- Research article
- Open Access
Selection and screening of drought tolerant high yielding chickpea genotypes based on physio-biochemical indices and multi-environmental yield trials
BMC Plant Biology volume 20, Article number: 171 (2020)
Chickpea is one of the major legume crops being cultivated in the arid and semi-arid regions of Pakistan. It is mainly grown on the marginal areas where, terminal drought stress is one of the serious threats to its productivity. For defining the appropriate selection criteria for screening drought tolerant chickpea genotypes, present study was conducted. Distinct chickpea germplasm was collected from different pulses breeding institutes of Pakistan and evaluated for drought tolerance at germination and early seedling stages, furthermore, at late vegetative growth stages physiochemical traits and multi-environment yield performance were also tested.
Chickpea genotypes under different environments, were significantly varied for different seedling traits, physio-chemical attributes and seed yield. Genotypes showing drought tolerance by performing better at an early seedling stages were not correspondingly high yielding. Screening for drought tolerance on seed yield basis is the most appropriate trait to develop the drought tolerant as well as high yielding chickpea genotypes. Results confirmed that traits of early growth stages were not reflecting the drought tolerance at terminal growth stages and also did not confer high yielding. NIAB-rain fed environment proved ideal in nature to screen the chickpea genotypes whereas, NIAB-lysimeter and Kalur Kot was least effective for selecting genotypes with high seed yield. Genotypes D0091–10, K010–10, D0085–10, K005–10, D0078–10, 08AG016, 08AG004, D0080–10, 09AG002, K002–10 and D0099–10 were high yielding and drought tolerant based on their performance across multiple hotspot environments.
The selected genotypes are intended for further evaluation for varietal approval to recommend for general cultivation on farmer fields in drought hit areas of Pakistan. Among physio-biochemical traits, higher proline, glycine betain, RWC and CMS were reflecting the higher capability to tolerate the drought stress in chickpea. Drought sensitive genotypes (K0037–10, 2204, K0052–10, 09AG015, K0042–10, CM709/06, K0068–10, K004–10, K0026–10 and K0063–10) were also identified in present study which were resourceful asset for using as contrasting parents in hybridization programs. To our knowledge, this is first report using an integrated approach involving, physio-biochemical indices, and multi-environmental yield trials, for comparison, screening and selection of chickpea genotypes for drought tolerance.
Crop plants are subjugated by wide spectrum of biotic and abiotic stresses which seriously hinders their optimal growth, development and reproduction. Scientific studies have proven that water deficiency adversely affected the crop growth and productivity that is a serious threat for agriculture . Presently about one-third of world’s population is living in water-deficit regions whereas and it is predicted that severity, duration and frequentness of drought stress is expected to increase due to elevated CO2 in atmosphere and climatic changes . Resilience of the legume crops in current prevailing weather extremities could be the futuristic adaptation in more severe climatic conditions .
Chickpea (Cicer arietinum L.) the second most important grain legume cultivated by resource poor farmers in the arid and semi-arid regions of the world especially Pakistan [27, 42]. It is mostly grown under rainfed conditions. Availability of water in rainfed regions is either in form of stored soil moisture in subtropical environment with summer-prevalent rainfall or only at the time of seasonal rainfall. In these conditions rainfed chickpea plantations encounters the serious yield losses due to terminal drought stress [40, 43, 44]. Drought stress is one of the major constraints for chickpea, which causes up to 50% yield losses . Drought is a complex abiotic stress, which affects various physiological and biochemical processes of crop plants. These disturbances cause growth reduction, decrease in chlorophyll contents, decrease in ascorbic acid, increase in proline accumulation, hydrogen peroxide [14, 27, 28]. These attributes can be used to screen genotypes for drought tolerance. Moreover, high relative water content (RWC) and low excised leaf water loss (ELWL) was related to drought resistance [4, 15]. So, all these are being used as screening tools in different crops.
Understanding of genetic manipulation to encounter the drought stress like, drought escape, drought avoidance and drought tolerance is necessary to develop drought tolerant or resistant crop plant [27, 39, 41]. Drought tolerance is complex phenomenon established through biosynthesis of osmolytes or compatible solutes, osmoprotectants and secondary metabolites and adjusting water relations [22, 27, 29]. Sustainable development of drought tolerant genotypes is necessary to fulfill the food demands of ever growing world population which possible through continuous evaluation of genotypes for stress tolerance in drought prone regions and selection of genotypes on the basis of yield performance [20, 30]. Key objectives of this study were to explore the drought tolerant chickpea genotypes with high yield performance for releasing as commercial cultivar.
Experimental material and Germplasm collection
Present study was comprised of the 55 chickpea genotypes (Table 1), collected from three research institutes including Nuclear Institute for Agriculture and Biology (NIAB), Faisalabad, Punjab, Pakistan, Arid Zone Research Institute (AZRI), Bakhar, Punjab, Pakistan and Pulses Research Institute (PRI), AARI, Faisalabad, Punjab, Pakistan. Performance of these chickpea genotypes for drought tolerance was evaluated through different studies to select high yielding genotypes for commercial cultivation. Experiments were conducted in growth chamber to determine physiological & biochemical index and in fields to located in hot spot regions of the country.
Growth chamber evaluation
For physiological indices studies were conducted in growth room under controlled conditions. Ten seeds of each genotype were sown in glass bowls containing polyethylene glycol (PEG-6000) of-0.60MPa solution prepared by dissolving 180 g of PEG in 1 l of distilled water [5, 31]. Seeds were placed on gauze in glass bowls (dia. 14 cm x depth 8 cm) containing 70 ml of the test solution. A control of these experiments was also maintained with 0.00 MPa. These bowls were kept in growth chamber at running 30/25 °C (day/night) temperature. Darkness was maintained during the experiment upto the seed germination. Germination data was recorded when the radicles were of 5mmin length. The germination stress tolerance index was calculated by determining the promptness index (P.I.) following the George .
P.l. = nd2 (1.00) + nd4 (0.75) + nd6 (0.50) + nd8 (0.25).
Where, nd2, nd4, nd6 and nd8were seeds germinated on the 2nd, 4th, 6thand 8th day of sowing, respectively.
