Spectral assessment of drought tolerance indices and grain yield in advanced spring wheat lines grown under full and limited water irrigation

Agricultural Water Management 182 (2017) 1–12

Contents lists available at ScienceDirect

Agricultural Water Management journal homepage: www.elsevier.com/locate/agwat

Spectral assessment of drought tolerance indices and grain yield in advanced spring wheat lines grown under full and limited water irrigation Salah E. El-Hendawy a,b,∗ , Wael M. Hassan c,d , Nasser A. Al-Suhaibani a , Urs Schmidhalter e a

Department of Plant Production, College of Food and Agriculture Sciences, King Saud University, 11451 Riyadh, Saudi Arabia Department of Agronomy, Faculty of Agriculture, Suez Canal University, 41522 Ismailia, Egypt c Department of Agricultural Botany, Faculty of Agriculture, Suez Canal University, 41522 Ismailia, Egypt d Department of Biology, College of Science and Humanities at Quwayiah, Shaqra University, Saudi Arabia e Department of Plant Sciences, Technische Universität München, Emil-Ramann-Str. 2, D-85350 Freising, Germany b

a r t i c l e

i n f o

Article history: Received 2 August 2016 Received in revised form 3 December 2016 Accepted 7 December 2016 Keywords: Canopy spectral reflectance High-throughput phenotyping Phenomics Proximal sensing techniques Vegetation index Water index Yield performance

a b s t r a c t Because wheat varieties exhibit a high genotype × environment interaction, several drought tolerance indices (DTIs) have been developed to assist breeders in selecting genotypes with good performance under contrasting water conditions. We compared the relative yield of advanced breeding wheat lines under both well-watered and limited water irrigation conditions using different DTIs and evaluated how spectral reflectance indices (SRIs), as rapid and non-destructive tools, can effectively monitor DTIs and grain yield. Sixty-five recombinant inbred lines (RILs) developed from a cross between drought-tolerant (Sakha 93) and drought-sensitive (Sakha 61) genotypes were subjected to full irrigation (FI) and limited water irrigation (LWI) in the 2014 (F6 ), 2015 (F7 ), and 2016 (F8 ) growing seasons. Eight vegetation- and water-SRIs calculated from canopy reflectance under FI and LWI, and taken at the heading and middle grain filling stages, were related to the DTIs and grain yield. We found that the yield performance of the RILs was not consistent across the two water regimes. Selection based on the DTIs, the stress susceptibility index and the tolerance index failed to identify RILs that had very low yields under both treatments. However, the mean productivity index (MPI) and the geometric mean productivity index (GMP) enabled us to identify RILs that produced desirable yields under both full and limited irrigation, and these drought tolerance indices further exhibited a high heritability. Across the three years of investigation and at the heading and middle grain filling stages, these DTIs were best described either by the vegetation-based dry matter content index (DMCI) or the water-based normalized multi-band drought index (NDMI), or a combination of both. In conclusion, our results demonstrate that a combination of near infrared (NIR) and shortwave infrared (SWIR)-based SRIs can be used as a fast and low-cost predictor for selecting wheat genotypes with superior yield under different water regimes. © 2016 Elsevier B.V. All rights reserved.

1. Introduction Water shortages currently impair almost every country in the world’s arid regions and they have become the norm rather than the exception in those regions. Most importantly, water shortages in arid regions will be further worsened due to abrupt cli-

∗ Corresponding author at: Plant Production Department, College of Food and Agriculture Sciences, King Saud University, 11451, Riyadh, Saudi Arabia. E-mail addresses: [email protected], [email protected], [email protected] (S.E. El-Hendawy). http://dx.doi.org/10.1016/j.agwat.2016.12.003 0378-3774/© 2016 Elsevier B.V. All rights reserved.

matic changes, continuous population growth, and rising incomes. According to Perry et al. (2009), as a consequence of climate change, most of the current water-scarce countries will become drier and warmer. Water shortage events have therefore gained increasing importance in both scientific and political contexts. Because the agriculture sector is the largest single user of fresh water, consuming about 75% of the available water supply (Prathapar, 2000), many governments in arid and semiarid regions have issued different regulations such as moratoriums on drilling new wells and mandates for the installation of water meters on pumping stations in order to restrict the irrigation water supply for this sector (Payero et al., 2008). These regulations will reduce the amount of

2

S.E. El-Hendawy et al. / Agricultural Water Management 182 (2017) 1–12

water that can be allocated for each crop because the increase in yield under sufficient irrigation may not compensate for the costs of this extra irrigation water. Thus, different strategies are urgently needed to improve wheat production, especially since the production of major crops could be reduced by more than 50% under limited water irrigation (El-Hendawy and Schmidhalter, 2010; Nezhad et al., 2012; El-Hendawy et al., 2014; El-Hendawy et al., 2015a). Although there are many strategies that can improve wheat production under limited water conditions, the development of new genotypes with higher yield potential and drought tolerance is still recognized as the most feasible strategy for addressing this challenge (Sinclair, 2011). Although many efforts have been made so far to improve the drought tolerance of wheat genotypes through molecular breeding, there have been very few successful examples of increasing the yields of wheat genotypes in farmers’ fields. A major constraint on accelerating wheat breeding for drought tolerance is the lack of effective evaluation tools for precise phenotyping of drought-related traits in breeding programs (Tester and Langridge, 2010; Mir et al., 2012; Elazab et al., 2015; Barakat et al., 2016). Unfortunately, the majority of the current tools for phenotyping drought-tolerance traits are destructive, expensive, and unaffordable, especially when a large number of genotypes are being evaluated (El-Hendawy et al., 2015a,b). There is therefore an urgent need for non-destructive, easy, rapid, practical, and economical phenotyping tools that can evaluate large numbers of genotypes in a relatively short time. One of the most promising techniques currently used to precisely monitoring traits related to drought tolerance for a large number of entries in a fast and nondestructive manner is the canopy reflectance sensing technique. The basic principle of this technique is based on the amount of light reflected from the canopy at a specific wavelength, which is a result of biochemical, physiological and structural properties of the canopy, providing several types of information that can be used to assess canopy chlorophyll content, canopy senescence, photosynthetic capacity, aboveground biomass, leaf area index, grain yield, and plant water status from a single spectrum (Aparicio et al., 2002; Babar et al., 2006; Prasad et al., 2007; Gutierrez et al., 2010; Weber et al., 2012; Erdle et al., 2013; Kipp et al., 2014; Lobos et al., 2014; El-Hendawy et al., 2015a; Elsayed et al., 2015). In cereals crop, the link between biomass and yield is established and well known. In the last decade, nondestructive devices based on chlorophyll fluorescence have been developed and tested for biomass and nitrogen assessment in several crops (Agati et al., 2013; Cerovic et al., 2015). Furthermore, the measurement of spectral reflectance with ground-based proximal sensing techniques has potentially be used as an easy, rapid, practical, and economical technique for assessing several phenotyping criteria related to drought tolerance. To date, plant breeders around the world consider grain yield per se as the main selection criterion for improving grain production under different environmental conditions. However, because of the substantial interactions between genotype and environment for this trait (Golabadi et al., 2006), repetitive evaluations of genotypes in different locations and in successive years are necessary to produce sufficiently accurate results when seeking to identify superior genotypes with high grain yield. In the absence of such a strategy, there is an increased risk of accidentally discarding good lines or retaining inappropriate genotypes in breeding trials. Further, measuring grain yield by conventional methods for a large set of entries and treatments is not an easy task. Because final grain yield is a function of many morphological and physiological characteristics that show significant differences among germplasm at different growth stages, several researchers have suggested that spectral reflectance approaches could be used to early predict grain yield prior to harvest in a rapid and nondestructive manner (Prasad

et al., 2007; Lin et al., 2012; Weber et al., 2012; Erdle et al., 2013; Hackl et al., 2014; Elsayed et al., 2015). Based on simple mathematical operations such as ratios and differences between the reflectance of the canopy at visible (VIS, 400–700 nm), near-infrared (NIR, 700–1300 ;1300nm), and short-wave infrared (SWIR, 1300–2500 nm) wavelengths, different spectral reflectance indices (SRIs) have been developed to predict different agronomic and physiological traits. Several researchers have suggested that measuring these SRIs periodically during different growth stages can be an effective way to rapidly and nondestructively predict grain yield under diverse environmental conditions (Weber et al., 2012; Erdle et al., 2013; Araus and Cairns, 2014; Lobos et al., 2014; El-Hendawy et al., 2015a; Elsayed et al., 2015). For example, Ma et al. (2001) found that under irrigated conditions, the green normalized difference vegetation index (GNDVI) had the highest association with soybean yield and explained up to 80% of the variability found in grain yield. The GNDVI and the near-infrared radiation (NIR)-based indices were also highly correlated with maize grain yield and explained 70–92% of yield variability at the middle grain filling stage under normal growing conditions (Shanahan et al., 2001). The normalized difference vegetation indices (NDVIs) were also well correlated with wheat grain yield under rainfed and irrigated conditions (Aparicio et al., 2002). The spectral indices related to normalized water indices (NWI-1, NWI-2, NWI-3 and NWI-4) were highly correlated with bread-wheat grain yield and explained more than 70% of the variation in grain yield under normal and water stress conditions in diverse studies (Marti et al., 2007; Prasad et al., 2007; Lobos et al., 2014; El-Hendawy et al., 2015a). Weber et al. (2012) reported that the spectral indices of wavelengths from 495 to 680 nm, from 680 to 780 nm, and at 900, 970, and 1450 nm, which are related to photosynthetic capacity, plant biomass, and plant water status, respectively, had the highest levels of association with maize grain yield under different water regimes. Lobos et al. (2014) reported that an NIR-based SRI such as the normalized difference moisture index (NDMI: 2200; 1100) was well correlated with the final wheat grain yield under limited water irrigation. Elazab et al. (2015) also found that the normalized green-red difference index (NGRDI) at anthesis was the trait best correlated with wheat grain yield under contrasting water conditions. However, because the response of genotypes to water stress varies with growth stage and with environmental conditions, care must be taken to identify a proper and consistent growth stage, as well as SRIs that can most effectively discriminate among genotypes in specific conditions and breeding trials. Generally, grain yield performance is not consistent for all genotypes across environments (Raman et al., 2012). When the grain yield of a large number of genotypes is evaluated under wellwatered and limited water irrigation conditions, the performance of genotypes for grain yield may be good or weak in both conditions, or good only in well-watered conditions, or in limited water conditions (Fernandez, 1992). Therefore, several drought tolerance indices (DTIs) that are based on mathematical relationships between normal and stress conditions have been proposed to identify desirable genotypes that perform well under a wide range of water treatments (Mohammadi et al., 2010; Cabello et al., 2013). The stress tolerance index (STI), stress susceptibility index (SSI), yield index (YI), tolerance index (TOL), mean productivity index (MPI), and geometric mean productivity (GMP) are examples of these selection indices and they have been applied in studies of many crops (Fernandez, 1992; Jafari et al., 2009; Singh et al., 2011; Drikvand et al., 2012; Cabello et al., 2013). It was found that MPI has an upward bias when the differences in grain yield between wellwatered and limited water irrigation treatments are large (Hossain et al., 1990). The GMP and STI are suitable indices when the performance of genotypes for grain yield is tested at both well-watered

