator of abiotic stress compared to other plant physiological and biochemical characteristics. It is also closely linked to other plant water status parameters such as water potential, osmotic potential, and turgor pressure [
51]. These facts about LRWC indicate that this trait may be considered an appropriate phenotypic screening criterion for evaluating the salt tolerance of wheat genotypes. In this study, the LRWC showed significant differences (
p < 0.001) among tested genotypes (
Table 1) and exhibited a moderate heritability in a broad sense (h
2 = 48.51%) (
Table 2). On the other hand, SSR markers showed a strong association with LRWC (R
2 = 0.62), with markers Barc44 and Wmc11 displaying a significant variation (R
2 = 0.54) in this trait (
Table 3). These findings indicate that LRWC could serve as a useful screening criterion for identifying salt-tolerant genotypes at both phenotypic and genotypic levels.
The result of this study also found that the Chlt showed significant differences (
p < 0.001) among tested genotypes (
Table 1 and
Figure 1) with a high heritability in a broad sense (h
2 = 61.84%) and a moderate GG (11.76%) (
Table 2). The results for Chlt, both phenotypic and genotypic, confirm its significance as a key screening criterion for assessing salt tolerance in wheat genotypes under actual field conditions. This observation is likely because the reduction in chlorophyll content is a typical response to salinity stress, with salt-sensitive genotypes showing a more significant decrease compared to salt-tolerant ones [
6,
52]. The decrease in chlorophyll content under salinity conditions may be caused by the combined osmotic and ionic toxicity of salinity stress. This can lead to increased activity of chlorophyll-degrading enzymes, such as reactive oxygen species (ROS), as well as the disruption of enzymes involved in chlorophyll synthesis. Ultimately, this results in reduced chlorophyll formation or accelerated chlorophyll degradation [
53]. However, other studies have indicated that salinity stress has a lesser effect on chlorophyll content, with no significant differences observed between salinity-treated and control groups. This is attributed to the development of smaller, thicker leaves with a higher concentration of chloroplasts per leaf area under salinity stress. Consequently, there is an increase in chlorophyll content per unit area [
6,
54]. This indicates that the chlorophyll content may not remain stable if salinity stress has a significant impact on leaf anatomy, such as the size and thickness of the leaf. This could explain why SSR markers only showed a weak relationship with chlorophyll content (R
2 = 0.39), as shown in
Table 3.
In general, plant breeders usually prioritize the final grain yield (GY) when assessing genotypes under abiotic stress. This is because GY reflects various crucial factors that develop at key growth stages during the crop growth cycle, such as biomass allocation, sunlight interception, and conversion. This makes GY a comprehensive indicator of genotype performance and stress tolerance [
15,
46,
55]. Consequently, GY can serve as a valuable screening criterion for evaluating the salt tolerance of genotypes. The results of this study confirm this statement and found that GY showed highly significant differences (
p < 0.001) among genotypes (
Table 1 and
Figure 1), with a high broad-sense heritability of 84.0% and a moderate GC of 10.32%. The PCV was also comparable to the GCV. Additionally, three SSR markers (Barc44, Cfd9, and Barc34) explained 66% of the variations in GY. These results confirm the importance of GY as a screening criterion for evaluating the salt tolerance of wheat genotypes.
Ability of Different Spectral Reflectance Indices as an Alternative Screening Criteria for Evaluating Salt Tolerance in Wheat Genotypes
In general, salinity stress causes significant changes in various biophysical and biochemical properties of plant canopies, resulting in alterations in their spectral signatures across the entire spectrum (400–2500 nm). These changes can be detected through various spectral regions. For instance, variations in leaf pigment content, plant health, and photosynthetic capacity are noticeable in the VIS (400–700 nm) spectrum. Changes in biomass accumulation, leaf structure, and leaf area index impact canopy reflectance in the NIR (7900–1300 nm) region. Changes in plant water status can be detected through specific water absorption bands in the NIR and SWIR (1300–2500 nm) regions. To capture these changes in a simple way, two types of SRIs have been developed. The first type, vegetation SRIs, includes wavelengths from the VIS, red-edge (700–850 nm), and NIR regions. These indices are effective in detecting variations in chlorophyll pigments, vegetative vigor, photosynthetic efficiency, and biomass accumulation. The second type, water SRIs, incorporates weak and strong water absorption bands from the NIR and SWIR regions, making them suitable for detecting changes in plant water status [
24,
28,
29,
30,
33,
34,
56,