Germination stress tolerance index (GSI) was calculated as described by ;
The seedlings were allowed grow for14 days, after harvesting the data for shoot and root lengths were recorded. After recording fresh weights of shoot and root, they were dried at 70 °C for 48 h in an oven and their dry weights were estimated on electrical digital balance. This data were used to calculate root and shoot length stress tolerance indices (RLSI and SLSI, respectively), fresh and dry weight stress tolerance indices (FWSI and DWSI, respectively) as under;
Cell membrane stability
Cell membrane stability (CMS) was determined according to the procedure developed by . Seeds of different genotypes were sown in two cemented tanks (3 × 3 m) filled with medium textured soil. After the period of 4 weeks water was removed from one tank while normal irrigation was applied to the other tank up to 4 weeks. Leaf samples were collected from plants grown in these two conditions. Leaf discs (1 g) of 0.5 cm size were cut from the fully developed upper leaflet placed in a 50 ml test tube which were washed three times with deionized water before imbedding in test solution of PEG. These leaflet discs were inundated in 30 ml of 40% PEG-6000 solution (T1) and in deionized water as a control (C1) then both were incubated at 10 °C for 24 h. After which the leaf discs were again washed three times with deionized water and floated in 30 ml deionized water for at 10 °C for 24 h. Then electrical conductivity (EC) of the liquid was then measured with conductivity meter. After which leaf discs l in the same solution were autoclaving for 20 min and EC was again measured at 25 °C (T2, C2) Cell membrane stability was calculated as under;
CMS = 100- Percent Injury.
Evaluation of physiological and biochemical markers
From the field studies at NIAB, Faisalabad, Pakistan, leaf samples were collected to determine different physiological and biochemical attributes necessary to estimate the genotypic variability. Following biochemical and physiological parameters were estimated:
Excised leaf water loss
After getting weight of leaves, excised leaf water loss (ELWL) was recorded in three steps,
Fresh weight (immediately after sampling)
After drying in an incubator at 28 °C at 50% R.H.(Relative humidity) for 6 h
Re-drying in an oven for 24 h at 70 °C as suggested by Clarke and Townley-Smith .
Excised leaf water loss was calculated from the following formula;
Relative water contents
Relative water contents (RWC) is possibly the proper measure of plant water status in terms of the physiological outcome of cellular water deficit. Methods of Barrs and Weatherley  were used to determine the RWC. For this parameter 100 mg leaf material was taken and kept in double distilled water in a petridish for 2 h to make the leaf tissue turgid. The turgid weight and dry weight of the leaf materials was measured and RWC was calculated with the help of formula given below.
Fresh chickpea leaves were used to determine proline contents according to the method of Bates et al. . The fresh leaf material (1 g) was homogenized in 3% aqueous solution of sulphosalicylic acid (10 ml). Leaf samples were centrifuged at speed of 12,000 rpm in centrifuge machine for 10 min and removed the residue. The supernatant was used (1 ml) was allowed to react with 1 ml of acid-ninhydrin (1.25 g ninhydrine in 30 ml glacial acetic acid and 20 ml 6 M orthophosphoric acid) p and 1 ml of glacialacetic acid in a test tube (25 ml) for 1 h at 100 °C. Reaction mixture was extracted with 10 mL toluene, which was mixed vigorously by passing a continuous stream of air for 1–2 min. Toluene was aspirated from chromophore. Aqueous phase was taken, warmed at room temperature and read at 520 nm absorbance using toluene as a blank. The proline concentration was estimated from a standard curve developed with different concentration of D-proline using following formula.
(μmole proline g− 1fresh weight =
(μg proline mL− 1x mL of toluene/115.5)/(g of sample/10)
Nitrate reductase activity
Nitrate reductase activity (NRA) in leaves of chickpea was determined according to reported method  in which potassium nitrate act as substrate. Enzyme from leaves was extracted with0.02 M phosphate buffer (pH 7.0). Fresh plant material (0.5 g) was homogenized in4.5 ml phosphate buffer (pH 7.0) and 0.5 ml of 0.02 M KNO3. Reaction mixture was incubated at 32 °C in dark for 1 h. The reaction was terminated by adding 1 ml of 1% sulfanilamide prepared in 2 N HCl and then 1 ml of 0.02% of aqueous solution of N 1-Naphthyl-ethylene diaminedihydrochloride was added. The optical density of the reaction mixture was measured at 542 nm on a spectrophotometer. Activity of NRA was determined as μmoles of NO2 g− 1F.W h− 1 by using standard curve developed with NaNO2.
Glycine betaine contents
Glycine betaine in leaf samples of chickpea genotypes were determined by using the protocol recommended by . Leaf samples of 0.5 g were finely ground and then vortexed with 20 ml deionized water for 48 h at 25oC. Filtrate from this solution was mixed with equal volume of 2 N sulphuric acid (1:1 ratio). Aliquot of 0.5 ml volume was taken and cooled for 1 h in ice water. Mixture was vortexed after the addition of 0.2 ml cold potassium iodide-iodine reagents. These samples were stored at 0-4 °C for 16 h and subsequently these were centrifuged 10,000 rpm for 15 min at 0 °C. Supernatant was discarded and then preiodite crystals were dissolved in 9 ml of 1, 2-dichloro ethane. Samples were incubated for the period of 2.0 to 2.5 h and then absorbance was determined at 365 nm with UV-visible spectrophotometer. Standard solutions of glycine betaine (50–200 μg/ml) were prepared in 2 N sulphuric acid and then subjected to spectrophotometer absorbance similar to the procedure followed for samples.
Multi-environment evaluation for seed yield
These 55 chickpea genotypes were also evaluated for yield performance across different environments for evaluating the genotype × environment interaction for grain yield. The field experiment was conducted in randomized complete block design with three replications under each environment. These targeted environments were NIAB-lysimeter, NIAB-irrigated, NIAB-rainfed, AZRI, AARI and Kaloorkot. NIAB-lysimeter, NIAB-irrigated and NIAB-rainfed environments were developed at Nuclear Institute for Agriculture and Biology, Faisalabad, by managing irrigation pattern. NIAB-lysimeter environment (triplicate) was given irrigation only at the soil preparation prior to the sowing and access of the rain water was also prevented by developing protective shelters . NIAB-irrigated environment was generated by provision of only two irrigations i.e. during field bed preparation and initiation of flowering. Rainfall was also accessible to the fields for NIAB-irrigated environment. NIAB-rainfed environment was demonstrated as only irrigation was provided at field bed preparation and second water source was only rain. Whereas, other three environments belongs to different hotspot locations for chickpea cultivation and uncertainty of water availability. Water source at these three locations was only rainfall except the initial irrigation for soil bed preparations. For comparison of yield performance well known commercial cultivars i.e. CM-2008, Noor-2009, Pb 2008 and Thall 2011 were used as local checks. Plot size was1.5 m × 1.2 m, while P × P and R × R distance was 30 cm and 15 cm, respectively. All other standard agronomic practices were followed across all of the subjected environments from sowing to harvesting except for provision of irrigation water. Upon harvesting yield per plot was calculated as g plot− 1 and converted to kg ha−1for subsequent analysis.