S.E. El-Hendawy et al. / Agricultural Water Management 182 (2017) 1–12

and limited water irrigation conditions, and they favor genotypes with higher yield potential under both conditions (Ramirez Vallejo and Kelly, 1998; Cabello et al., 2013). Drikvand et al. (2012) also reported that the most effective DTIs for the identification of genotypes under well-watered and limited water irrigation conditions were the GMP, MPI, and STI. To the best of our knowledge, no study has tried to spectrally evaluate the relative yield performance of wheat genotypes under a range of water treatments by testing the relationships of SRIs to DTIs. The objective of this study was therefore to evaluate the grain yield of advanced breeding wheat lines under full irrigation and limited water irrigation, to test the relative yield performance of these lines through the DTI, and to assess how the SRI as a rapid and non-destructive tool can be used effectively to monitor the DTI and grain yield. 2. Materials and methods 2.1. Plant materials A set of 65 recombinant inbred lines (RILs) developed from crosses between Sakha 93 and Sakha 61 was used in this study. The two Egyptian parents, Sakha 93 and Sakha 61, were introduced and characterized as drought-tolerant and drought-sensitive cultivars, respectively, by the Agricultural Research Center, Ministry of Agriculture and Land Reclamation, Egypt. The 65 RILs and their parents were evaluated under full irrigation and limited water irrigation treatments in the years 2013/2014 (F6 ), 2014/2015 (F7 ), and 2015/2016 (F8 ). 2.2. Experimental conditions, design, and water treatments The field experiments were conducted at the Dierab Research Station of the College of Food and Agriculture Sciences, King Saud University, Riyadh, Saudi Arabia (24◦ 25 N, 46◦ 34 E, 400 m a.s.l.). Dierab is characterized by an arid climate with an average daily temperature and average precipitation during the wheat growing cycle of about 19.7 ◦ C and 30 mm, respectively. The soil texture of the experimental field is a loamy sand (82.4% sand, 9.5% silt, and 8.1% clay) with a field capacity, permanent wilting point, and plant available water capacity in the 0–40 cm surface layer of 0.124, 0.067, and 0.057 m3 m−3 , respectively, along with a bulk soil density of 1.5 g cm−3 and a pH of 8.2. The experiments were laid out in a randomized complete block split-plot design with three replicates; the irrigation treatments were kept in the main plots and the wheat RILs in subplots. Each subplot’s size was 4 m in length by 1.2 m in width (4.8 m2 in total area). The seeds of each RIL were planted on December 12 of each season in a six-row subplot and at a seeding rate of 330 seeds m−2 . Nitrogen and phosphorus fertilizers were applied at a rate of 180.0 and 31.0 kg ha−1 , respectively. Nitrogen fertilizer as urea (46.0% N) was applied in three equal doses at sowing, the beginning of tillering, and the beginning of booting growth stages. Whole phosphorus was applied prior to sowing as calcium super phosphate (15.5% P2 O5 ). The plots were protected from weeds and diseases throughout the growing season by recommended agronomic practices (El-Hendawy et al., 2015a). The wheat entries were evaluated under two water treatments that were imposed two weeks after sowing. The water treatments were full irrigation (FI) and limited water irrigation (LWI). The FI and LWI treatments were irrigated whenever the amount of evaporation water from a class A pan accumulated to 50 and 150 mm, respectively. The irrigation in the FI treatment was applied at the tillering (ZS 24), stem elongation (ZS 32), flag leaf emergence (ZS 37), booting (ZS 49), heading (ZS 59), complete emergence of inflo-

3

rescence (ZS 68), grain milk (ZS 73), and grain dough (ZS 83) of Zadoks’ decimal code of wheat growth stages (Zadoks et al., 1974), with the total amount of water, 7600 m3 ha−1 , being equivalent to 765 mm. The irrigation in the LWI treatment was applied at tillering (ZS 24), booting (ZS 49), and complete emergence of inflorescence (ZS 68), with the total amount of water, 2550 m3 ha−1 , being equivalent to 255 mm. The irrigation water was applied via the furrow system. The irrigation system had one water-emitting tube at each plot to deliver constant and equal amounts of water to each plot. The amount of water was monitored with a discharge gauge and regulated through manually operated control valves. 2.3. Grain yield and drought tolerance indices After physiological maturity, three internal rows in each subplot, each 3 m in length (1.8 m2 in total area), were harvested by hand and threshed. After that, total grain yield was weighed and expressed as tons ha−1 after adjustment to a water content of 15.5%. Based on the grain yield under the FI and LWI treatments, various drought tolerance indices (DTIs), YI, SSI, STI, TOL, MPI, and GMP, were calculated to estimate the stability in yield under FI and LWI conditions. The formulas for these indices are presented in Table 1. 2.4. Spectral reflectance measurements The absolute spectral reflectance of the canopy at the heading (ZS 59) and middle grain filling stages (ZS 75) was measured in the spectral range of 350–2500 nm using a hand-held ASD FieldSpec Spectroradiometer (Analytical Spectral Devices Inc., Boulder, CO, USA). The reflectance measurements were made on cloudless days between 10.00 and 15.00 h (in arid and semiarid regions the weather remains fairly stable between these hours in winter time), and the spectrometer was calibrated using a white plate of barium sulfate (BaSO4 ) approximately every 15 min in order to avoid problems originating from sun angles during the day. Using a 2.3 mm diameter and 25◦ full conical angle optical fiber, three shots from different locations in each plot were taken for each measurement. The optical fiber was placed approximately 80 cm above the top of the canopy in the nadir position. Because there were time differences of 3–6 days among the tested lines and irrigation treatments in reaching the heading (ZS 59) and middle grain filling (ZS 75) stages, each reading of absolute spectral reflectance of the canopy was taken at the middle of this range to ensure minimal influence on the reading. A total of 2150 continuous wavelength bands were taken per measurement and used to calculate four vegetation indices, namely, green normalized difference vegetation index (GNDVI), red normalized difference vegetation index (RNDVI), photochemical reflectance index (PRI), and dry matter content index (DMCI) and four water indices, namely, normalized water index 3 (NWI-3), normalized differences moisture index (NDMI), normalized difference infrared index (NDII), and normalized multi-band drought index (NMDI). The formulas for these indices are presented in Table 1. 2.5. Data analysis Data were subjected to analysis of variance (ANOVA) appropriate for a randomized complete block split-plot design for DTI (years and genotypes were considered fixed effects and replications were considered random effects), and a randomized complete block split- split plot design for grain yield and SRI (years, irrigation treatments, and genotypes were considered fixed effects and replications were considered random effects). For grain yield and DTI, combined analyses were performed across three years. For different SRIs, combined analyses were done across years at individual

4

S.E. El-Hendawy et al. / Agricultural Water Management 182 (2017) 1–12

Table 1 Description of the drought tolerance indices based on grain yield and the spectral reflectance indices used in this study. Drought tolerance indices

Formula

Reference

Yield index (YI) Stress susceptibility index ((SSI) Stress tolerance index (STI) Tolerance index (TOL) Mean productivity index (MPI) Geometric mean productivity (GMP)

YI = YLWI /Y´ LWI SSI = [(1 − (YLWI /YFI )/(1 − (Y´ LWI /Y´ FI )] STI = [(YFI × YLWI )/(Y´ LWI )2 ] TOL = YFI − YLWI MPI = (YFI + YLWI )/2 GMP = (YFI × YLWI )0.5

Gavuzzi et al. (1997) Fischer and Maurer (1978) Fernandez (1992) Hossain et al. (1990) Rosielle and Hambling (1981) Fernandez (1992)

Spectral reflectance indices

Formula

Reference

Green normalized difference vegetation index (GNDVI) Red normalized difference vegetation index (RNDVI) Photochemical reflectance index (PRI) Dry matter content index (DMCI) Normalized water index 3 (NWI-3) Normalized differences moisture index (NDMI) Normalized difference infrared index (NDII) Normalized multi-band drought index (NMDI)

(R780 –R550 )/(R780 + R550 ) (R780 − R670 )/(R780 + R670 ) (R531 − R570 )/(R531 + R570 ) (R2305 − R1495 )/(R2305 + R1495 ) (R970 − R880 )/(R970 + R880 ) (R2200 − R1100 )/(R2200 + R1100 ) (R860 − R1650 )/(R860 + R1650 ) R860 − (R1640 − R2130 )/R860 + (R1640 − R2130 )

Aparicio et al. (2002) Raun et al. (2001) ˜ Penuelas et al. (1997) Romero et al. (2012); Wang et al. (2013) Babar et al. (2006); Prasad et al. (2007) Lozano et al. (2007) Hardisky et al. (1983) Wang and Qu (2007) −

−

Where, YFI and YLWI are the grain yield of a wheat germplasm grown under full irrigation and limited water irrigation treatments, respectively. YFI and Y LWI are the mean grain yields of all wheat germplasms under full irrigation and limited water irrigation treatments, respectively.

growth stages and across years and growth stages under individual irrigation treatments. The best relationships of SRI with grain yield and DTI within each growth stage and irrigation treatment were analyzed using Pearson linear correlation. Genotypic correlations were used to estimate the relationship between the DTIs and the SRIs taken under the FI and LWI treatments and measured at the heading and middle grain filling stages, and between grain yield of FI and LWI and SRI within each growth stage and irrigation treatment. This correlation was calculated using the following formula: √ rg = (Covxy )/ (Varx × Vary )

(1)

where Cov and Var indicate components of covariance and variance between two traits, respectively. Heritability was estimated for each trait using mean squares of combined analysis for genotypes and error. The heritability was calculated according to Falconer (1989) using the following equation: H2 = ␴2 g/␴2 p

(2)

where ␴2 g and ␴2 p indicate genotypic and phenotypic variance, respectively.

