Plant vigour QTLs co-map with an earlier reported QTL hotspot for drought tolerance while water saving QTLs map in other regions of the chickpea genome

Background Terminal drought stress leads to substantial annual yield losses in chickpea (Cicer arietinum L.). Adaptation to water limitation is a matter of matching water supply to water demand by the crop. Therefore, harnessing the genetics of traits contributing to plant water use, i.e. transpiration rate and canopy development dynamics, is important to design crop ideotypes suited to a varying range of water limited environments. With an aim of identifying genomic regions for plant vigour (growth and canopy size) and canopy conductance traits, 232 recombinant inbred lines derived from a cross between ICC 4958 and ICC 1882, were phenotyped at vegetative stage under well-watered conditions using a high throughput phenotyping platform (LeasyScan). Results Twenty one major quantitative trait loci (M-QTLs) were identified for plant vigour and canopy conductance traits using an ultra-high density bin map. Plant vigour traits had 13 M-QTLs on CaLG04, with favourable alleles from high vigour parent ICC 4958. Most of them co-mapped with a previously fine mapped major drought tolerance “QTL-hotspot” region on CaLG04. One M-QTL was found for canopy conductance on CaLG03 with the ultra-high density bin map. Comparative analysis of the QTLs found across different density genetic maps revealed that QTL size reduced considerably and % of phenotypic variation increased as marker density increased. Conclusion Earlier reported drought tolerance hotspot is a vigour locus. The fact that canopy conductance traits, i.e. the other important determinant of plant water use, mapped on CaLG03 provides an opportunity to manipulate these loci to tailor recombinants having low/high transpiration rate and plant vigour, fitted to specific drought stress scenarios in chickpea. Electronic supplementary material The online version of this article (10.1186/s12870-018-1245-1) contains supplementary material, which is available to authorized users.


Background
Chickpea (Cicer arietinum L.), the second most important grain legume crops in the world [1], is widely cultivated on residual soil moisture in the arid and semi-arid agricultural systems of the world. Terminal water deficit is one of the major constraints limiting the chickpea crop productivity [2] and has been reported to cause yield losses upto 50% in chickpea [3].
Deeper and more profuse rooting has been hypothesized to be the major factor contributing to yield increase under water limited environments in chickpea, where the assumption was made that these root traits would increase water extraction [4][5][6][7][8]. However it was also shown that chickpea genotypes with deeper and more profuse rooting did not extract more water from the soil profile [9]. Rather, tolerant chickpea genotypes turned out to be those able to somewhat limit water use at vegetative stage and making more water available for the grain filling period [9,10]. Similar results have been reported in other crops (e.g. in pearl millet [11], in sorghum [12]). Therefore, the central hypothesis of the present study is that, given the limited seasonal water budget, any trait allowing water conservation during vegetative growth (e.g. canopy conductivity, canopy size and development) extends the duration of water extraction during pod filling and so increases productivity of chickpea crop under terminal water stress [9][10][11][13][14][15].
In chickpea, the availability of large scale genomic resources has paved the way to dissect the mechanisms underlying various stresses adaptations [16,17]. A recent mapping study in chickpea reported a genomic region on CaLG04 referred as a "QTL hotspot" that harbours several drought tolerance traits including rooting depth [18]. Introgression of this region into elite variety JG 11 improved yield under drought [19]. This reported "QTL hotspot" region (spanning~29 cM) was originally associated with seven SSR markers [20]. Further, this "QTL-hotspot" region was refined to~14 cM, with additional 49 SNP markers, [20] using genotyping-by-sequencing (GBS). Skim sequencing (with bin used as markers, based on recombination break points) approach then allowed to fine map this region to~300 Kb [21]. An intriguing feature of the preliminary steps of this research was also the mapping of a major QTL for shoot weight on CaLG04, which co-mapped with a QTLs for root traits (depth and length density), from a study where these traits were assessed in PVC tubes [22]. Interestingly, the percentage of phenotypic variation explained by this QTL was more for the shoot dry weight than for the root traits, suggesting that this QTL region was a QTL for vigour, but this hypothesis was not followed further.
In chickpea, the studies of physiological traits allowing water conservation (e.g. canopy conductivity, canopy size & development; [9,10]) are very scarce, partially because an accurate assessment of leaf area is a rate limiting step. Recognizing this obstacle, a high throughput phenotyping platform was developed to measure canopy development traits [23]. The high throughput platform was used to phenotype the RIL population (ICC 4958 × ICC 1882), from which the "QTL-hotspot" was reported, for plant vigour traits (leaf area, plant height, rate of leaf area increase) and water saving traits (conductance), as a mean to re-investigate the map location of these traits with regards to the QTL hotspot earlier identified [18].
Therefore, the overall objective of this study was: i) to assess the phenotypic variation in traits involved in the control of plant water use either from canopy development or canopy conductance, and explore their functional associations in a RIL mapping population previously used for mapping the "drought tolerance QTL" (ICC 4958 × ICC 1882), ii) to map these drought adaptive traits and assess their interactions, iii) to conduct comparative mapping study using differently saturated genetic maps.