Data acquired from growth chamber investigation was subjected to analysis of variance (ANOVA) under completely randomized design (CRD) with three replications . Data acquired for biochemical and physiological parameters from field experiment was subjected to analysis of variance under randomized complete block design (RCBD). All of the targeted six different environments were separately subjected to analysis of variance to determine the significance of environmental effects on genotypic differences. Multi-environment yield data for different chickpea genotypes was subjected to analysis of variance under factorial treatment structure where, genotypes and environments were two different factors. Genotypic selection was most authenticated by use of Principal Component Analysis (PCA) based biplots ( a, b ;) for these studies separately and combined. GGE biplot analysis was used for assessment of pattern for genotype × environment interaction. Depiction of average environment, best environment, average performing genotype and best genotypes is facilitated by GGE biplot analysis across diverse environments [45, 46].
Partitioning of variability and assessment of significance for differences
Variability for growth chamber evaluation was partitioned into various components to determine the significance of genotypic differences for growth indices. Highly significant genotypic differences were observed for GSI, DWSI, FWSI, RLSI and SLSI in this study (Table 2). Dissection of variability for physiological and biochemical markers showed, the highly significant genotypic differences for CMS, ELWL, Electrolyte leakage, Glycine-betaine, NRA, Proline and RWC (Table 2).ANOVA for each environment separately showed that genotypes were significantly different in their yield performance at AARI, AZRI, K. Kot and NIAB-rainfed. However, genotypes were insignificantly different for yield performance under NIAB-lysimetric and NIAB-irrigated environments (Table 2). Combined ANOVA across the environments showed that genotypes, environments and genotype × environment interaction (GEI) were significantly different for grain yield (Table 3). Significance of GEI showed that performance of the genotypes was different across various environments.
Basic summary statistics and genotypic means were estimated for all studied traits to examine the variability is acquired data. Mean for DWSI was 40.46, that was within the range of 5.85 to 104.32 in growth chamber study. Mean for EWSI was 32.51, that was also within the range of 4.35 to 76.29. Similarly, GSI values were ranged from 33.48 to 133.78 with the mean value of 85.31. RLSI has the mean 46.80 within the range of 8.37 to 113.25. Mean value for SLSI was 69.90 within the range of 17.35 to 257.19. Other statistical parameters like, standard deviation, standard error of mean, First, 2nd and 3rd quartiles for traits studied in growth chamber were also given in Table 4.
Biochemical and physiological traits were also subjected to summary statistics. Mean for CMS was 85.114% with the range of 75.00 to 95.47%. Highest value for ELWL was 9.00% and lowest was 7.67% whereas, mean was 8.19%. Mean for ECCL was 4.799% while minimum value was 1.997% and the highest was 20.91%.Glycinebetaine (GB) has mean value 1.59 μmol/g within the range of 1.00 μmol/g to 2.609 μmol/g dry weight. The lowest value for NRA was 1.504 μmolNO2 g− 1 fresh weight h− 1and the highest value was 5.065 μmolNO2 g− 1 fresh weight h− 1while mean was 4.095μmolNO2 g− 1 fresh weight h− 1. Proline concentrations were ranging from 2.69to 71.49 μmolg− 1 fresh weight with the mean value of 21.36μmolg− 1 fresh weight. Mean across all genotypes for RWC was 70.46% while the lowest value was 35.71% and the highest was 120.00%. Other statistical parameters like, standard deviation, standard error of mean, 1st, 2nd and 3rd quartiles for traits studied in growth chamber were also given in Table 4.
Means for yield of chickpea genotypes across all of the six different locations were also accessed. Mean yield at AARI was 1639 kg/hac with the range of 679.0 to 3273 kg/hac.0. AZRI produced the mean yield 1749 kg/ha for all genotypes while lowest yield was 919.0 and highest was 2793.0. Mean yield for K. Kot was 1016 with the range of 400.0 to 1663.0. NIAB-lysimter produced the mean yield of 206.88 with the range of 51.00 to 506.0. Mean yield at NIAB-irrigated was 2868 with the range of 1584.0 to 3898.0. NIAB-rainfed produced the mean yield of 2569.6 while the range was 1170.0 to 3439.0 (Table 4). Other statistical parameters like, standard deviation, standard error of mean, 1st, 2nd and 3rd quartiles for traits studied in growth chamber were also given in Table 4.
Mean values for all of the studied chickpea genotypes were given in Table 5. Genotypes 2 (CM776/06), 7 (CM848/06), 10 (09AK055), 23 (K0017–10), 24 (K0019–10), 25 (K0024–10), 26 (K0026–10), 29 (K0031–10), 30 (K0032–10), 31 (K0037–10), 32 (K0042–10), 37 (K0068–10), 38 (K0069–10), 39 (09AG019), 45 (08AG015), 54 (D0097–10) and 55 (D0099–10) showed the 100% GSI under growth chamber conditions. The lowest GSI was observed 33.33, 44.44 and 66.67% for genotypes 44 (09AG006), 50 (D0085–10) and 28 (K0030–10), respectively (Table 5). However, two chickpea genotypes i.e. 38 (K0069–10) and 39 (09AG019) showed the increased germination percentage having GSI value of 133.3%under lower osmotic stress conditions (Table 5).