3.2. Grain yield performance under full irrigation and limited water irrigation conditions The mean values of grain yield for the 65 recombinant inbred lines (RILs) and their two parents ranged from 4.60 to 9.71 tons ha−1 , from 3.35 to 8.81 tons ha−1 , and from 3.75 to 8.60 tons ha−1 under FI, and from 2.72 to 6.44 tons ha−1 , from 1.94 to 5.11 tons ha−1 , and from 2.39 to 5.50 tons ha−1 under LWI, in 2014, 2015, and 2016, respectively (Table 4). Under FI, out of 65 lines, the grain yield of eleven lines in 2014 and in 2015 and of fifteen lines in 2016 was higher than the grain yield of the drought-tolerant parent Sakha 93. Meanwhile, twenty-two lines in 2014, twenty-nine lines in 2015, and thirty-three lines in 2016 had lower grain yield than the drought-sensitive parent Sakah 61. Under LWI, no line had a higher grain yield than the drought-tolerant parent Sakha 93, whereas the grain yield of eleven, nineteen, and nine lines in 2014, 2015, and 2016, respectively, was lower than the yield of the drought-sensitive parent Sakah 61 (Table S1). Importantly, the performance of the tested lines for grain yield was not consistent across the two different irrigation treatments. Some lines were only favorable under FI, and their yield declined by more than 50% under LWI. Some lines yielded very well while others yielded low under both irrigation treatments. Furthermore, some lines yielded well under FI and their yield was reasonable and comparable to the drought-tolerant parent Sakha 93 under LWI (Table 4). 3.3. Drought tolerance indices (DTIs) reflect the relationships of limited water irrigation yields to full irrigation yields

3. Results 3.1. Analysis of variance Mean squares from the analysis of variance combined over years for grain yield, drought tolerance indices (DTIs), and spectral reflectance indices (SRIs) measured at the heading and middle grain filling stages, as well as combined over growth stages and years for the SRIs taken under full irrigation (FI) and limited water irrigation (LWI), are presented in Tables 2 and 3, respectively. The irrigation treatment and/or genotype main effect in the combined analysis was significant for grain yield, DTI, and all SRIs. The genotype × irrigation, genotype × growth stage, genotype × irrigation × year, and genotype × growth stage × year interactions were highly significant for all SRIs. No irrigation × year, genotype × year, or growth stage × year interactions occurred for most evaluated SRIs.

To compare the grain yield of lines under LWI with their grain yield under FI, six DTIs, YI, SSI, STI, TOL, MPI, and GMP, were calculated for each growing season (Table 4). The drought-tolerant parent Sakha 93, which yielded well under both irrigation treatments, exhibited high values for the YI, STI, MPI, and GMP, and low values for the SSI and TOL. The reverse was true for the droughtsensitive parent Sakah 61, the yield of which declined by more than 50% under LWI (Table 4 and Tables S2, S3, and S4). The values of the DTIs for the 65 RILs depended on the yield potential under both irrigation treatments, as well as on the percentage of reductions in yield due to LWI. Generally, the RILs whose yield declined by more than 50% under LWI registered the highest values for the SSI and TOL and the lowest values for the other indices. The RILs that yielded very low under both irrigation treatments exhibited the lowest values for YI, STI, MPI, and GMP. The RILs that yielded very well under FI and yielded low under LWI tended to have higher values for all

S.E. El-Hendawy et al. / Agricultural Water Management 182 (2017) 1–12

5

Table 2 Combined analysis of variance (mean squares values) across years for grain yield and drought tolerance indices, and for spectral reflectance indices (SRIs) measured at the heading and middle grain filling stages.

Grain yield Drought tolerance indices

Heading growth stage

Middle grain filling stage

Traits

Year (Y)

Irrigation (I)

I*Y

Genotype (G)

G*Y

G*I

df

2

1

2

66

132

66

**

156.7 0.000002 0.00346 0.1939 0.8853 78.352 82.212 1.9443 4.8692 0.0289 0.0689 0.00145 0.02314 0.0691 0.05853 0.4227 0.4869 0.00403 0.02994 0.00460 0.5251 0.1528 0.7468

YI SSI STI TOL MPI GMP GNDVI RNDVI PRI DMCI NWI-3 NDMI NDII NMDI GNDVI RNDVI PRI DMCI NWI-3 NDMI NDII NMDI

2070.2

0.4426

0.6289** 0.9201* 0.2242*** 0.00011* 0.00305** 0.2082** 0.2430** 6.1951*** 4.156*** 13.579*** 3.196* 1.0069*** 0.00214** 3.0009*** 0.4485*** 7.4139***

0.2810* 0.4561 0.0141* 0.00698 0.00167* 0.0201 0.0785* 0.0623 1.173* 2.352 0.0328* 0.0335 0.0035* 1.5519 0.3565 0.1843

**

14.85 0.3993*** 0.6585*** 0.3831*** 8.7835*** 7.4266*** 7.1845*** 0.0397*** 0.0524*** 0.00212*** 0.00904*** 0.00077*** 0.0263*** 0.01618*** 0.0436*** 0.0375*** 0.0458*** 0.00235*** 0.01660*** 0.00105*** 0.1429*** 0.0669*** 0.07134***

0.2718 0.00686 0.01739 0.00643 0.2626 0.1359 0.1231 0.0061* 0.0142 0.00039 0.00044 0.00010* 0.0033 0.0029 0.0030 0.0064 0.0082* 0.00044 0.00089 0.00005* 0.0082* 0.0032 0.0047

G*I*Y 132 **

4.391

0.1312

0.0058*** 0.0089*** 0.00029*** 0.00048*** 0.00017*** 0.0067*** 0.0035*** 0.0172*** 0.0101*** 0.0093*** 0.00042*** 0.00092*** 0.00003*** 0.0062*** 0.0084*** 0.0062***

0.0045*** 0.0101*** 0.00025*** 0.00026*** 0.000096*** 0.0028*** 0.0031*** 0.0031*** 0.0056*** 0.0061*** 0.00031*** 0.00062*** 0.000039*** 0.0058*** 0.0036*** 0.0026***

The full names of the abbreviations of DTI and SRI are listed in Table 1. *, **,*** Statistically significant at p ≤ 0.05, 0.01, 0.001, respectively. Table 3 Combined analysis of variance (mean squares values) across years and growth stages for different spectral reflectance indices (SRIs) taken under full irrigation (FI) and limited water irrigation (LWI) treatments.

Full irrigation (FI)

limited water irrigation (LWI)

Traits

Year (Y)

Growth stage (GS)

GS*Y

Genotypes (G)

G*Y

G*GS

G*GS*Y

df

2

1

2

66

132

66

132

GNDVI RNDVI PRI DMCI NWI-3 NDMI NDII NMDI GNDVI RNDVI PRI DMCI NWI-3 NDMI NDII NMDI

0.4218 2.2011 0.04971 0.00343 0.00667 0.0856 0.11757 0.42271 0.5907 2.4793 0.00693 0.04068 0.00327 0.7948 0.01274 0.18801

9.6971*** 32.2702*** 0.84418*** 0.22304** 0.05257*** 20.1075*** 2.2553*** 3.5965*** 19.0062*** 70.6693*** 4.9864*** 2.2087*** 0.04855*** 5.2696*** 2.8175*** 2.7642***

2.6550* 2.8562 0.00596 0.06927 0.000754 0.05388* 0.23496* 0.40152 0.15618* 0.62771 0.01725* 0.02601 0.000481 1.1859* 0.29156* 0.03970

0.04641*** 0.05598*** 0.0030*** 0.00935*** 0.00063*** 0.06928*** 0.04537*** 0.02092*** 0.03088*** 0.03941*** 0.00179*** 0.01606*** 0.00119*** 0.07346*** 0.02704*** 0.10811***

0.00735* 0.01238 0.00072 0.00037 0.00009* 0.00649 0.00362 0.00310 0.00481 0.00823* 0.00032 0.00096 0.00006* 0.00455* 0.00249 0.00450

0.00947*** 0.01021*** 0.00014 0.00542*** 0.00013*** 0.03476*** 0.02048*** 0.00665*** 0.00629*** 0.01069*** 0.00024*** 0.0011*** 0.00007*** 0.00461*** 0.00207*** 0.00271***

0.00702*** 0.01291*** 0.000117 0.00027*** 0.000083*** 0.00537*** 0.00515*** 0.00336*** 0.00336*** 0.00503*** 0.00023*** 0.00062*** 0.000051*** 0.00359*** 0.00155*** 0.00250***

The full names of the abbreviations of SRIs are listed in Table 1. *, **,*** Statistically significant at p ≤ 0.05, 0.01, 0.001, respectively.

