Selection and screening of drought tolerant high yielding chickpea genotypes based on physio-biochemical indices and multi-environmental yield trials

Background 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. Results 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. Conclusions 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.


(Continued from previous page)
Conclusions: 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.
Keywords: Chickpea, Drought tolerance, Physiological traits, And screening criteria Background 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 [34]. 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 CO 2 in atmosphere and climatic changes [3]. Resilience of the legume crops in current prevailing weather extremities could be the futuristic adaptation in more severe climatic conditions [13].
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 [42]. 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 [18]. P.l. = nd 2 (1.00) + nd 4 (0.75) + nd 6 (0.50) + nd 8 (0.25). Where, nd 2 , nd 4 , nd 6 and nd 8 were seeds germinated on the 2nd, 4th, 6 th and 8th day of sowing, respectively.
Germination stress tolerance index (GSI) was calculated as described by [5]; GSI ¼ P:I: of stress seeds P:I: of control seeds Â 100 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

RLSI
Cell membrane stability (CMS) was determined according to the procedure developed by [37]. 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 (T 1 ) and in deionized water as a control (C 1 ) 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

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,

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 [9] 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.

Proline content
Fresh chickpea leaves were used to determine proline contents according to the method of Bates et al. [10]. 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.

Nitrate reductase activity
Nitrate reductase activity (NRA) in leaves of chickpea was determined according to reported method [32] 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 KNO 3 . 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 NO 2 g − 1 F.W h − 1 by using standard curve developed with NaNO 2 .

Glycine betaine contents
Glycine betaine in leaf samples of chickpea genotypes were determined by using the protocol recommended by [19]. 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, NIABirrigated and NIAB-rainfed environments were developed at Nuclear Institute for Agriculture and Biology, Faisalabad, by managing irrigation pattern. NIABlysimeter 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 [35]. 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 −1 for subsequent analysis.

Statistical analysis
Data acquired from growth chamber investigation was subjected to analysis of variance (ANOVA) under completely randomized design (CRD) with three replications [36]. 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 ( [22] a, b [7];) 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].

Results
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, Glycinebetaine, 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.

Mean comparison
Basic summary statistics and genotypic means were estimated for all studied traits to examine the variability is acquired data. Mean for DWSI was 40 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 μmolNO 2 g − 1 fresh weight h − 1 and the highest value was 5.065 μmolNO 2 g − 1 fresh weight h − 1 while mean was 4.095μmolNO 2 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.   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 (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).
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, NIABlysimeter, 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).

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 (

Discussion
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 [23];).
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 [38].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 [1].
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.
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 physiochemical 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 ( [22] 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 . [22] 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.

Conclusions
Present investigation indicated that chickpea genotypes were significantly different for various seedling, physiobiochemical 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.