57]. This close relationship between SRIs and plant characteristics indicates that SRIs can be used as non-destructive and cost-effective screening criteria for evaluating salt tolerance in wheat genotypes. This offers an alternative to traditional morpho-physiological traits commonly used in breeding programs to evaluate salt tolerance. Moreover, SRIs can be a viable substitute for traditional phenotyping methods in indirectly assessing the performance of wheat genotypes under salinity stress conditions. To investigate the ability of SRIs as alternative screening criteria, a genetic analysis, heritability, cluster analysis, and association with SSR markers were tested for SRIs and compared with those of morpho-physiological traits. SRIs can be used as alternative screening criteria to evaluate salt tolerance if they are comparable to direct destructive traits in terms of genetic analysis and heritability. In general, to make reliable selections based on additive gene action, it is recommended to consider broad-sense heritability, the GCV, and GA together. Johnson et al. [
58] emphasized the importance of evaluating both heritability and GG, as high heritability does not always translate to a high GG. The results of this study found highly significant differences among genotypes in different SRIs (
p < 0.001) (
Table 1 and
Figure 2 and
Figure 3). Most SRIs had high heritability and a moderate GG, with similar values for the GCV and PCV (
Table 2). The SRIs effectively clustered check salt-tolerant genotypes Sakha 93 and Kharachia-65 together and separated check salt-sensitive genotype Sakha 61 (
Figure 4B). SSR markers showed a strong association with SRIs (R
2 ranged from 0.56 to 0.89), with specific markers (Barc44, Gwm350, Gwm 335, and Cfd9) displaying the most variation in SRIs (
Table 3). Importantly, the Mantel test revealed a significant positive correlation between the phenotypic clustering based on different SRIs and SSR-based clusters (r = 0.269,
p < 0.0001). The heritability and genetic gain results for various SRIs indicate that genetic factors play a significant role in the variation in these indices among genotypes. This suggests that phenotypic selection can effectively estimate these SRIs and their phenotypic performance could be valuable for selection in the context of genetic improvement [
59,
60]. Therefore, SRIs can serve as a dependable screening criterion for evaluating salt tolerance in wheat genotypes, eliminating the requirement for destructive morpho-physiological traits. Mohi-Ud-Din et al. [
19], Gutierrez et al. [
33], El-Hendawy et al. [
35], Prasad et al. [
61], and Babar et al. [
62] found similar results under drought stress conditions. They reported that significant genetic gains could be achieved by incorporating SRIs’ measurements during selection in wheat breeding programs, especially when the selection is carried out in mid- and late-breeding generations (F5 to F7).
In this study, we also explore the potential of using SRIs as a substitute for conventional phenotyping techniques to indirectly assess the four morpho-physiological traits. This will help us quickly and non-invasively evaluate the performance of multiple genotypes under salinity stress conditions. Several studies have used SRIs as effective tools to indirectly estimate various plant traits such as chlorophyll content, plant biomass, RWC, and GY in wheat and other cereal crops [
1,
18,
19,
33,
35,
63,
64,
65,
66]. However, most of these studies were conducted under normal growing conditions with a wide range of phenotypic variability in genotypes or drought stress conditions. To our knowledge, only a few studies have investigated the performance of SRIs for indirectly estimating different morpho-physiological traits under salinity stress conditions.
Previous studies have demonstrated that different vegetation SRIs and water SRIs can accurately predict a significant portion of the variability in GY and other plant traits such as plant biomass, chlorophyll content, and water content across various crops and environments [
26,
33,
35,
56,
57,
66,
67]. However, there is a discrepancy in studies regarding the effectiveness of vegetation SRIs versus water SRIs in estimating different plant traits. Some studies indicate that water SRIs, incorporating wavelengths from NIR and SWIR regions, are better at capturing genotypic variability in GY and plant biomass than vegetation SRIs, which use VIS, red-edge, and NIR wavelengths, under diverse environmental conditions [
33,
61,
62]. However, other studies show that vegetation SRIs perform better or equally well compared to water SRIs in estimating the GY of spring wheat under water deficit stress conditions [
35,
67,
68,
69]. This study found that both types of SRIs were effective in assessing PDW and GY under salinity stress conditions, with the vegetation SRIs and water SRIs showing a moderate relationship with PDW (R
2 ranged from 0.46 to 0.60 and 0.38 to 0.56) and a strong relationship with GY (R
2 ranged from 0.69 to 0.80 and 0.69 to 0.90), respectively (
Table 4). These results indicate that genotypic variations in plant biomass and GY under salinity conditions can be assessed during the vegetative stage using vegetation SRIs and water SRIs without the need to wait for plants to reach maturity. This suggests that SRIs can be a rapid and straightforward phenotyping tool to effectively assess the salinity tolerance in larger populations in breeding programs within a shorter time frame. The common negative effects of salinity stress, such as reduced biomass, chlorophyll degradation, and changes in leaf structure and water content, may explain why both vegetation SRIs and water SRIs incorporating green, red, red-edge, NIR, and SWIR bands were successful in explaining variations in PDW and GY among tested genotypes in this study.