DTIs compared to the drought-sensitive parent Sakah 61. The RILs that had desirable yields under both irrigation treatments had comparable DTI values, as did the drought-tolerant parent Sakha 93 (Table 4). 3.4. Relationships between drought tolerance indices (DTIs) and spectral reflectance indices (SRIs) The relationships between DTI and SRI were calculated after combining the data from the three growing seasons (Table 5). In general, the SRIs did not exhibit any relationship with the SSI and TOL indices. At the heading stage, all the SRIs calculated from measurements taken under LWI correlated better with YI, STI, MPI, and GMP than did those calculated from measurements taken under FI. At the middle grain filling stage, the water-SRI taken under LWI exhibited higher coefficients of determination than the vegetation-

SRI taken under the same conditions with YI, STI, MPI, and GMP. In contrast, the strength of the determination coefficients of the relationships between the vegetation-SRI taken under FI and the YI, STI, MPI, and GMP were significant and comparable to those of the water-SRI taken under the same conditions (Table 5). The DMCI was the only vegetation-SRI to show significant relationships with YI, STI, MPI, and GMP under both irrigation treatments and at both phenological growth stages. 3.5. Relationships between grain yield and spectral reflectance indices (SRIs) At the heading stage, all the SRIs that were calculated from measurements taken under FI had a weak relationship with grain yield for both the FI and LWI treatments (R2 values ranging from 0.12 to 0.39), except for the DMCI, which exhibited a better fit with the

6

S.E. El-Hendawy et al. / Agricultural Water Management 182 (2017) 1–12

Table 4 Range of the three replications and means ± standard deviation of the sixty five recombinant inbred lines (RILs) and their two parents for grain yield under full irrigation (FI) and limited water irrigation (LWI) and drought tolerance indices during three growing years. Years

Genotypes

2014

Sakha 93 Sakha 61 RILs

2015

Sakha 93 Sakha 61 RILs

2016

Sakha 93 Sakha 61 RILs

Basic statistics

Range Mean Range Mean Range Mean Range Mean Range Mean Range Mean Range Mean Range Mean Range Mean

Grain yield (ton ha−1 )

Drought tolerance indices

FI

LWI

YI

SSI

STI

TOL

MPI

GMP

8.17–8.89 8.61 ± 0.39 6.42–7.20 6.79 ± 0.39 4.60–9.71 7.29 ± 1.26 7.12–7.39 7.23 ± 0.14 5.89–6.30 6.04 ± 0.23 3.35–8.81 6.00 ± 1.24 7.30–7.71 7.46 ± 0.22 5.98–7.50 6.70 ± 0.76 3.75–8.60 6.49 ± 1.21

6.30–6.80 6.53 ± 0.25 3.43–4.14 3.74 ± 0.36 2.72–6.44 4.63 ± 0.90 5.28–5.48 5.38 ± 0.10 2.90–3.06 2.96 ± 0.10 1.94–5.11 3.45 ± 0.78 5.49–5.56 5.57 ± 0.09 2.77–2.90 2.85 ± 0.07 2.39–5.50 3.83 ± 0.78

1.35–1.45 1.40 ± 0.05 0.74–0.89 0.80 ± 0.08 0.58–1.39 1.00 ± 0.19 1.52–1.59 1.55 ± 0.04 0.83–0.87 0.85 ± 0.02 0.56–1.47 0.99 ± 0.22 1.40–1.48 1.45 ± 0.04 0.71–0.76 0.74 ± 0.03 0.62–1.43 1.00 ± 0.20

0.66–0.68 0.67 ± 0.01 1.12–1.42 1.24 ± 0.16 0.43–1.84 0.98 ± 0.29 0.55–0.67 0.60 ± 0.06 1.13–1.28 1.21 ± 0.08 0.41–1.72 0.98 ± 0.28 0.55–0.71 0.62 ± 0.08 1.32–1.49 1.40 ± 0.09 0.53–1.62 0.98 ± 0.26

1.01–1.15 1.06 ± 0.08 0.42–0.57 0.48 ± 0.08 0.26–1.02 0.65 ± 0.20 1.04–1.11 1.07 ± 0.04 0.47–0.52 0.49 ± 0.03 0.20–0.98 0.59 ± 0.21 0.96–1.00 0.98 ± 0.02 0.38–0.52 0.45 ± 0.07 0.23–1.00 0.60 ± 0.20

1.87–2.29 2.08 ± 0.21 2.77–3.32 3.05 ± 0.28 0.76–5.64 2.66 ± 1.04 1.64–2.11 1.85 ± 0.24 2.83–3.39 3.08 ± 0.28 0.81–5.44 2.55 ± 1.05 1.64–2.22 1.89 ± 0.30 3.21–4.60 3.85 ± 0.70 0.89–5.36 2.66 ± 0.99

7.24–7.85 7.57 ± 0.31 5.03–5.67 5.26 ± 0.35 3.79–7.55 5.96 ± 0.97 6.28–6.34 6.30 ± 0.03 4.42–4.61 4.50 ± 0.10 2.74–6.32 4.73 ± 0.89 6.46–6.60 6.51 ± 0.08 4.37–5.20 4.78 ± 0.42 3.18–6.72 5.16 ± 0.89

7.17–7.78 7.50 ± 0.31 4.81–5.46 5.04 ± 0.37 3.70–7.37 5.79 ± 0.95 6.21–6.25 6.23 ± 0.02 4.15–4.28 4.22 ± 0.07 2.66–5.95 4.52 ± 0.87 6.40–6.51 6.44 ± 0.57 4.07–4.66 4.37 ± 0.30 3.12–6.51 4.96 ± 0.87

The full names of the abbreviations of DTIs are listed in Table 1.

Table 5 Correlation coefficients between of the relationships between drought tolerance indices (DTIs) and spectral reflectance indices (SRIs) taken under full irrigation (FI) and limited water irrigation (LWI) treatments and measured at heading and middle grain filling stages. The data are averaged over three years. Coefficients of determination are indicated. Water treatments

SRI

Heading growth stage

Middle grain filling stage

YI

SSI

STI

TOL

MPI

GMP

YI

SSI

STI

TOL

MPI

GMP

Vegetation-SRI 0.40* 0.36* 0.42* 0.68**

0.001 0.002 0.004 0.002

0.57** 0.53* 0.65** 0.90***

0.14 0.13 0.16 0.21

0.62** 0.57** 0.66** 0.93***

0.58* 0.56* 0.65** 0.92***

0.60** 0.62** 0.56* 0.69**

0.002 0.001 0.001 0.006

0.77*** 0.83*** 0.80*** 0.86***

0.14 0.16 0.13 0.12

0.83*** 0.88*** 0.78*** 0.90***

0.81*** 0.86*** 0.77*** 0.90***

0.03 0.02 0.02 0.01

0.84*** 0.84*** 0.90*** 0.94***

0.07 0.09 0.09 0.11

0.84*** 0.87*** 0.89*** 0.93***

0.86*** 0.87*** 0.90*** 0.94***

0.46* 0.34* 0.49* 0.56*

0.005 0.002 0.008 0.003

0.60** 0.49* 0.66** 0.80***

0.06 0.07 0.08 0.20

0.57** 0.46* 0.63** 0.87***

0.58** 0.46* 0.64** 0.84***

0.03 0.06 0.08 0.06

0.57* 0.48* 0.46* 0.45*

0.03 0.01 0.01 0.01

0.57** 0.47* 0.44* 0.43*

0.60** 0.50* 0.48* 0.46*

0.73*** 0.73*** 0.70*** 0.71***

0.01 0.01 0.01 0.01

0.95*** 0.93*** 0.93*** 0.92***

0.12 0.11 0.12 0.10

0.94*** 0.92*** 0.91*** 0.89***

0.94*** 0.93*** 0.91*** 0.90***

0.02 0.01 0.03 0.02

0.87*** 0.89*** 0.85*** 0.95***

0.10 0.09 0.07 0.01

0.88*** 0.91*** 0.85*** 0.93***

0.88*** 0.92*** 0.86*** 0.94***

0.70*** 0.66** 0.67** 0.72***

0.01 0.01 0.01 0.01

0.90*** 0.89*** 0.88*** 0.92***

0.09 0.15 0.12 0.11

0.86*** 0.91*** 0.88*** 0.92***

0.87*** 0.90*** 0.88*** 0.92***

FI

GNDVI RNDVI PRI DMCI

LWI

GNDVI RNDVI PRI DMCI

FI

NWI-3 NDMI NDII NMDI

0.71*** 0.70*** 0.72*** 0.74*** Water-SRI 0.52** 0.48* 0.50** 0.44*

LWI

NWI-3 NDMI NDII NMDI

0.70*** 0.72*** 0.72*** 0.75***

The full names of the abbreviations of DTIs and SRIs are listed in Table 1. *, **, *** Statistically significant at p ≤ 0.05, 0.01, 0.001, respectively.

grain yield for both the FI and LWI treatments (R2 values ranging from 0.62 to 0.71). The vegetation-SRIs and water-SRIs that were calculated from measurements taken under LWI exhibited significant and comparable coefficients of determination with the grain yield for both the FI and LWI treatments, but the DMCI and NMDI correlated better with grain yield than did the other SRIs (Table 6). At the middle grain filling stage, the GNDVI, RNDVI, and PRI measured under LWI were the only SRIs having weak relationships with grain yield for both the FI and LWI treatments (R2 values ranging from 0.16 to 0.36). Comparing all SRIs, in general, the waterSRIs exhibited higher determination coefficients (R2 values ranging from 0.51 to 0.71) with grain yield for both the FI and LWI treatments than the vegetation-SRIs (R2 values ranging from 0.16 to 0.66), except for the DMCI, which had comparable coefficients of determination (R2 values ranging from 0.53 to 0.70) to those of the water-SRIs. Considering the water-SRIs together, the NMDI demon-

strated higher levels of relationship with grain yield than the other water-SRIs (Table 6). 3.6. Relationships of mean SRI values to grain yield over two growth stages, three years, and two irrigation treatments Fig. 1 shows the functional relationship of grain yield for both the FI and LWI treatments to the mean SRI values over two growth stages, three years, and two irrigation treatments. In general, in comparison with any individual measurement, all the SRIs efficiently explained the grain yield variability when the data of the SRIs for growth stages, years, and irrigation treatments were combined. The DMCI was the index that correlated best with grain yield for the FI treatment (R2 = 0.81), followed by the RNDVI (R2 = 0.77), NDMI (R2 = 0.76), and GNDVI (R2 = 0.75). All water-SRIs provided higher coefficients of determination for the relationships with grain