Canopy conductance traits
Summary statistics The two parental genotypes (ICC 4958 and ICC 1882) and RILs (progenies) showed significant difference for all canopy conductance traits (T, TR, eT, eTR & R-3D/PLA) in both years (2014-2015; Table 2). For example, T was one among those showing the largest phenotypic variation, i.e. a 5-fold range variation in both years (Table 2). In addition, TR also showed 2-fold range of variation (Fig. 1b). Continuous variation and normal frequency distribution was found for all traits (Additional file 2 E, F, G & Hdata not shown for R-3D/PLA). Transpiration and 3D-leaf area were tightly correlated (r 2 = 0.68) until the LAI reached a value of 1 (25 DAS). Thereafter, this relationship became much weaker (r 2 = 0.22) when the plants reached an LAI between 1 and 2 (38 DAS; see Fig. 2 a & b). At this stage, TR became much more closely related to T (r 2 = 0.92), whereas this relationship was weaker (r 2 = 0.62) when the LAI was less than 1 (25 DAS; see Fig. 2 c & d). Hence it was interpreted that at a low LAI, leaf area was the main driver of T. By contrast, at a high LAI, TR was the main driver of T. Since the average VPD during the transpiration measurement was high (3.76 kPa), this was interpreted to be caused by TR differences under high VPD.

Trait correlation and their relationships
Simple Pearson correlation analysis Phenotypic correlation coefficients of ICC 4958 x ICC 1882 population are presented in Additional file 4. As expected there were strong relationships within both groups of traits, but also between traits across groups. As expected, 3DL, LAI and SDW (plant vigour traits) were positively correlated with T and eT (canopy conductance traits), whereas 3DL, PL, LAI and SLA were negatively correlated with TR (see Additional file 4). Interestingly, most plant vigour traits were negatively correlated with R-3D/PLA (Canopy structure). By contrast, R-3D/PLA was positively correlated with TR and eTR. A significant correlation was observed among plant vigour traits. For example, plant vigour score (VIG) was significantly Principal component analysis (PCA) A principal component analysis was used to identify the relationships between parameters, and group these in a more comprehensive manner. Three principal components (PC) explained 62% of the total variation observed among the RIL population, using BLUPs phenotypic data across years (Additional files 5 and 6). PC1 (34%) had a strong positive loading from SLA and a strong negative loading from 3DL (Additional file 6), which agrees well with the strong negative correlations between these traits (Additional file 4). PC2 (17%) had a strong positive loading from PL and 3DL (plant vigour traits), whereas most canopy conductance traits had a strong negative loading. This also agreed well with the strong negative correlations between plant vigour (PL and 3DL) and canopy conductance (TR and eTR) traits (Additional file  (Additional file 7), and four minor QTL were distributed on CaLG01 (2 QTLs and PVE 5%), CaLG02 (1 QTL and PVE 5%), CaLG07 (1 QTL and PVE 5%; Additional file 7). For eTR, one M-QTL (LOD 6 & PVE 11%) was identified on CaLG04 with favourable allele from ICC 1882. This QTL was located just outside the "QTL hotspot" region (Additional file 7  Fig. 3).