Genotypes K004–10, K0026–10, K0032–10, 09AG019 and D0091–10 had highest values for RLSI with means of 108.3, 93.76, 79.26, 98.21 and 113.0%respectively, whereas genotypes CM709/06, CM776/06, 2175, K005–10, K0016–10, K0030–10 and 09AG006 had the lowest RLSI 20.72, 19.57, 26.12, 23.68, 14.43, 25.85 and 8.42%, respectively (Table 5).GenotypesK0017–10, K0019–10, K0032–10, 09AG019, 09AG002 and 08AG015 have 257.7, 106.9, 102.4, 110.6and114.6%, respectively being the highest values for SLSI whereas, genotypes K005–10, K0042–10, 09AG006 and D0094–10 had the lowest values for SLSI 17.45, 23.55, 20.8, and 23.33%, respectively (Table 5). The lowest FWSI values were observed in 09AG006 (4.356%) followed by D0081–10 (8.038%),D0094–10 (8.176%), K005–10 (9.296%), D0079–10 (10.27%), K0016–10 (13.65%), K003–10 (15.98%), and CM709/06 (16.92%). The highest FWSI was observed in K0019–10 (76.67%), followed by K0032–10 (68.00%), 2204 (66.89%) and K0017–10 (64.86%) (Table 5). The maximum DWSI was observed for K0019–10 (104.2%), followed by K0017–10 (96.04%), D0097–10 (74.88%), 09AK055 (74.25%) and 2204 (69.48%) whereas, the minimum DWSI was observed in 09AG006 (5.94%), followed by D0094–10 (13.67%), D0079–10 (15.31%), D0080–10 (15.51%), D0081–10 (16.43%), K005–10 (18.30%) and K0016–10 (19.44%) (Table 5).
The highest value for CMS was observed in 09AG019 (95.47%), followed by K0019–10 (93.73%), 09AG002 (93.40%), D0080–10 (92.43%), K008–10 (91.53%), 08AG004 (91.20%) and D0091–10 (91.18%) whereas, the lowest CMS was observed in K0037–10 (78.17%), 2204 (77.50%), K005–10 (77.43%), K0014–10 (75.27%) and K004–10 (75.00%) (Table 5). The highest RWC was found in CM795/06 (120.0%), followed by K0016–10 (115.8%), CM709/06 (97.65%), K012–10 (95.41%), CM1528/03 (90.32%), K0026–10 (86.81%), whereas it was the lowest inK0017–10 (44.41%), followed by K005–10 (48.12%),K0013–10 (50.80%), and K0027–10 (53.22%), (Table 5). The highest ELWL value of 9.00% was observed in CM1004/06, followed by CM795/06, K0019–10, D0078–10 and D0094–10 whereas, the lowest 7.67% was observed in K0027–10,followed by K0037–10 and K0052–10 (Table 5).
The highest value of ECCL was observed in 08AG015 (20.91%), followed by CM731/06 (11.11%) and 2124 (10.78%) whereas, the lowest was in D0094–10 (2.00%) followed by D0091–10 (2.17%),K0019–10 (2.21%),K0030–10 (2.22%), D0081–10 (2.26%), K008–10 (2.33%), K0031–10 (2.38%), K012–10 (2.58%), K0013–10 (2.77%), and K0026–10 (2.91%), (Table 5).
The highest glycinebetaine (GB) contents were observed in CM1004/06 (2.61 μmol/g), followed by K0026–10 (2.32 μmol/g), D0096–10 (2.23 μmol/g), K0032–10 (2.15 μmol/g), 09AK055 (2.07 μmol/g), and K010–10 (2.02 μmol/g),, whereas, the lowest GB contents were found in K0030–10 (1.15 μmol/g), followed by K0014–10 (1.16 μmol/g), K0063–10 (1.19 μmol/g), K0024–10 (1.31 μmol/g), K005–10 (1.31 μmol/g), K004–10 (1.33 μmol/g) and 2124 (1.33 μmol/g) (Table 5).
Proline contents were the highest in 2175 (23.83 μmol/g), followed by K0032–10 (23.51 μmol/g), CM687/06 (23.33 μmol/g), and K0068–10 (20.22 μmol/g), whereas, it was the lower in D0079–10 (1.79 μmol/g) followed by D0080–10 (1.90 μmol/g), CM731/06 (1.98 μmol/g), K005–10 (2.01 μmol/g), D0078–10 (2.02 μmol/g), 09AG006 (2.03 μmol/g), 2204 (2.04 μmol/g), D0091–10 (2.08 μmol/g), CM795/06 (2.16 μmol/g),, and D0094–10 (2.18 μmol/g), respectively (Table 5).
The highest NRA was recorded in K0031–10 (5.06 μ mol NO2.g− 1FW h− 1), followed byK0017–10 (5.01 μ mol NO2.g− 1FW h− 1), and 08AG004 (5.01 μ mol NO2.g− 1FW h− 1) whereas the lower NRA was observed in D0079–10 (2.25 μ mol NO2.g− 1FW h− 1), followed by D0080–10 (2.35 μ mol NO2.g− 1FW h− 1), CM709/06 (2.41 μ mol NO2.g− 1FW h− 1) and K0049–10 (2.43 μ mol NO2.g− 1FW h− 1),respectively (Table 5).
Average grain yield across all of the studied environments showed that genotype D0091–10 (2283 kg ha− 1) maintained the highest yield followed by, K010–10 (2081 kg ha− 1), 2175 (2073 kg ha− 1), CM1528/03 (2043 kg ha− 1), and D0085–10 (2015 kg ha− 1) whereas, genotypeK0037–10 (1217 kg ha− 1) had the lowest grain yield followed byK0063–10 (1285 kg ha− 1) (Table 5).
Principal component biplot analysis
PCA biplots were generated for each experiment separately and cumulatively. PCA biplot for early growth indices depicted the 80.89% (PC1: 65.62%, PC2: 15.28%) of the variability in the raw data (Fig. 1). Angle of the trait vectors was reflecting the correlation of variables. If angle between two trait vectors is <90o then correlation is positive, if angle is >90o then correlation is negative while if equivalent to 90o then traits are said be independent of each other. Growth indices were mostly had positive correlation except for the correlation of SLSI and GSI. FWSI and DWSI which have strong positive correlation. Traits vectors were labeled for the mean values of the corresponding traits which are facilitating the selection of genotypes for the lowest or the highest mean values. Among studied chickpea genotypes, 23 (K0017–10), 39 (09AG019), 24 (K0019–10), 15 (K004–10), 30 (K0032–10), 12 (2204), 10 (09AK055), 6 (CM795/06), 25 (K0024–10) and 54 (D0097–10) had the highest mean values and better performance while the genotypes 44 (09AG006), 47 (D0079–10), 48 (D0080–10), 52 (D0094–10), 16 (K005–10), 1 (CM709/06), 28 (K0030–10) and 22 (K0016–10) has the lowest mean values and relatively poor performaningfor studied indices of early growth stages (Fig. 1).