S.E. El-Hendawy et al. / Agricultural Water Management 182 (2017) 1–12

7

Table 6 Correlation coefficients between of the relationships between grain yield measured under full irrigation (FI) and limited water irrigation (LWI) treatments and spectral reflectance indices (SRIs) taken under FI and LWI treatments and measured at the heading and middle grain filling stages in 2014, 2015 and 2016. Water treatments

SRI

Heading growth stage

Middle grain filling stage

Grain yield under FI 2014

Grain yield under LWI

Grain yield under FI

Grain yield under LWI

2015

2016

2014

2015

2016

2014

2015

2016

2014

2015

2016

Vegetation-SRI FI

GNDVI RNDVI PRI DMCI

0.28 0.31 0.39 0.69***

0.32 0.38 0.38 0.66**

0.38 0.35 0.38 0.72***

0.21 0.24 0.33 0.62**

0.27 0.21 0.34 0.69***

0.28 0.27 0.20 0.63**

0.59* 0.66** 0.50* 0.69***

0.63** 0.61** 0.55* 0.66**

0.57* 0.61** 0.51* 0.70***

0.55* 0.51* 0.51* 0.66**

0.50* 0.50* 0.44* 0.65**

0.41* 0.48* 0.39* 0.63**

LWI

GNDVI RNDVI PRI DMCI

0.60** 0.53* 0.56* 0.67**

0.50* 0.59* 0.60** 0.65**

0.66** 0.67** 0.65** 0.71***

0.60*** 0.55* 0.63** 0.66**

0.62** 0.56* 0.58* 0.65**

0.18 0.20 0.16 0.67**

0.31 0.29 0.36 0.69**

0.33 0.27 0.33 0.67**

0.21 0.22 0.20 0.50*

0.26 0.19 0.24 0.53*

0.29 0.18 0.34 0.50*

FI

NWI-3 NDMI NDII NMDI

0.56* 0.54* 0.66** 0.69** Water-SRI 0.21 0.12 0.15 0.17

0.18 0.18 0.19 0.15

0.24 0.14 0.12 0.13

0.22 0.17 0.24 0.26

0.27 0.24 0.29 0.20

0.29 0.28 0.28 0.29

0.66** 0.66** 0.65** 0.69**

0.65** 0.66** 0.66** 0.67**

0.65** 0.68** 0.64** 0.71***

0.63** 0.66** 0.61** 0.70***

0.65** 0.65** 0.63** 0.68**

0.64** 0.65** 0.63** 0.66**

LWI

NWI-3 NDMI NDII NMDI

0.63** 0.65** 0.52* 0.68**

0.59* 0.65** 0.53* 0.66**

0.60** 0.64** 0.49* 0.68**

0.60** 0.68** 0.56* 0.71***

0.63** 0.66** 0.59* 0.69***

0.61** 0.61** 0.56* 0.71***

0.64** 0.64** 0.59* 0.70***

0.58* 0.63** 0.61** 0.68**

0.57* 0.62** 0.53* 0.65**

0.64** 0.58* 0.63** 0.66**

0.62** 0.55* 0.51* 0.69***

0.58** 0.53* 0.52* 0.68**

The full name of the abbreviations of the SRIs are listed in Table 1. *, **, *** Statistically significant at p ≤ 0.05, 0.01, 0.001, respectively.

yield for the LWI (R2 ≈ 0.77) compared to the well-known vegetation indices (GNDVI, RNDVI, and PRI) (Fig. 1). 3.7. Genotypic correlations between SRI and DTI and grain yield Tables 7 and 8 demonstrate the genetic correlations between DTIs and SRIs, and between grain yield and SRIs, respectively, using correlation coefficients calculated from values averaged over three years for each trait. In general, all SRIs measured under FI and LWI treatments and measured at the heading and middle grain filling stages had significant positive and negative genotypic correlations with all DTIs, except for the SSI and TOL, and with grain yield for both the FI and LWI treatments. Genotypic correlations increased between DTI or grain yield and the vegetation-SRIs taken under FI, as the growth of genotypes transitioned from the heading stage to the middle grain filling stage; the opposite held true for the vegetation-SRIs calculated under LWI. All water-SRIs calculated under LWI showed a strong genotypic correlation with the DTIs and grain yield at both growth stages, whereas the genotypic correlations between the water-SRIs calculated under FI and the DTIs and grain yield was much stronger at the middle grain filling stages than at the heading growth stage (Tables 7 and 8). 3.8. Broad-sense heritability of traits The broad-sense heritability (H2 ) was estimated across years for grain yield and the DTIs (Table 9), and across years and irrigation treatments or across years and growth stages for all SRIs (Table 10). The grain yield and all DTIs gave the highest values of heritability. The heritability values for grain yield and DTI ranged from 78.3 to 92.6% (Table 9). At the heading growth stage, the SRI showing a high value of heritability was the DMCI (78.9%). Meanwhile, at the middle grain filling stage, the highest values for heritability were found for the DMCI (74.1%) and all water-SRIs (H2 ranged from 66.1 to 74.6%). High values of heritability for the SRIs calculated under FI were recorded for the PRI and DMCI, while for SRIs calculated under LWI, high values were recorded for the DMCI and all water-SRIs (Table 10).

4. Discussion Most studies point out that a high yield potential under optimal conditions does not necessarily result in improved yield under limited water irrigation. That observation agrees with the results of our study where the performance of the tested lines and their parents for grain yield was not consistent across the two irrigation treatments. Because of this fact, the use of drought tolerance indices (DTIs) has been recommended by several researchers in order to take into account the observed values of the target trait under both normal and stress conditions, and to integrate that information across treatments when evaluating the responses of genotypes to stresses (Singh et al., 2011; Raman et al., 2012). However, the effectiveness of DTIs in the evaluation process and in discriminating among genotypes with regard to drought tolerance is based on the relative yield performance of genotypes in both normal and stress conditions together. For instance, the STI was able to identify genotypes that yielded well under both normal and stress conditions (Fernandez, 1992). Thus, a high STI value implies a higher stress tolerance and yield potential for a given genotype. The reverse is true for SSI and TOL. Rizza et al. (2004) reported that a selection based on minimum yield decrease under limited water irrigation with respect to normal conditions (TOL) failed to identify the best genotypes. The MPI and GMP may be useful for identifying lines that yield well under normal conditions and yield reasonably well under limited water irrigation (Hossain et al., 1990; Raman et al., 2012). It seems from the results of the DTIs that these indices are more useful for identifying drought tolerant genotypes and for integrating the information of a given genotype across treatments and years if: (1) the tested genotypes produce high yield under FI and yield reasonably well under LWI; and (2) the differences in grain yield between FI and LWI are not too large. Based on these two reasons, the results of this study indicate that the YI, STI, MPI, and GMP indices can be used as the best predictors to identify genotypes that produce desirable yields in both normal and stress conditions. In agreement with this finding, other studies have reported that the YI, STI, MPI, and GMP indices were the more accurate criteria for

S.E. El-Hendawy et al. / Agricultural Water Management 182 (2017) 1–12

Grain yield (ton ha-1)

Grain yield (ton ha-1)

Grain yield (ton ha-1)

Grain yield (ton ha-1)

8

12 11 10 9 8 7 6 5 4 3 2 1 24 0 - . 12 11 10 9 8 7 6 5 4 3 2 1 02 -0 . 12 11 10 9 8 7 6 5 4 3 2 1 0 0 .4 12 11 10 9 8 7 6 5 4 3 2 1 5 0 .3

2

-0.

22

-0 .

20

18 16 -0. -0 . DM CI

-0 .

14

0 .3

2

0 .3

6

0 .4

FI y = 5.88 + 93.4x r 2 = 0.73 * ** 2 ** L W I y = 3.51 + 60 .6x r = 0.64

-0.

01

0 .0

1 2 0 .0 0 .0 PRI

0

0 .0

3

0 .0

5

0 .5

0

5 0 .5 RNDVI

0 .6

0

0 .6

5

4

0 .2

0

0 .4

5

0 0 .5 GNDVI

0 .5

5

0 .6

0

4 0 .4 NMDI

0 .4

8

0 .5

2

0 .5

6

0

0 .2

2

0 .2

4

0 .2

6

2 8 8 4 6 0 0 .2 0 .3 0 .3 0 .3 0 .3 0 .3 N D II

FI y = -1.43 - 16 .7 x r 2 = 0.76 * ** L W I y = -1.5 4 - 11.5x r 2 = 0.74 ** *

-0 .

FI y = -4.79 + 23.3x r 2 = 0.75 *** 2 * ** L W I y = -3.54 + 15 .4x r = 0.69

0 .4

0

FI y = -0.26 + 23 .4x r 2 = 0 .74 * ** 2 * ** L W I y = -0.8 7 + 16 .6x r = 0.77

FI y = -4 .76 + 21 .3x r 2 = 0.77 ** * L W I y = -3.27 + 13 .6x r 2 = 0.65 * *

0 .4

* **

FI y = -1.8 6 + 18.7 x r = 0.74 L W I y = -2.02 + 13.3x r 2 = 0.77 ***

FI y = -1.11 - 41 .3x r 2 = 0.81 ** * 2 ** * L W I y = -0.99 - 2 6.6x r = 0.70

-0 .