Co-localization of plant vigour and drought tolerance related traits
Map position of plant vigour traits reported here was compared to map position of roots and drought tolerance traits reported earlier [18,20,21]. With the low density marker map, plant vigour traits co-localized with several root traits [eg. root length density, root dry weight/total plant dry weight ratio); see [18] from the previously reported "QTL-hotspot" region ( Fig. 4-I-A [20]) on CaLG04, which gave also a refined "QTL hotspot" region ( Fig. 4-II-A, B&C; Additional file 11).
Asserting QTL location and size in different genetic maps Different density genetic maps showed QTLs for plant vigour traits on CaLG04 and co-located with the "QTLhotspot" region. Their size within the "QTLhotspot" region using the low density (29 cM size), high density (1 5 cM size) and ultra-high density maps ("QTL-hotspot" a & b (see more details on [21]) together~300Kb size) on CaLG04 is discussed in this section.
For plant vigour related traits (VIG, 3DL, PL, PH, PHG, LAI, SDW), 28 and 32 M-QTLs were mapped on the low and high density maps, and their size ranged from 1 cM to 8.0 cM on the low density map and 0.8 cM to 5.6 cM on the high density map. For the same traits, the 15 M-QTLs that were mapped using the ultra-high density marker map (Table 3) had a size ranging from 0.14 cM to 0.15 cM. For instance, Fig. 5-I, II, III, IV-A, B&C showed plant vigour traits (VIG, 3DL, PH and SDW) in three different genetic maps. It showed that gradually LOD and PVE increased with marker density and simultaneously QTL size decreased, being fine-tuned down to 300Kb with the ultra-high density marker map. More details on major and minor QTLs for plant vigour in different density genetic maps are presented in Additional files 7, 10 and 11. In addition, low density genetic map along with plant vigour traits QTLs position are shown in Additional file 12A.
For canopy conductance traits (TR, eTR, T, eT and R-3D/PLA), several QTLs were identified on different linkage groups (CaLG01, LG03, LG04, LG05, LG06, LG07 & LG08) across the genome. A total of 18 and 20 M-QTLs were mapped on different LGs using low and high density maps, respectively (see more details in Additional files 10 and 11). The QTL size ranged from 1 cM to 15.0 cM size in low density map and high density ranged from 0.3 cM to 5.0 cM size (Table 3). Two M-QTLs were mapped on CaLG03 (TR) and just outside the CaLG04 "QTL-hotspot" region (eTR) using ultra-high density map. The QTL size ranged from 0.08 cM (TR) to 0.48 cM (eTR) size (Table 3). For TR, three M-QTLs with 5-13 cM were identified on CaLG07 using low density marker (Table 3). In the high density marker, no M-QTL was detected for TR. But, six minor QTLs were identified on CaLG03 (2QTLs; 4.9-5.1 cM), CaLG07 (1QTL; 2.0 cM), CaLG06 (1QTL; 10.3 cM) and CaLG04 (2.3-11.9 cM; Additional file 11). On the ultra-high density map, one M-QTL for TR was mapped on CaLG03 (0.08 cM). For TR, mapping position varied between low and ultra-high density markers. This might be most of the similar alleles between CaLG03 and CaLG07 (Table  3) (Table 3 & see more details in Additional file 11). There was no QTL was detected with the ultra-high density markers map (Table 3). More details of canopy conductance traits major and minor QTLs in different density genetic maps are presented in Additional files 7, 10 and 11. In addition, low density genetic map along with canopy conductance traits QTLs position is shown in Additional file 12B, C & D.

Discussion
The summary of the main results is as follows: i) Genetic variation of 16 phenotypic traits revealed two clusters of plant vigour and canopy conductance traits and their association was clarified with PCA analysis and correlation. ii) Using the ultra-high density map, M-QTLs for plant vigour traits predominantly mapped on CaLG04 and these co-mapped with a previously refined "QTLhotspot" region (~300Kb) for drought tolerance traits. Canopy conductance traits were mapped in CaLG03 (TR) and CaLG04 (eTR). iii) The refined "QTL-hotspot" region (Bin-Map) was further sub-divided into a "QTLhotspot-a" and "QTL-hotspotb" regions. While both "QTL-hotspot" sub-regions co-mapped with previous study [21], the phenotyping data at a lower level of plant organization gathered here led us to interpret that region 'a' (139.22Kb or 0.23 cM) could be a locus for branching and tissue/organ expansive processes while region 'b' (153.36Kb or 0.22 cM) could be interpreted as a locus for physiological processes related to biomass accumulation. iv) As marker density increased QTL number and size decreased (~29 cM to 0.22 cM); and LOD and PVE (%) increased for most of the QTLs. v) Most of plant vigour traits had alleles from high vigour parent ICC 4958 whereas in the case of canopy conductance traits (eTR and TR) the favourable alleles were contributed by the low vigour parent ICC 1882. vi) Plant vigour traits mapped mostly on CaLG04 whereas canopy conductance traits mapped on CaLG03, providing an opportunity to manipulate these loci to tailor recombinants having lower transpiration rate and high plant vigour desirable for water limited environments.