PCA biplot for physiological and biochemical traits of the chickpea genotypes was reflecting the 95.31% (PC1: 64.99%, PC2: 30.32%) of the total variability in data. Proline contents and GB contents had strong positive correlation between each other. Electrolyte leakage, RWC and ELWL were also positively correlated whereas, CMS and NRA were negatively correlated with each other. Among studied chickpea genotypes, 11 (2175), 4 (CM687/06), 30 (K0032–10), 37 (K0068–10), 36 (K0066–10), 10 (09AK055), 53 (D0096–10), 14 (K003–10), 8 (CM1528/03) and 15 (K004–10) had the highest NRA, GB and Proline contents. Genotypes 6 (CM795/06), 22 (K0016–10), 1 (CM709/06) and 19 (K012–10) had the highest values for electrolyte leakage, RWC, ELWL and CMS. Among all of the chickpea genotypes, 27 (K0027–10), 23 (K0017–10), 16 (K005–10), 18 (K010–10) and 7 (CM848/06) had the lowest mean values for all of the studied physiological and biochemical traits (Fig. 1).
PCA biplot for grain yield was reflecting the 70.90% of the total variability of yield data. AARI was proved to be entirely distinct environment, whereas, AZRI and K. Kot were positively correlated in their discrimination ability to assort the chickpea genotypes on yield performance. Genotypes 3 (CM1004/06), 9 (2124), 11 (2175), 17 (K008–10) and 8 (CM1528/03) were the highest yielding in AARI conditions. Genotypes 51 (D0091–10), 50 (D0085–10) and 18 (K010–10) were higher yielder across all of the environments. Genotypes 31 (K0037–10), 34 (K0052–10), 12 (2204), 32 (K0042–10), 26 (K0026–10), 1 (CM709/06), 40 (09AG015), 37 (K0068–10), 35 (K0063–10) and 15 (K004–10) were lower yielder across all of the environments except AARI (Fig. 1).
Cumulative PCA biplot for all of the studied traits showed the 69.89% (PC1: 40.99%, PC2: 28.90%) of the total variability in raw data. Wide spectrum distribution of trait vectors in this biplot showed that different traits of early growth stages, physio-chemical nature and yield performance have differential correlation with each other. GSI, FWSI, DWSI traits has positive correlation with RWC, Proline, GB, ELWL and yield at AARI environment. Yield performance at AARI environment was positively correlated with higher mean values of the some traits of early growth stages (GSI, FWSI and DWSI) and physiological and biochemical traits (RWC, Proline, GB, ELWL and Electrolyte leakage) background. Yield performance across all other environments, NIAB-lysimeter, NIAB-rainfed, AZRI, K. Kot had strong positive correlation with CMS. Yield performance of chickpea genotypes under NIAB-irrigated environment was distinctive from other environments due to higher vector angles. While, yield performance of subjected chickpea genotypes under AARI environment had negative correlation with reference to other environments (Fig. 1). Some of the traits atearly growth stages (GSI, FWSI, DWSI, SLSI, RLSI) and physio-chemical nature (Proline, RWC, NRA, ELWL, Glycine betaine and Electrolyte leakage) were negatively correlated with the yield performance of chickpea genotypes at AZRI, NIAB-rainfed, NIAB-NIAB-irrigated, NIAB-lysimeter and K. Kot environments (Fig. 1).
Among studied chickpea genotypes, 29 (K0031–10), 12 (2204), 30 (K0032–10), 38 (K0069–10), 24 (K0019–10), 1 (CM709/06), 32 (K0042–10), 15 (K004–10), 33 (K0049–10) and 35 (K0063–10) had higher mean performance for SLSI, RLSI, NRA, GSI, FWSI and DWSI however, these genotypes were lower in grain yield because these are allocated on the the lowest mean vector spoke of the yield environments. Chickpea genotypes 3 (CM1004/06), 9 (2124), 17 (K008–10), 11 (2175), 8 (CM1528/03) and 13 (K002–10) were high yielding under AARI environment whereas, genotypes 48 (D0080–10), 49 (D0081–10), 41 (08AG004), 39 (09AG019), 40 (09AG015) and 44(09AG006) were high yielding under NIAB-lysimeter, NIAB-rainfed, AZRI and K. Kot environments. Genotypes 49 (D0081–10), 18 (K010–10), 10 (09AK055) and 16(K005–10) were high yielding under NIAB-irrigated environment (Fig. 1).
GGE biplot analysis
Comparison biplots for environments and genotypes were generated by using the GGE biplot analysis. Environment and genotype comparison biplots were reflecting the 73.82% (PC1: 48.92% and PC2: 24.90%) of total variability in chickpea genotypes under subjected environments for grain yield. Arrow on average environment axis at the center of concentric circles showed the position of ideal environments to evaluate the performance of chickpea genotypes. Among studied environments, NIAB-rainfed environment was closer to the ideal environment followed by AZRI and NIAB-irrigated environments. NIAB-lysimeter and K. Kot were poor environments to evaluate the chickpea genotypes because these are positioned farthest away from the center of concentric circles (Fig. 2).
Arrow on the average environment axis at the center of concentric circles for genotypic comparison is reflecting the position of ideal genotype for subjected environments. Genotypes positioned closer to ideal genotype are preferable for selection under studied environments whereas, genotypes positioned farthest away from ideal genotype are poor performing based on the average yield performance across all environments. Genotype 51 (D0091–10), 18 (K010–10), 50 (D0085–10), 16 (K005–10), 46 (D0078–10), 42 (08AG016), 41 (08AG004), 48 (D0080–10), 43 (09AG002), 13 (K002–10) and 55 (D0099–10) were high yielding based on the average performance across all environments with reference to the ideal genotype. Genotypes 31 (K0037–10), 12 (2204), 34 (K0052–10), 40 (09AG015), 32 (K0042–10), 1 (CM709/06), 37 (K0068–10), 15 (K004–10), 26 (K0026–10) and 35 (K0063–10) were low yielding based on the average yield performance across all environments with reference to the ideal genotype (Fig. 2).
Drought stress is a global problem which adversely affects the performance of different crop plants. Development of drought tolerant cultivars is prerequisite to encounter the prevailing drought stress on sustainable basis. Evaluation of the chickpea responses to drought stress showed the significant variations for different traits of early growth stages, physio-chemical nature and grain yield. Significance of the variations depicted the differential responses of the genotypes to variable environments at different growth stages. These variations could be attributed to the differences in their genetic makeup of the studied genotypes. Different researchers have also reported the differential responses of the chickpea genotypes under different environments and at different growth stages ([1, 21, 22, 38] a, b ;).