The study revealed that vegetation SRIs (R
2 = 0.40–0.68) provided more accurate estimates of Chlt compared to water SRIs (R
2 = 0.19–0.28). Among the vegetation SRIs examined, the MTVI and OSAVI were the most precise (R
2 = 0.68 and 0.62, respectively) in estimating Chlt, followed by the Chl
red-edge and EVI (R
2 = 0.53 for both) (
Table 4). Similarly, previous studies have demonstrated that SRIs incorporating wavelengths in the red, red-edge, and NIR regions are more reliable indicators of leaf chlorophyll content than indices in other regions. This is because these wavelengths are effective at detecting the absorption characteristics of photosynthetic pigments like chlorophyll-a and -b, α- and β-carotenes, and lutein, which are affected by salinity stress [
70,
71,
72,
73]. The superior performance of the MTVI, OSAVI, and EVI in estimating Chlt content in wheat compared to other vegetation SRIs may be attributed to these indices effectively addressing saturation issues across a range of LAI values from two to eight and taking into account internal and surface structural effects on leaf surface reflectance. On the other hand, other vegetation SRIs like the BNDVI, GNDVI, and RNDVI are less effective in estimating chlorophyll content, as they may reach saturation levels when LAI values are higher than two [
72,
73,
74,
75,
76,
77].
The study found that both vegetation SRIs and water SRIs are equally effective in estimating LRWC and showed a moderate relationship (R
2 = 0.23–0.42 and 0.30–0.44, respectively) with this plant trait (
Table 4). The water bands in the NIR and SWIR regions are known to be effective for detecting the water status of plants. The water absorption bands in NIR region, particularly at 970 nm, can penetrate deeper into the canopy, allowing for the estimation of plant water status [
33]. Moreover, the water bands in the SWIR regions such as 1400, 1450, 1650, 1920, 1950, and 2250 nm are sensitive to changes in plant water status and less affected by noise from internal leaf structure [
78]. This may explain why the water SRIs that combine NIR and SWIR wavelengths were successful in estimating LRWC under salinity conditions in this study. The study’s findings indicate that variations in LRWC could also be successfully estimated by vegetation SRIs that incorporate wavelengths related to pigment levels, photosynthetic capacity, growth vigor, structural leaf compounds, leaf cellular structure, and biomass accumulation rather than being influenced by alterations in plant water content. Kovar et al. [
79] found similar results, indicating that the green wavelength in the VIS spectrum was effective in estimating the plant water status, which was expressed by the relative water content, leaf water potential, and equivalent water thickness of soybean under varying irrigation levels. Under stress conditions, plants experience a decrease in cell turgor due to reduced water content. This leads to cell shrinkage, lower chlorophyll levels, and increased reflection in the green spectral region [
80]. Colovic et al. [
30] found that some vegetation SRIs incorporating red-edge wavelengths were also effective in estimating changes in water status in sweet maize. This could explain why vegetation SRIs accurately estimated LRWC as water SRIs under salinity stress in the present study.
To gain a better understanding of the relationships between morpho-physiological traits, different SRIs, and genotypes, a PCA was conducted. The results revealed that traits, vegetation SRIs, and four water SRIs (RWI, NDWI, NDMI, and SWSI) were closely linked and associated with salt-tolerant genotypes and the top five RILs exhibiting high values for traits and SRIs. In contrast, the remaining three water SRIs (NWI, DMCI, and NMDI) were grouped separately and were closely linked to a salt-sensitive genotype, specifically the bottom five RILs with lower values for traits and most SRIs (
Figure 6). The results indicate that both vegetation SRIs and water SRIs are successful in differentiating between salt-tolerant and salt-sensitive genotypes, similar to traditional morpho-physiological traits that are measured destructively. The PCA further validated that various SRIs can be effective, non-destructive, and affordable tools for phenotyping in wheat breeding programs under salinity stress conditions, particularly as these SRIs varied significantly among genotypes and were closely linked to SSR markers.
Due to the limited number of wavelengths included in each SRI, many of them are unable to effectively address the effects of soil background, saturation of the leaf area index, chlorophyll content, and biomass on the performance of the index of interest. This inefficiency hinders the accurate estimation of morpho-physiological traits using individual SRIs [
35,
73,
75]. Additionally, the same set of SRIs exhibits substantial multicollinearity among them, which limits their accuracy in predicting plant traits when dealing with treatments or genotypes that have similar reflectance and absorption patterns. Hence, as previously mentioned in the studies, combining PLSR with multiple SRIs has been shown to improve the accuracy of estimating various plant traits [
35,
81,
82,
83]. The findings of this study also validated that the PLSR models utilizing all vegetation and water SRIs improved the accuracy of the estimation of various morpho-physiological traits in both cal. and val. when compared to using individual SRIs (refer to
Table 4 and
Table 5 for comparison). This improvement could be attributed to the PLSR models in this study incorporating data from 14 SRIs and determining the optimal number of factors to accurately represent the calibration data without overfitting.