60

-0 .

55

-0 .

50

45 -0 . NDMI

-0.

40

-0 .

35

-0.

30

FI y = 1 .48 - 149 .7x r 2 = 0 .74 * ** 2 ** * L W I y = 0.37 - 105 .6 x r = 0.77

0 5 0 0 .0 4 5 0 .0 4 0 0 .0 3 5 0 .0 3 0 0 .0 2 5 0 .0 2 0 0 .0 1 5 N W I-3

Fig. 1. Relationship between grain yield (y) for full irrigation (FI) and limited water irrigation (LWI) with different spectral reflectance indices (SRIs) (x). Grain yield was estimated as the mean of three years whereas the SRIs was estimated as mean value of two growth stages, three years and two irrigation treatments. The full name of the abbreviation of the SRI are listed in Table 1. **, *** Statistically significant at p ≤ 0.01, 0.001, respectively.

selecting genotypes with high grain yield under both normal and stress conditions because they take into account the yield potential under both conditions (Drikvand et al., 2012; Cabello et al., 2013; Eivazi et al., 2013). To calculate these DTIs based on grain yield, one must wait until harvesting the experimental plots in order to determine the grain yield, and one also must test the yield performance under both favorable and stress conditions, which is time consuming and labor intensive, especially when plant breeders need to evaluate a large number of entries. However, on the other hand, the main advan-

tage for the estimation of the DTI rather than of the grain yield per se is that these indices provide better opportunities for breeders to select genotypes that do well under all situations ranging from irrigated control to limited water irrigation. Therefore, early prediction of the DTI through early estimation of the grain yield by rapid and non-destructive reflectance measurements would be useful for plant breeders in the early generations of breeding programs as they attempt to select genotypes with higher productive capacity under a wide range of environments.

S.E. El-Hendawy et al. / Agricultural Water Management 182 (2017) 1–12

9

Table 7 The genotypic correlation (rg ) between drought tolerance indices (DTI) and spectral reflectance indices (SRIs) taken under full irrigation (FI) and limited water irrigation (LWI) treatments and measured at the heading and middle grain filling stages. Water treatments

SRI

Heading growth stage

Middle grain filling stage

YI

SSI

STI

TOL

MPI

GMP

YI

SSI

STI

TOL

MPI

GMP

Vegetation-SRI 0.65* 0.61* 0.66* −0.84**

0.03 0.04 0.06 0.05

0.78** 0.75** 0.82** −0.96***

0.39 0.37 0.40 −0.38

0.81*** 0.77** 0.82*** −0.98***

0.79** 0.76** 0.81** −0.97***

0.78** 0.81** 0.83** −0.84**

−0.04 −0.01 −0.01 0.08

0.89** 0.94*** 0.99*** −0.95***

0.38 0.41 0.41 −0.35

0.93*** 0.97*** 0.98*** −0.97***

0.92*** 0.95*** 0.98*** −0.97***

−0.17 −0.14 −0.13 0.12

0.94*** 0.94*** 0.96*** −0.98***

0.27 0.30 0.31 −0.33

0.94*** 0.95*** 0.96*** −0.98***

0.95*** 0.96*** 0.97*** −0.99***

0.69* 0.60* 0.71** −0.76**

−0.08 −0.05 −0.09 −0.05

0.79** 0.72** 0.83** −0.91***

0.25 0.28 0.29 −0.46

0.77** 0.70** 0.81** −0.95**

0.78** 0.70** 0.82** −0.93***

0.19 0.24 −0.30 −0.25

−0.78** −0.71** 0.70** 0.68*

−0.18 −0.10 0.04 0.08

−0.78** −0.70** 0.69** 0.66**

−0.80** −0.72** 0.71** 0.69**

−0.87** −0.86** 0.86** 0.86**

0.10 0.10 −0.08 −0.11

−0.99*** −0.98*** 0.99*** 0.98***

−0.35 −0.33 0.35 0.31

−0.99*** −0.97*** 0.98*** 0.96***

−0.99*** −0.98*** 0.98*** 0.97***

0.12 0.11 −0.17 −0.13

−0.95*** −0.96*** 0.94*** 0.99***

−0.32 −0.32 0.27 0.32

−0.96*** −0.97*** 0.94*** 0.98***

−0.96*** −0.97*** 0.95*** 0.98***

−0.85** −0.82** 0.83** 0.86**

0.11 0.03 −0.06 −0.09

−0.97*** −0.96*** 0.96*** 0.97***

−0.30 −0.39 0.35 0.33

−0.95*** −0.97*** 0.96*** 0.97***

−0.95*** −0.96*** 0.96*** 0.98***

FI

GNDVI RNDVI PRI DMCI

LWI

GNDVI RNDVI PRI DMCI

FI

NWI-3 NDMI NDII NMDI

0.86** 0.85** 0.86** −0.87** Water-SRI −0.75** −0.70** 0.72** 0.68*

LWI

NWI-3 NDMI NDII NMDI

−0.85** −0.86** 0.86** 0.88***

The full names of the abbreviations of DTIs and SRIs are listed in Table 1. *, **,*** Statistically significant at p ≤ 0.05, 0.01, 0.001, respectively. Table 8 The genotypic correlation (rg ) between grain yield measured under full irrigation (FI) and limited water irrigation (LWI) treatments and spectral reflectance indices (SRIs) taken under FI and SWS treatments and measured at the heading and middle grain filling stages. Water treatments

SRI

Heading growth stage

Middle grain filling stage

Grain yield under FI

Grain yield under LWI

Grain yield under FI

Grain yield under LWI

Vegetation-SRI 0.77** 0.74** 0.79** −0.90***

0.65* 0.62* 0.66* −0.84**

0.86** 0.90*** 0.91*** −0.88**

0.78** 0.81** 0.83** −0.85**

0.86** 0.86** 0.86** −0.84**

0.68* 0.64* 0.74** −0.91***

0.69** 0.60* 0.71** −0.76**

−0.75** −0.70** 0.72** 0.68**

−0.89*** −0.87*** 0.89** 0.86**

−0.87*** −0.86*** 0.86** 0.86**

−0.85** −0.86** 0.86** 0.88**

−0.84** −0.89*** 0.87** 0.87**

−0.85** −0.82** 0.83** 0.86**

FI

GNDVI RNDVI PRI DMCI

LWI

GNDVI RNDVI PRI DMCI

FI

NWI-3 NDMI NDII NMDI

0.82** 0.85** 0.86** −0.82** Water-SRI −0.67** −0.57* 0.54* 0.54*

LWI

NWI-3 NDMI NDII NMDI

−0.86** −0.86** 0.82** 0.87**

The full name of the abbreviations of the SRIs are listed in Table 1. *, **, *** Statistically significant at p ≤ 0.05, 0.01, 0.001, respectively. Table 9 Genotypic variance (␴2 g ), environmental variance (␴2 e ), phenotypic variance (␴2 p ) and broad sense heritability across years for grain yield and drought tolerance indices (DTIs).

Grain yield Drought tolerance indices

Traits

␴2 g

␴2 e

␴2 p

Heritability (%)

YI SSI STI Tol MPI GMP

1.6354 0.0436 0.0712 0.0419 0.9468 0.8101 0.7846

0.1312 0.0069 0.0174 0.0064 0.2626 0.1359 0.1231

1.7666 0.0505 0.0886 0.0483 1.2094 0.9460 0.9077

92.6 86.4 80.4 86.7 78.3 85.6 86.4

The full name of the abbreviations of the SRIs are listed in Table 1.

4.1. Relationship of drought tolerance indices and spectral indices The drought tolerance indices SSI and TOL were not related to the spectral indices (water- or vegetation-based indices), whereas

the index YI was moderately related to the spectral indices. Generally, the drought tolerance indices STI, MPI, and GMP were best related to the spectral indices, although with varying degrees of strength.

10

S.E. El-Hendawy et al. / Agricultural Water Management 182 (2017) 1–12

Table 10 Genotypic variance (␴2 g ), environmental variance (␴2 e ), phenotypic variance (␴2 p ) and broad sense heritability across years and irrigation water treatments or across years and growth stages for different spectral reflectance indices.

Heading growth stage

Middle grain filling stage

FI

LWI

Traits

␴2 g

␴2 e

␴2 p

Heritability (%)

GNDVI RNDVI PRI DMCI NWI-3 NDMI NDII NMDI GNDVI RNDVI PRI DMCI NWI-3 NDMI NDII NMDI GNDVI RNDVI PRI DMCI NWI-3 NDMI NDII NMDI GNDVI RNDVI PRI DMCI NWI-3 NDMI NDII NMDI

0.003911 0.004700 0.000206 0.000971 0.000075 0.002611 0.001453 0.004500 0.003544 0.004411 0.000227 0.001776 0.000107 0.015233 0.007033 0.007638 0.004377 0.004786 0.000320 0.001009 0.000061 0.007101 0.004469 0.001951 0.003058 0.003820 0.000173 0.001716 0.000127 0.007763 0.002832 0.011734

0.004500 0.010100 0.000250 0.000260 0.000096 0.002800 0.003100 0.003100 0.005600 0.006100 0.000310 0.000620 0.000039 0.005800 0.003600 0.002600 0.007020 0.012910 0.000117 0.000270 0.000083 0.005370 0.005150 0.003360 0.003360 0.005030 0.000230 0.000620 0.000051 0.003590 0.001550 0.002500

0.008411 0.014800 0.000456 0.001231 0.000171 0.005411 0.004553 0.007600 0.009144 0.010511 0.000537 0.002396 0.000146 0.021033 0.010633 0.010238 0.011397 0.017696 0.000437 0.001279 0.000144 0.012471 0.009619 0.005311 0.006418 0.008850 0.000403 0.002336 0.000178 0.011353 0.004382 0.014234

46.5 31.8 45.1 78.9 43.8 48.3 31.9 59.2 38.8 42.0 42.2 74.1 73.2 72.4 66.1 74.6 38.4 27.0 73.2 78.9 42.3 56.9 46.5 36.7 47.6 43.2 43.0 73.5 71.3 68.4 64.6 82.4

The full name of the abbreviations of the SRIs are listed in Table 1.