Phenotyping at different level of plant organization
The vigour traits (3DL, PL, SDW, PH, and 3D-LG) were tightly linked to plant water use traits. These traits were reported to be linked to crop biomass production and then crop yield [25,26]. The co-localization study clearly demonstrated the close relationship between traits from the present study at a lower level of plant organization (eg. 3D-Leaf area, growth rate) and the agronomic traits (eg.shoot biomass, harvest index) studied previously by Varshney et al. [18]. Canopy development traits had also a clear effect on crop production [25]. Although phenotyping of traits at a lower level of plant organization is usually laborious and time-consuming process, it was facilitated by the use of a high throughput phenotyping platform (LeasyScan). Most of the plant vigour traits had high heritabilities, making them suitable for breeding applications. The high vigour parent ICC 4958 had higher biomass and water use (absolute T) than the low vigour parent ICC 1882. By contrast, the high vigour parent had lower transpiration rate (TR; g of water transpired per unit of leaf area) than low vigour parent. Hence, the cause for such response in water-use was the difference in leaf area (vigour/canopy development). The effect of such combination, having high vigour and lower TR would be then of high value to test across time and geographical scale using crop simulation analysis. Crop simulation modelling of water saving traits (eg. limited transpiration rate) has indeed shown a clear yield advantage under terminal drought stress conditions (Soybean- [27], Maize- [28] and Sorghum- [15]).

Co-localization of plant vigour traits and previously identified drought tolerance traits in different genetic maps
Early plant vigour is an important trait for water limited environments. It may contribute to shading of the soil surface, thereby reducing evaporation of water from the soil and leaving more water available for the crop [25,29,30]. In the present study, most of the plant vigour traits had several M-QTLs on CaLG04 and co-mapped with the earlier reported fine mapped "QTL hotspot" region [18,20,21] with QTLs for root traits. The alleles for these vigour traits were contributed by high vigour parent ICC 4958. Here is a first detailed study reporting the co-mapping of plant vigour traits with root and socalled drought tolerance traits. This is also a confirmation of the earlier observations that shoot dry weight and root length density QTL co-mapping in preliminary results [22]. This result, therefore, suggests that the drought tolerance reported earlier to be associated with that QTL in the hotspot region (241 Low density SSR marker- [18]; 1007-High density GBS markers- [20] and Ultra-high density Bin maps- [21]) would actually be conferred by plant vigour aspects. Such result was also predicted by a crop simulation study [31] that concluded that in the short duration environments where chickpea cultivation is now mostly cultivated, a high plant vigour associated with faster rooting would be necessary to reach the water available deep in the soil profile. Similarly, in recent pearl millet mapping studies [32,33] it was reported that plant vigour traits also co-localized with agronomic traits related to terminal drought   tolerance [34]; drought index of stover yield, grain yield, biomass yield and harvest index [35][36][37]. Similarly, another study in a high-resolution cross (HRC) population of pearl millet showed that plant vigour traits (3D-leaf area, plant growth rate, plant height) measured from LeasyScan co-localized with yield traits measured in the field under different water stress treatments (Tharanya et al-unpublished data. The present study suggests that high root length density obtained earlier [18] could be more easily proxied by vigour traits at the canopy level, which would then ease the phenotyping of that particular trait. Overall, plant vigour traits might lead to high biomass, which would then link to higher yield potential. Therefore, the genotypes that have alleles from ICC 4958 would be beneficial for water limited conditions.