Different chickpea traits were targeted for identification of drought tolerant genotype among wide range of available germplasm. Literature indicated that evaluation and screening of germplasm under drought stress on the basis of physiological and biochemical parameters is prerequisite for crop improvement .GSI of different genotypes was unaffected, reduced or increased under water stress condition compared to the control treatment. Germination of the genotypes was affected differently under the provision of two different osmotic potentials (Table 5). Genotypes CM776/06, CM848/06, 09AK055, K0017–10, K0019–10, K0024–10, K0026–10, K0031–10, K0032–10, K0037–10, K0042–10, K0068–10, K0069–10, 09AG019, 08AG015, D0097–10 and D0099–10 showed the 100% GSI which showed their tolerance response under two different osmotic potentials for seed emergence. Genotypes 09AG006, D0085–10 and K0030–10 showed the most the lowest germination low osmotic potential whereas, genotype K0069–10 and 09AG019 showed the surprisingly increase in germination at lower osmotic potential. Such erratic behavior of germination percentage under stress condition (Table 5) also depends on seed viability, i.e., seed storing after harvesting, maintenance of seed moisture etc. [5, 6], so this type of seeds GSI could not show any linear or direct association with drought stress. The previous findings also suggested that only GSI could not be used for screening genotypes for drought tolerance .
Parameters of early growth stages were also used for evaluation of genotypic response under different water stress treatments. Growth parameters were studied as growth indices to reflect the relative value under both osmotic stress conditions provided by PEG solution. Different genotypes showed the differential (increased, decreased or unaltered) responses for these growth indices (FWSI, DWSI, RLSI and SLSI). Genotypes having the values for growth indices equal to 100, were showing the equivalent performance under both osmotic conditions whereas, values for growth indices near to zero are reflecting the highest reduction in performance under stress condition (Table 5). Genotypes having growth indices values more than 100are showing the better performance of genotypes for intended growth traits under stress condition (Table 5). Based on the buffering capacity to minimize the reduction of the different growth indices, genotypes 23 (K0017–10), 39 (09AG019), 24 (K0019–10), 15 (K004–10), 30 (K0032–10), 12 (2204), 10 (09AK055), 6 (CM795/06), 25 (K0024–10) and 54 (D0097–10) were osmotic stress tolerant at early growth stages. It was also previously reported that genotypic responses are variable under different stress conditions [26, 33] at early growth stages which indicate that consistent genotypic evaluation is prerequisite for development of new high yielding stress tolerant cultivars.
Tolerance at early growth stages is only useful if it harbored the high yield or it showed the positive association with other physio-biochemical traits. Therefore, performance of chickpea genotypes was further evaluated for physio-biochemical traits and yield performance. Significant variability in chickpea genotypes for different physiological and biochemical traits was evident in present study. Several other researchers have also reported the variability in physiological and biochemical traits of chickpea [2, 8, 11, 12, 16, 17]. Based on physiological and biochemical traits, genotypes CM687/06, K0032–10, K0068–10, K0066–10, 09AK055, D0096–10, K003–10, CM1528/03, K004–10, CM795/06, K0016–10, CM709/06 and K012–10 were better performer.
Genotypes with better performance at early growth stages were not necessarily having the better outcome for physio-chemical traits. Available findings indicated that, genotypes with better performance in physio-chemical traits not necessarily showed the better performance at early growth stages. Results of this study showed that among the better performing genotypes at early growth stages, 15 were higher in RWC and proline accumulation, 30 were higher in CMS, RWC and proline accumulation, 10 has higher glycine betain, proline, CMS and RWC whereas, 12 have lower proline and CMS values (Table 5). Researchers reported that higher proline, glycine betain, RWC and CMS values were reflecting the higher capability to tolerate the drought stress whereas, the lowest values for electrolyte leakage were depicting the capability for drought tolerance ( [2, 8, 11, 16, 17];).
Average yield across all of the subjected environments revealed that genotypes D0091–10, D0085–10, K010–10, 2175 and CM1528/03 were high in grain yield. Being high yielding under diverse and severe environmental conditions these genotypes can be declared drought tolerant. But these genotypes were not proved to be drought tolerant at early growth stages due to more reduction in growth performance at germination and seedling stages (Table 5). These genotypes were also not having extensively higher in physio-biochemical characteristics at pre-reproductive growth stages. These results showed that better performance of chickpea genotypes at early growth stages is not guaranteeing the high yield or drought tolerance at terminal growth stages. It may be corroborated that better growth performance at seedling stage under drought stress may induce the earliness in reproductive maturity which may confers yield penalty ( a,b [23, 27];).
PCA biplotis the most effective multivariate analysis evaluate the traits interaction and genotypic performance. PCA biplots analysis for SLSI, RLSI, NRA, GSI, FWSI and DWSI showed negative correlation with average grain yield of chickpea genotypes (Fig. 1). These findings further confirmed that selection of genotypes at early growth stages and based on different physio-chemical traits is not providing any surety for higher grain yield and terminal drought stress tolerance. PCA biplot was also extensively used by several researchers to dissect the traits correlation in different crop plants [7, 23,24,25].
Genotype × environment interaction (GEI) is seriously affected the performance of genotypes across different environments. Multi-environment trials are necessitating the manifestation of suitable biometrical tool to evaluate the interaction of genotypes with different environments, therefore, GGE biplot analysis was used in present study. In present study, NIAB-rainfed proved to be ideal environment for discrimination of chickpea genotypes for grain yield performance. Ideal environment is only referred on context based or theoretical perspectives but not on practical perspectives. NIAB-rainfed should be preferred for further studies to differentiate the genotypes for their yield performance. Among other environments, NIAB-lysimeter and K. Kot proved to be least capable of differentiating the genotypes for their yield performance (Fig. 2). Mean yield performance of chickpea genotypes was also very poor on NIAB-lysimeter and K. Kot environments due to very severe nature of water deficit at these environments. NIAB-lysimeter and K. Kot environments were depicting the similar genotypic performance therefore one of these environments could be excluded from further studies to cut short the management expenses and to alternatively add other environments in the studies. Maqbool et al . also evaluated the chickpea genotypes under similar type of environments in different years and found the poor yield potential of lysimter conditions and better yield potential of NIAB-rainfed conditions.