The best performing and most consistent index across the two growth stages under full and limited water irrigation was the vegetation index DMCI. Generally, the investigated water indices performed even better at the middle grain filling stage under both full irrigation and limited water irrigation, and comparably well under limited water irrigation at the heading stage, but yielded only moderate relationships under full irrigation. It seems therefore that the DMCI is the most preferred index across the two water regimes and development growth stages, since it is most closely related to selected drought tolerance indices, particularly to the STI, MPI, and GMP. The overall relationships can further be improved by combining the DMCI, which is most suitable for the heading stage, with the best performing water stress indices, NWI-3 and NDMI, for the middle grain stage across both water regimes, respectively. These results indicate that it is indeed possible by reflectance measurements to make early estimates of the DTI that take into account the yield potential of a genotype under both normal and stress conditions, and that have the ability to identify breeding lines that produce yield reasonably well under all drought conditions.

4.2. Relationships between grain yield and spectral indices Developing higher-yielding genotypes of wheat still remains the main target for breeders as it is directly linked to both attainable and on-farm yields. Different studies have reported that early prediction of grain yield is indeed possible under normal and/or stress conditions by measuring spectral reflectance indices prior to harvest (Prasad et al., 2007; Weber et al., 2012; Lobos et al., 2014; El-Hendawy et al., 2015a; Elsayed et al., 2015; Elazab et al., 2015). Similar conclusions, as described above for the relationship between spectral indices and drought tolerance indices, can be

drawn regarding the relationship between spectral indices and grain yield across the three years of our study. Again, the vegetation index DMCI was, across the two growth stages and the two water regimes, the best performing one. A comparable performance was also observed for the water-based index NDMI, which best predicted the yield under both water scenarios except at the heading growth stage. This index surpassed the DMCI index particularly under LWI at the middle grain stage. A combination of both indices at the respective stages or under the respective stress scenarios can further contribute to improved predictions of the grain yield. Our results partly agree with previous studies reporting that the usefulness of the spectral reflectance indices in predicting grain yield is enhanced at the grain filling stage of the crop, especially when the SRIs were calculated from measurements taken under well-watered conditions. For example, Lobos et al. (2014) found that the association between SRI and grain yield of spring bread wheat was increased and stronger at the middle grain filling stage compared to the heading stage, when the LAI decreased to values around 2 and the plants started to become senescent. By contrast, our results reveal that vegetation-based indices like the DMCI perform well at the heading stage, whereas the water-stressbased indices fail to identify relationships under full irrigation. For instance, the wavelength 2305 nm was found to best estimate ˜ et al., 2005). the dry matter content of 37 different species (Riano Romero et al. (2012) also found that the wavelengths 1495 nm and 2305 nm were well correlated with the dry matter content of tree leaves. The improved performance of the water-based indices at the grain filling stage is probably related to the senescence of the plants and thus to decreasing water content. Over the three years of this study, in general, the indices based on near infrared and shortwave infrared consistently demonstrated closer relationships with grain yield compared with the indices

S.E. El-Hendawy et al. / Agricultural Water Management 182 (2017) 1–12

based only on visible wavelengths (Table 6). This result may be because of the fact that the indices based on NIR and SWIR were more related to the water status of the plants, which is a critical issue under arid conditions, and even under FI because of the high rate of evapotranspiration. The NIR wavelengths have the ability to penetrate more into the canopy and are more sensitive to changes in the internal structure of the leaf (Campbell, 2002). They can therefore assess the water status of the canopy as well as morphological changes in the internal anatomy related to turgidity (Yao et al., 2014). The wavelengths within the SWIR penetrate less far into the canopy, which makes them less sensitive to LAI changes and more sensitive to changes in leaf water content (Ceccato et al., 2002). These reasons may explain why the indices based on NIR and SWIR in this study, such as all the water-SRIs, gave good predictions of grain yield in both phenological stages, and especially under LWI treatment, when compared with the indices based on VIS only. In this study, it was evident that the mean grain yield (mean of three years) showed very strong associations with the overall mean for the SRIs (averaged over the two growth stages, three years, and two irrigation treatments), and the determination coefficients for these associations were higher than those calculated for the individual years and growth stages (Fig. 1). Combining the three years identifies the DMCI as the single best-performing index across the two growth stages and the two water regimes, being surpassed only by the NMDI and related water stress indices under LWI. The extended range of values resulting from the combination of values has probably contributed to this observation. This result reinforces the results of other studies that found very strong associations between grain yield and SRI when more dates were averaged over growth stages and years (Babar et al., 2006; Prasad et al., 2007; Lobos et al., 2014).

4.3. Overall relationships of spectral indices to drought tolerance indices and grain yield This study identified the spectral indices most closely related to both drought tolerance indices and grain yield. Given that both parameters were best described by either the DMCI or the NDMI, or a combination of both, a powerful combination of spectral indices can allow us to deduce optimized tolerance indices that are closely related to final yield under full irrigation and limited water irrigation. It is advantageous that already at the heading growth stage, enhanced relationships exist for both the drought tolerance indices and the final grain yield, and these are more closely related at that time than at the middle grain filling stage.

5. Conclusions Our results show that selection based on the drought tolerance indices YI, STI, MPI, and GMP will make it possible to identify wheat genotypes that produce desirable yields in both normal and stress conditions. The differences in these drought tolerance indices and in grain yield for advanced breeding wheat lines can be predicted using SRIs. The relative importance of the SRIs in predicting grain yield not only depends on phenological growth stage but is also related to the type of SRI (vegetation- vs. water-SRI) and to the wavelength bands incorporated into these SRIs. Very strong associations between grain yield and SRIs were observed when the SRIs were averaged over growth stages, years, and water treatments.

11

Acknowledgements The authors would like to extend their sincere appreciation to the Deanship of Scientific Research at King Saud University for funding this Research Group No. (RG-1435-032). Appendix A. Supplementary data Supplementary data associated with this article can be found, in the online version, at http://dx.doi.org/10.1016/j.agwat.2016.12. 003. References Agati, G., Foschi, L., Grossi, N., Guglielminetti, L., Cerovic, Z.G., Volterrani, M., 2013. Fluorescence-based versus reflectance proximal sensing of nitrogen content in Paspalum vaginatum and Zoysia matrella turfgrasses. Eur. J. Agron. 45, 39–51. Aparicio, N., Villegas, D., Araus, J.L., Casadesus, J., Royo, C., 2002. Relationship between growth traits and spectral vegetation indices in durum wheat. Crop Sci. 42, 1547–1555. Araus, J.L., Cairns, J., 2014. Field high-throughput phenotyping—the new crop breeding frontier. Trends Plant Sci. 19, 52–61. Babar, M.A., van Ginkel, M., Klatt, A.R., Prasad, B., Reynolds, M.P., 2006. The potential of using spectral reflectance indices to estimate yield in wheat grown under reduced irrigation. Euphytica 150, 155–172. Barakat, M., El-Hendawy, S., Al-Suhaibani, N., Elshafei, A., Al-Doss, A., Al-Ashkar, A., Ahmed, E., Al-Gaadi, K., 2016. The genetic basis of spectral reflectance indices in drought-stressed wheat. Acta Physiol. Plant 38, 227–238. Cabello, R., Monneveux, P., De Mendiburu, F., Bonierbale, M., 2013. Comparison of yield based drought tolerance indices in improved varieties, genetic stocks and landraces of potato (Solanum tuberosum L.). Euphytica 193, 147–156. Campbell, J.B., 2002. Introduction to Remote Sensing, third ed. Guilford Press, New York. Ceccato, P., Gobron, N., Flasse, S., Pinty, B., Tarantola, S., 2002. Designing a spectral index to estimate vegetation water content from remote sensing data: part 1: theoretical approach. Remote Sens. Environ. 82, 188–197. Cerovic, Z.G., Ben Ghozlen, N., Milhade, C., Obert, M., Debuisson, S., Le Moigne, M., 2015. Non-destructive diagnostic test for nitrogen nutrition of grapevine (Vitis vinifera L.) based on Dualex leaf-clip measurements in the field. J. Agric. Food Chem. 63, 3669–3680. Drikvand, R., Doosty, B., Hosseinpour, T., 2012. Response of rainfed wheat genotypes to drought stress using drought tolerance indices. J. Agric. Sci. 4 (7), 126–131. Eivazi, A.R., Mohammadi, S., Rezaei, M., Ashori, S., Hossien, P.F., 2013. Effective selection criteria for assessing drought tolerance indices in barley (Hordeum vulgare L.) accessions. Inter. J. Agron. Plant Prod. 4 (4), 813–821. El-Hendawy, S.E., Schmidhalter, U., 2010. Optimal coupling combinations between irrigation frequency and rate for drip-irrigated maize grown on sandy soil. Agric. Water Manag. 97, 439–448. El-Hendawy, S.E., Kotab, M.A., Al-Suhaibani, N., Schmidhalter, U., 2014. Optimal coupling combinations between the irrigation rate and glycinebetaine levels for improving yield and water use efficiency of drip-irrigated maize grown under arid conditions. Agric. Water Manage. 140, 69–78. El-Hendawy, S.E., Al-Suhaibani, N., Salem, A., Ur Rehman, S., Schmidhalter, U., 2015a. Spectral reflectance indices as a rapid nondestructive phenotyping tool for estimating different morphophysiological traits of contrasting spring wheat germplasms under arid conditions. Turk. J. Agric. For. 39, 572–587. El-Hendawy, S., Al-Suhaibani, N., Al-Gaadi, K., Ur Rehman, S., 2015b. Capability of multiple selection criteria to evaluate contrasting spring wheat germplasms under arid conditions. Pak. J. Bot. 47 (6), 2093–2105. Elazab, A., Bort, J., Zhou, B., Serret, M.D., Nieto-Taladriz, M.T., Araus, J.L., 2015. The combined use of vegetation indices and stable isotopes to predict durum wheat grain yield under contrasting water conditions. Agric. Water Manage. 158, 196–208. Elsayed, S., Rischbeck, P., Schmidhalter, U., 2015. Comparing the performance of active and passive reflectance sensors to assess the normalized relative canopy temperature and grain yield of drought-stressed barley cultivars. Field Crops Res. 177, 148–160. Erdle, K., Mistele, B., Schmidhalter, U., 2013. Spectral high-throughput assessments of phenotypic differences in biomass and nitrogen partitioning during grain filling of wheat under high yielding Western European conditions. Field Crops Res. 141, 16–26. Falconer, D.S., 1989. Introduction to Quantitative Genetics. Richard Clay Ltd., Bungay Suffolk, Great Britain, pp. 129–185. Fernandez, G.C.J., 1992. Effective selection criteria for assessing plant stress tolerance. In: Kus, E.G. (Ed.), Adaptation of Food Crop Temperature and Water Stress. Proceedings of 4th International Symposium. Asian Vegetable and Research and Development Center, Shantana Taiwan, pp. 257–270. Gavuzzi, P., Rizza, F., Palumbo, M., Campaline, R.G., Ricciardi, G.L., Borghi, B., 1997. Evaluation of field and laboratory predictors of drought and heat tolerance in winter cereals. Plant Sci. 77, 523–531.