Binmap QTL hotspot region
With the ultra-high density marker, the refined QTLhotspot was sub-divided into two sub-regions "QTLhotspot"-"a" & "QTL-hotspot"-"b". Our interpretation, on the basis of the phenotyping at a lower level of plant organization done in the present study, is that these two regions could control two domains of physiological processes. "QTL-hotspot"-"a" region, which had QTL for traits related to vigour and growth rate (PH and VIG), could be interpreted as a region coding for branching and expansive processes. We interpret the possible effect on the branching from the two fairly opposite phenotypes of the parents of the population used here, i.e. highly branched ICC 1882 with low height versus less branched but taller ICC 4958. More work would be needed to decipher in more details the possible interaction between height and branching. The interpretation of the expansive processes comes from recent genetic work on regions controlling leaf expansion in maize [38], and where vigour could simply be consequences of differences in the expansive processes leading to larger organ sizes and quicker development. Interestingly, this region 'a' was earlier reported to harbour QTL for pod number per plant, 100-seed weight and plant height [21], although a finer analysis of the plant processes possibly involved was not done. It was particularly interesting to see that this region led to seed size differences, which then raises the question whether seed size is not itself controlled by expansive processes at the time of embryo development and seed formation. "QTL-hotspot"-"b" region could then simply be a locus controlling physiological processes involved in biomass accumulation, which was corroborated by the QTLs found here for 3DL, LAI and SDW, or for biomass traits SDW and RTR traits in Kale et al. [21].  Twelve candidate genes were reported from this fine mapped "QTL hotspot" region (see, [21]) stating that most of the genes were involved in abiotic stress tolerance. The same genes were also reported to be associated with plant growth and development related functions (e.g. genes of serine threonine-protein kinases, E3 ubiquitin ligases, Leucine-rich repeat extension (LRXs), Protein IQ domain and Vicilin 47 K and Cotyledon vascular pattern (CVP2) genes that were reported to be associated with drought stress adaptation by Kale et al., [21] were also reported to be associated with plant growth and development related process [39][40][41][42][43][44][45][46][47][48]. These reports additionally suggest that earlier reported "QTL-hotspot" region to be associated more likely with vigour related traits.

Ideotyping of plant vigour and canopy conductance genomic regions
An ideal ideotype for water limited environment would be the one having higher plant vigour (the proxy for higher biomass and yield) potential with restriction of transpiration under high VPD conditions. These combinations would achieve higher water use efficiency, eventually soil moisture conservation, and then ultimately lead to crop production success. The plant vigour traits were mapped on CaLG04 and the canopy conductance (eg. TR) traits were present on CaLG03. These two genomic regions contributed more than 75% QTLs for plant water use (vigour and conductance) traits. Therefore, CaLG04 (plant vigour) and CaLG03 (canopy conductance) provide an opportunity to manipulate these loci to tailor recombinants having alleles with lower transpiration rate along with high plant vigour. This ideotype might be useful in enhancing the water stress adaptation in chickpea. Similar kind of ideotyping was recommended in pearl millet [32,33]. Recent modelling reports on sorghum [15] showed that alteration of leaf area (plant vigour components) and transpiration rate increased grain yield under severe stress conditions. This  Figure 5-I represent plant vigour QTL peak; 5-II represent 3D-leaf area peak; 5-III represent plant height QTL peak and 5-IV represent shoot dry weight QTL peak study suggests that plant vigour and transpiration rate trait assessed in the current study might also have an effect on crop production success in specific target environments.

Conclusion
The present study has shown that a previously identified "QTL hotspot" region on LG04 of chickpea and harbouring QTL for root traits and so-called terminal drought tolerance in chickpea was a vigor locus, with favourable alleles from high vigour parent ICC 4958. Our phenotypic analysis at a lower level of plant organization led us to interpret that this locus may be divided into two subregions, one coding for expansive processes and one for biomass accumulation. Another genomic region on CaLG03 harboured QTL for canopy conductance traits (e.g. TR). Plant vigour and canopy conductance traits were somewhat negatively related but being mapped on different chromosome provides an opportunity to manipulate these loci to tailor recombinants having lower transpiration rate and high plant vigour which would be useful enhancing the drought adaptation in chickpea. In addition, potential genomic region on CaLG04 with simple vigour traits (e.g vigour score) could be used for breeding programs through marker assisted backcross (MAB) to devolep improved variety. Enrichment of the marker density reduced QTL size and increased in LOD and PVE% for all plant vigour and canopy conductance traits.