GGE biplot analysis revealed that genotype D0091–10, K010–10, D0085–10, K005–10, D0078–10, 08AG016, 08AG004, D0080–10, 09AG002, K002–10 and D0099–10 were relatively superior in yield performance under drought stressed hotspot diverse environmental conditions (Fig. 2). Therefore, dissemination of these genotypes should be promoted in chickpea grown regions to have the higher yield potential under terminal drought conditions. Multi-environment hotspot evaluation of chickpea germplasm besides selection of high yielding genotypes also facilitates the identification of poor yielding or drought sensitive genotypes. Genotypes K0037–10, 2204, K0052–10, 09AG015, K0042–10, CM709/06, K0068–10, K004–10, K0026–10 and K0063–10 were poor yielding anddrought sensitive in present study according to multi-environment evaluation. These genotypes could efficiently be used as contrasting parents in different hybridization breeding programs intended for genetic improvement of chickpea genotypes for drought tolerance.
Present investigation indicated that chickpea genotypes were significantly different for various seedling, physio-biochemical traits and yield performance across different environments. PCA biplot suggested that genotypes with better performance at early growth stages were not correspondingly high yielding. NIAB-rainfed environment proved ideal in nature to discriminate the chickpea genotypes whereas, NIAB-lysimeter and K. Kot was least discriminating the genotypes on yield basis. Based on the yield performance, genotypes D0091–10, K010–10, D0085–10, K005–10, D0078–10, 08AG016, 08AG004, D0080–10, 09AG002, K002–10 and D0099–10 were found to be high yielding and drought tolerant across multiple hotspot environments. These genotypes are intended for further evaluation for varietal approval to recommend for general cultivation on farmer fields in drought affected areas. Among physio-biochemical traits, higher proline, glycine betain, RWC and CMS were reflecting the higher capability to tolerate the drought stress in chickpea. Drought sensitive genotypes (K0037–10, 2204, K0052–10, 09AG015, K0042–10, CM709/06, K0068–10, K004–10, K0026–10 and K0063–10) were also identified in present study which are resourceful asset for using as contrasting parents in hybridization programs.
Availability of data and materials
The datasets used and/or analysed during the current study are available from the corresponding author on reasonable request.
Nuclear Institute for Agriculture and Biology
Nuclear Institute for Food and Agriculture
Arid Zone Research Institute
Ayub Agriculture Research Institute
University of Agriculture
Relative water content
Excised leaf water loss
Pulses Research Institute
Root length stress tolerance index
Shoot length stress tolerance index
Fresh weight stress tolerance index
Dry weight stress tolerance index
Cell membrane stability
Germination stress tolerance index
Nitrite reductase activity
Completely randomized design
Randomized complete block design
Principal Component Analysis
Genotype × environment interaction
Ahmad S, Ahmad R, Ashraf MY, Waraich A. Sunflower (Helianthus annus L.) response to drought stress at germination and seedling growth stages. Pak. J. Botany. 2009;41(2):647–54.
Ali MA, Jabran K, Awan SI, Abbas A, Ehsanullah M, Zulkiffal T, Acet JF, Rehman A. Morpho-physiological diversity and its implications for improving drought tolerance in grain sorghum at different growth stages. Aust J Crop Sci. 2011;5(3):311–20.
Anonymous 2011. FAO Statistical Database. http://faostat.fao.org.
Araghi GS, Assad MT. Evaluation of four screening techniques for drought resistance and their relationship to yield reduction ratio in wheat. Euphytica. 1998;103:293–9.
Ashraf MY, Naqvi MH, Khan AH. Evaluation of four screening techniques for drought tolerance in wheat (TriticumaestivumL.). Acta Agron Hung. 1996;44:213–20.
Ashraf MY, Ali Y, Qureshi TM. Effect of salinity on photosynthetic efficiency and yield of rice genotypes. Pak J Biol Sci. 1998;1:72–4.
Aslam M, Maqbool MA, Zaman QU, Shahid M, Akhtar MA, Rana AS. Comparison of different tolerance indices and PCA biplot analysis for assessment of salinity tolerance in lentil (Lens culinaris) genotypes. Int J Agric Biol. 2017;19:470–8.
Bajji M, Kinet JM, Lutts S. The use of the electrolyte leakage method for assessing cell membrane stability as a water stress tolerance test in durum wheat. Plant Growth Regul. 2002;36:61–70.
Barrs HD, Weatherley PE. A re-examination of the relative turgidity technique for estimating water deficit in leaves. Aust J Biol Sci. 1962;15:413–28.
Bates LS, Waldren RP, Tears ID. Rapid determination of free proline for water stress studies. Plant Soil. 1973;39:205–7.
Ceyhan E, Kahraman A, Onder M, Ates MK, Karadas S, Topak R, Avci MA. Physiological and biochemical responses to drought stress of chickpea genotypes. WASET. 2012;6(6):352–7.
Clarke JM, Townley-Smith TF. Heritability and relationship to yield of excised leaf water retention in durum wheat. Crop Sci. 1986;26:289–92.
Cutforth HW, McGinn SM, McPhee KE, Miller PR. A daptation of pulse crops to the changing climate of the northern Great Plains. Agron J. 2007;99:1684–99.
Dencic S, Kastori R, Kobiljski B, Duggan B. Evaluation of grain yield and its components in wheat cultivates and landraces under near optimal and drought conditions. Euphytica. 2000;113:43–52.
Dhanda SS, Sethi GS. Inheritance of excised leaf water loss and relative water content in bread wheat (Triricumaestivum). Euphytica. 1998;104:39–47.
Dhanda SS, Sethi GS. Tolerance to drought stress among selected Indian wheat cultivars. J Agric Sci. 2002;139(3):319–26.
Farooq S, Azam F. Co-existence of salt and drought tolerance in Triticeae. Hereditas. 2002;135:205–10.
George DW. High temperature seed dormancy in wheat (Triticumaestivum L.). Crop Sci. 1967;7:249–53.
Greive CM, Grattan SR. Rapid assay for determination of water soluble quaternary amino compounds. Plant Soil. 1983;70:303–7.
Heatherly LG, Elmore RW. Managing inputs for peak production. In: Specht JE, Boerma HR, editors. Soybeans: Improvement, Production, and Uses. Agronomy Monographs, vol. 16. 3rd ed. Madison: ASA-CSSA-SSSA; 2004. p. 451–536.