12

S.E. El-Hendawy et al. / Agricultural Water Management 182 (2017) 1–12

Golabadi, M., Arzani, A., Mirmohammadi Maibody, S.A.M., 2006. Assessment of drought tolerance in segregating populations in durum wheat. Afr. J. Agric. Res. 1, 162–171. Gutierrez, M., Reynolds, M.P., Raun, W.R., Stone, M.L., Klatt, A.R., 2010. Spectral water indices for assessing yield in elite bread wheat genotypes in well irrigated, water stressed, and high temperature conditions. Crop Sci. 50, 197–214. Hardisky, M.A., Klemas, V., Smart, R.M., 1983. The influence of soil salinity, growth form, and leaf moisture on the spectral radiance of Spartina alterniflora canopies. Photogramm. Eng. Remote Sens. 49, 77–83. Hackl, H., Hu, Y., Schmidhalter, U., 2014. Evaluating growth platforms and stress scenarios to assess the salt tolerance of wheat plants. Funct. Plant Biol. 40, 409–424. Hossain, A.B.S., Sears, R.G., Cox, T.S., Paulsen, G.M., 1990. Desiccation tolerance and its relationship to assimilate partitioning in winter wheat. Crop Sci. 30, 622–627. Jafari, A., Paknejad, F., Al-Ahmadi, M., 2009. Evaluation of selection indices for drought tolerance of corn (Zea mays L.) hybrids. Int. J. Plant Prod. 3, 33–38. Kipp, S., Mistele, B., Schmidhalter, U., 2014. Identification of staygreen and early-senescence phenotypes in high-yielding winter wheat and their relationship to grain yield and grain protein concentration using high-throughput phenotyping techniques. Funct. Plant Biol. 41, 227–235. Lin, L., Chen, J., Cai, C., 2012. High rate of nitrogen fertilization increases the crop water stress index of corn under soil drought. Commun. Soil Sci. Plan. 43, 2865–2877. Lobos, G.A., Matus, I., Rodriguez, A., Romero-Bravo, S., Araus, J.L., Pozo, A.D., 2014. Wheat genotypic variability in grain yield and carbon isotope discrimination under Mediterranean conditions assessed by spectral reflectance. J. Integr. Plant Biol. 56, 470–479. Lozano, F.J., Suárez-Seoane, S., de Luis, E., 2007. Assessment of several spectral indices derived from multi-temporal Landsat data for fire occurrence probability modeling. Remote Sens. Environ. 107, 533–544. Ma, B.L., Dwyer, L.M., Costa, C., Cober, E.L., Morrison, M.J., 2001. Early prediction of soybean yield from canopy reflectance measurements. Agron. J. 93, 1227–1234. Marti, J., Bort, J., Slafer, G.A., Araus, J.L., 2007. Can wheat yield be assessed by early measurements of normalized difference vegetation index? Ann. Appl. Biol. 150, 253–257. Mir, R.R., Zaman-Allah, M., Sreenivasulu, N., Trethowan, R., Varshney, R.K., 2012. Integrated genomics, physiology and breeding approaches for improving drought tolerance in crops. Theor. Appl. Genet. 125, 625–645. Mohammadi, R., Armion, M., Kahrizi, D., Amri, A., 2010. Efficiency of screening techniques for evaluating durum wheat genotypes under mild drought conditions. Int. J. Plant Prod. 4 (1), 11–24. Nezhad, K.Z., Weber, W.E., Roder, M.S., Sharma, S., Lohwasser, U., Meyer, R.C., Saal, B., Borner, A., 2012. QTL analysis for thousand-grain weight under terminal drought stress in bread wheat (Triticum aestivum L.). Euphytica 186, 127–138. Payero, J.O., Tarkalson, D.D., Irmak, S., Davison, D., Petersen, L.L., 2008. Effect of irrigation amounts applied with subsurface drip irrigation on corn evapotranspiration, yield, water use efficiency, and dry matter production in a semiarid climate. Agric. Water Manage. 88, 895–908. Perry, C., Steduto, P., Allen, R.G., Burt, C.M., 2009. Increasing productivity in irrigated agriculture: agronomic constraints and hydrological realities. Agric. Water Manage. 96, 1517–1524.

˜ Penuelas, J., Isla, R., Filella, I., Araus, J.L., 1997. Visible and near-infrared reflectance assessment of salinity effects on barley. Crop Sci. 37, 198–202. Prasad, B., Carver, B.F., Stone, M.L., Babar, M.A., Raun, W.R., Klatt, A.R., 2007. Potential use of spectral reflectance indices as a selection tool for grain yield in winter wheat under Great Plains conditions. Crop Sci. 47, 1426–1440. Prathapar, S.A., 2000. Water shortages in the 21 st century. In: Cadman, H. (Ed.), The Food and Environment Tightrope. Australian Centre for International Agricultural Research, Canberra Australia, pp. 125–133. Raman, A., Verulkar, S., Mandal, N., Variar, V., Shukla, V., Dwivedi, J., Singh, B., Singh, O., Swain, P., Mall, A., Robin, S., Chandrababu, R., Jain, A., Ram, T., Hittalmani, S., Haefele, S., Hans-Peter, P., Kumar, A., 2012. Drought yield index to select high yielding rice lines under different drought stress severities. Rice 5, 31–43. Ramirez Vallejo, P., Kelly, J.D., 1998. Traits related to drought resistance in common bean. Euphytica 99, 127–136. Raun, W.R., Solie, J.B., Johnson, G.V., Stone, M.L., Lukina, E.V., Thomason, W.E., Schepers, J.S., 2001. In-season prediction of potential grain yield in winter wheat using canopy reflectance. Agron. J. 93, 131–138. ˜ D., Vaughan, P., Chuvieco, E., Zarco-Tejada, P.J., Ustin, S.L., 2005. Estimation Riano, of fuel moisture content by inversion of radiative transfer models to simulate equivalent water thickness and dry matter content: analysis at leaf and canopy level. IEEE Trans. Geosci. Remote Sens. 43, 819–826. Rizza, F., Badeckb, F.W., Cattivellia, L., 2004. Use of a water stress index to identify barley genotypes adapted to rain-fed and irrigated conditions. Crop Sci. 44, 2127–2137. Romero, A., Aguado, I., Yebra, M., 2012. Estimation of dry matter content in leaves using normalized indexes and PROSPECT model inversion. Int. J. Remote Sens. 33, 396–414. Shanahan, J.F., Schepers, J.S., Francis, D.D., Varvel, G.E., Wilhelm, W.W., Tringe, J.S., Schlemmer, M.R., Major, D.J., 2001. Use of remote sensing imagery to estimate corn grain yield. Agron. J. 93, 583–589. Sinclair, T.R., 2011. Challenges in breeding for yield increase for drought. Trends Plant Sci. 16, 289–293. Singh, B.U., Rao, K.V., Sharma, H.C., 2011. Comparison of selection indices to identify sorghum genotypes resistant to the spotted stemborer Chilo partellus (Lepidoptera: noctuidae). Int. J. Trop. Insect Sci. 31, 38–51. Tester, M., Langridge, P., 2010. Breeding technologies to increase crop production in a changing world. Science 327, 818–822. Wang, L., Raymond, E.J.R., Qu, J.J., Hao, X., Daughtry, C.S.T., 2013. Remote sensing of fuel moisture content from ratios of narrow-band vegetation water and dry-matter indices. Remote Sens. Environ. 129, 103–110. Weber, V.S., Araus, J.L., Cairns, J.E., Sanchez, C., Melchinger, A.E., Orsini, E., 2012. Prediction of grain yield using reflectance spectra of canopy and leaves in maize plants grown under different water regimes. Field Crop Res. 128, 82–90. Yao, X., Jia, W., Si, H., Guo, Z., Tian, Y., Liu, X., 2014. Exploring novel bands and key index for evaluating leaf equivalent water thickness in wheat using hyperspectra influenced by nitrogen. PLoS One 9 (6), 1–11. Zadoks, J.C., Chang, T.T., Konzak, C.F., 1974. A decimal code for the growth stages of cereals. Weeds Res. 14, 412–415.