Plant material
The genetic material was a set of 232 recombinant inbred lines from a population derived by single seed descent method from the cross between ICC 4958 and ICC 1882 and advanced to F10+ generation [18]. Genotype ICC 4958 is a drought tolerant breeding line developed by Jawaharlal Nehru Krishi Vishwa Vidyalaya, Jabalpur, and Madhya Pradesh, India. It has a large root system, early vigour is early to reach 50% of flowering (608 cumulative degrees) and maturity (1650 cumulative degrees). The ICC 1882 landrace was collected in India and added to the ICRISAT's genebank in 1973. It has a small root system, late vigour, is later to reach 50% of flowering (779 cumulative degrees) and maturity (1806 cumulative degrees) compared to ICC 4958 [8,18,49]. These two parental lines were contrasting for root traits and plant vigour i.e. were used for mapping population development. Additional detail account on parental lines and mapping population are provided in Varshney et al. [18].

Crop Phenotyping Plant growth conditions
Phenotyping was performed from November to December 2014 & 2015 in the LeasyScan facility [23]. Plants were sown during the post-rainy chickpea sowing window (November). Plants were grown in 27 cm diameter plastic pots filled initially with 9 kg of dry black soil (Vertisol) collected from ICRISAT farm. Each experimental unit in the LeasyScan platform was composed of 2 pots, each containing 4 healthy plants. These experimental units being of 65 × 40 cm, i.e. approximately 0.25 m 2 , the sowing density was 32 plant m − 2 , which is equivalent to the sowing density in the field. In other words, phenotyping was done on a crop canopy that had close similarities with a field situation. Sowing was done with 6-8 seeds per pot and seedlings were thinned to maintain four homogeneous seedlings per pot at 12 days after sowing (DAS). Fertilizers were provided with single super phosphate (SSP) as basal dose at the rate of 0.3 g/ kg of soil. The experimental design was an Alpha lattice with 4 replications and 24 blocks of 10 genotypes in each replication to avoid geographical variations. Plants were maintained under well water conditions the throughout experiment. During the crop grown period, 11/35.8°C minimum and maximum temperature and 17.2/93.2% relative humidity were observed.

Phenotypic traits evaluated
Sixteen phenotypic traits were measured and categorized into three groups: (i) Canopy traits (measured by Lea-syScan) (ii) Transpiration traits (measured by gravimetric balance system) and (iii) Biomass traits.
i) Canopy traits LeasyScan PlantEye ® scanners measured canopy development related traits [3DLeaf area (3D-L), projected leaf area (PL) and plant height (PH)] on the hourly basis during crop growth periods. Using these traits, plant growth rate related traits [3D-Leaf area growth rate (3D-LG), projected leaf area growth rate (PLG), plant height growth rate (PHG)] were calculated. Plant growth rate (3DLG, PLG, PHG) was calculated based on the average difference in respective leaf area and plant height between consecutive days during the exponential growth phase. The leaf area index (LAI) was estimated as the projected leaf area PL divided by the area of the pots in the sector. Plant vigour score was estimated by visual eye basis, on a scale from 1 (low vigour) to 5 (high vigour) at 20 DAS after sowing, all four replications being scored by one person eye visual score. Similar protocal was reported in other crop species such as wheat [50] and maize [51]. Residual (canopy structure) was calculated by using 3D-leaf area and projected leaf area.
ii) Transpiration traits Transpiration (evapotranspiration (eT)) was measured by a gravimetric method (see [32]). The pots were watered abundantly and drained overnight to attain field capacity. An extra 20 pots without plants were also brought to field capacity and were there to evaluate soil evaporation. Following day, plants were manually weighed (Model FCB 24 K0.2B, KERN & Sohn GmbH, Baligen, Germany.). All four replications were weighed between 6 and 7 am (Initial weight; average VPD~0.8 kPa). Pots were weighed again late afternoon between 3 and 4 pm (final weight; average VPD~3.76 kPa), following the same sequence of pot weighing as in the morning. Evapotranspiration was calculated by the difference between initial and final pot weight. Further, plant transpiration (T) was estimated by subtracting an estimate of soil evaporation (pot without plant soil evaporation). Briefly, it was assumed that soil evaporation in planted pot would be maximum with zero plant cover, and would be zero at a leaf area index of 2.
Therefore, the projected leaf area was used to infer a LAI. Briefly, LAI = PL/area of the pots in the sector. At the time of eT measurements and transpiration values were estimated from this correction. While this may have induced some error, we made the assumption the method would be correct for genotypic comparison and QTL analysis. Transpiration rate (TR) and evapotranspiration rate (eTR) were calculated by transpiration and evapotranspiration divided by 3D-leaf area and time [52].
iii) Biomass traits At the end of the experiment (canopy covered maximum in the pot; 35 DAS), shoot samples were harvested and over dried at 65°C for 48 h. Further, shoot dry weights (SDW) were weighed using gravimetric balance (KERN 3Kg) method. Specific leaf area (SLA) was estimated by leaf area divided by shoot dry weight. Specific leaf weight (SLW) was estimated by 1/SLA (inverse of SLA).
QTL analysis-single locus QTL analysis was conducted independently using three genetic maps developed earlier [18,20,21] and phenotyping data generated in this study. QTL Cartographer version 2.5, composite interval mapping (CIM) method was employed [53]. For ultra-high density bin markers, inclusive composite interval mapping-Additive mapping (ICIM-ADD) method was used for identification of QTLs using IciMapping software (v3.2; [54]). LOD threshold was set by using 1000 permutation and p value ≤0.05. Constructed linkage map was visualized using Mapchart 2.2 [55] software. When the PVE (phenotypic variation explained) was above 10%, QTLs were considered major QTLs (M-QTLs) and PVE below 10% were minor QTLs.