Kramer PJ. Drought stress and the origin of adaptations. In: Turner NC, Kramer PJ, editors. Adaptations of plants to water and high temperature stress; 1980.
Maqbool MA, Aslam M, Ali A, Mahmud ST, Babar F, Qamar Z. Drought tolerance indices based evaluation of chickpea advanced lines under different water treatments. Res. Crops. 2015;6:336–44.
Maqbool MA, Aslam M, Ali H, Shah TM. Evaluation of advanced chickpea (Cicerarietinum L.) accessions based on drought tolerance indices and SSR markers againstdifferent water treatments. Pak J Bot. 2016;48:1421–9.
Maqbool MA, Aslam M, Ali H, Shah TM, Atta BM. GGE biplot analysis basedselection of superior chickpea (Cicerarietinum L.) inbred lines under variable waterenvironments. Pak J Bot. 2015b;47:1901–8.
Maqbool MA, Aslam M, Ali H, Shah TM, Farid B, Zaman QU. Drought tolerance indices based evaluation of chickpea advanced lines under different water treatments. Res Crops. 2015a;16:336–44.
Misra N, Dwivedi UN. Genotypic differences in salinity tolerance of green gram cultivars. Plant Sci. 2004;166:1135–42.
Maqbool MA, Aslam M, Ali H. Breeding for improved drought tolerance in chickpea (Cicer arietinum L.). Plant Breed. 2017;136:300–18.
Mukherjee SP, Choudhuri MA. Implications of water stress-induced changes in the levels of endogenous ascorbic acid and hydrogen peroxide in Vigna seedlings. Physiol Plant. 1983;58:166–70. https://doi.org/10.1111/j.1399-3054.1983.tb04162.x.
Nguyen HT, Babu RC, Blum A. Breeding for drought resistance in rice: physiology and molecular genetics considerations. Crop Sci. 1997;37:1426–34.
Pathan MS, Lee JD, Shannon JG, Nguyen HT. Recent advances in breeding for drought and salt stress tolerance in soybean. In: Jenks MA, Hasegawa PM, Jain SM, editors. A dvances in molecular breeding toward drought and salt tolerant crops. New York: Springer; 2007. p. 739–73.
Premachandra GS, Saneoka H, Ogta S. Nutrio-physiological evaluation of the polyethylene glycol test of cell membrane stability in maize. Crop Sci. 1989;29:1292–7.
Ramarao CS, Patil VH, Dhak BD, Kadrekar SB. A simple in vivo method for the determination of nitrite reductase activity in rice roots. Z Pflanzenphysiol. 1983;109:81–5.
Romo S, Labrador E, Dopico B. Water stress regulated gene expression in Cicerarietinum seedlings and plants. Plant Physiol Biochem. 2001;39:1017–26.
Sharma KK, Lavanya M. Recent developments in transgenics for abiotic stress in legumes of the semi-arid tropics. In JIRCAS Work Rep. 2002;23:61–73.
Sloane RJ, Patterson RP, Carter TE. Field drought tolerance of a soybean plant introduction. Crop Sci. 1988;30(1):118–23.
Steel RGD, Torrie JH. Principles and procedures of statistics: a biometrical approach. 3rd ed. New York: McGraw -Hill; 1997.
Sullivan CY. In: Rao NGP, House LR, editors. Mechanism of heat and drought resistance in grain sorghum and methods of measurement. New Dehli: Sorghum in the seventies, Oxford & 113H publ, co; 1972. p. 247–64.
Talebi R, Ensafi MH, Baghebani N, Karami E, Mohammadi K. Physiological responses of chickpea (CicerarietinumL.) genotypes to drought stress. Environ. Exper. Biol. 2013;11:9–15.
Toker C, Lluch C, Tejera NA, Serraj R, Siddique KHM. Abiotic stresses. In: Yadav SS, Redden R, Chen W, Sharma B, editors. Chickpea Breeding and Management. Wallingford: CABI; 2007a. p. 474–96. Turner N.C. 1981.
Toker C, Lluch C, Tejera NA, Serraj R, Siddique KHM. Abiotic stresses. In: Yadav SS, Redden R, Chen W, Sharma B, editors. Chickpea breeding and management. Wallingford: CABI; 2007b. p. 474–96.
Turner NC, Wright GC, Siddique KHM. Adaptation of grain legumes (pulses) to water limited environments. Adv Agron. 2001;71:193–23.
Varshney RJ, Thudi M, Nayak SN, Gaur PM, Kashiwagi J, Karishnamurhy L, Jaganathan D, Koppolu K, Bohra A, Tripathi S, Rathore A, Jukanti AK, Jayalakshmi V, Vimula A, Singh SJ, Yasin M, Sheshshayee MS, Viswanatha KP. Genetic dissection of drought tolerance in chickpea (Cicerarietinum L.). Theo Appl Genet. 2014;127:445–62.
Yadav SS, Kumar J, Yadav SK, Singh S, Yadav VS, Turner NC, Redden R. Evaluation of Helicoverpa and drought resistance in desi and kabuli chickpea. Plant Genet Res. 2006a;4:198–203.
Yadav SS, Redden RJ, Chen C, Sharma B. Chickpea breeding and management. In: Oxford shire. U.K.: CAB International; 2006b. ISBN-13: 9781845932145.
Yan W, Tinker NA. An integrated system of biplot analysis for displaying, interpreting and exploring genotype by - environment interactions. Crop Sci. 2005;45:1004–16.
Yan W, Tinker NA. Biplot analysis of multi-environment trial data: principles and applications. Can J Plant Sci. 2006;86:623–45.
The authors are thankful to helping staff for their assistance in field experimentation and lab. Analysis.
Irem Waqar, Junior Scientist Mahmud Shah, at NIAB
This study is a part of a research project entitled, “Improvement of chickpea productivity through the identification of drought tolerant and disease resistant chickpea genotypes for marginal lands” sponsored by the Punjab Agricultural Research Board, Lahore through a research project PARB-120. We are thankful to the PARB for providing the financial support for execution of this research work.
Ethics approval and consent to participate
Consent for publication
The authors declare that they have no competing interests.
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
About this article
Cite this article
Shah, T.M., Imran, M., Atta, B.M. et al. Selection and screening of drought tolerant high yielding chickpea genotypes based on physio-biochemical indices and multi-environmental yield trials. BMC Plant Biol 20, 171 (2020). https://doi.org/10.1186/s12870-020-02381-9
- Drought tolerance
- Physiological traits
- And screening criteria