Interactions QTL analysis-multi-loci
The QTL interactions influencing the traits were identified using Genotype Matrix Mapping software (GMM; v. 2.1; [56], http://www.kazusa.or.jp/GMM). Using GMM, two and three loci interactions were tested. GMM analysis showed interactions between loci and different linkage groups of plant vigour and canopy conductance related traits. The current study identified allelic interactions that contributed to either a positive (increase) or negative (decrease) effect on the phenotypic value of the trait. In most cases, single locus QTL identified using GMM analysis were similar to those identified with CIM analysis, even though two approaches use different algorithms. In the following text, symbols "AA", "BB" and "stand for alleles originated from the high vigour parent (AA; ICC 4958) and low vigour parent (BB; ICC 1882)" and not distinguished from any parent (−), respectively.

Statistical analysis
To find the phenotypic variations and their significance in the population, ANOVA was performed for all observed parameters individually using GENSTAT 14.0 (VSN International Ltd., Hemel Hempstead, UK). Similarly, to find the phenotypic variations and their significance in parental lines were analyzed with statistical program package CoStat version 6.204 (Cohort Software, Monterey, CA, USA). Oneway ANOVA was carried out to test for genotypic difference between the genotypes. Means were compared using Tukey-Kramer test and Least Significant Difference (at P ≤ 0.05). Normal histograms with frequency distribution analysis for phenotypic traits were done using SPSS 16 desktop version (IBM, SPSS Statistical software). Principal component analysis (PCA) was used to visualize the relationships between traits in a multidimensional space using R software (version 2.11.1). To find the trait correlation of all phenotypic traits, simple Pearson correlation was performed using R software (version 2.11.1). For QTL and PCA analysis, Best Linear Unbiased Predictors (BLUPs) data were estimated by using GENSTAT 14.0 were used. The clustering analysis was performed by PCA loadings using R software (version 2.11.1). Genotypic and residuals mean square components were obtained from ANOVA through GENSTAT 14.0, which was used to calculate the broad sense heritability (h 2 ). The broad-sense heritability (h 2 ) was calculated as h 2 = σ Thiyagarajan Thirunalasundari is working as Professor and Head, Department of Industrial Biotechnology, Bharathidasan University at Tiruchirappalli in India. Pooran M. Gaur is Principal Scientist (Chickpea Breeding) at ICRISAT in India. He has extensive expertise on chickpea mapping population development and marker assisted breeding. Rajeev K Varshney is Research Program Director -Genetic Gains at ICRISAT in India. Heis internationally recognized for his contribution in genome sequencing of pigeonpea, chickpea, peanut, pearl millet, sesame, mungbean and azuki bean and development of molecular breeding products.
Ethics approval and consent to participate Not applicable -Data were generated from our own trials. The genetic material that was tested was readily available at ICRISAT and did not required us any kind of permit or request.

Consent for publication Not applicable
Competing interests