Legumes are very important sources of vegetable protein for human food and animal feed and these plants do not require nitrogen fertilization. Drought represents one of the main factors which limits legume productivity in the Mediterranean basin. A better understanding of the mechanisms for adaptation to water deficit in legumes is a prerequisite for any research aiming to improve legume yields. The slowness of genetic improvements in most legumes of economic interests is due to some unfavorable traits such as their large genome, their complex ploidy, and the difficulties of their transformation and regeneration. M. truncatula thus emerged as a model legume because of several unique characteristics (Barker et al., 1990). It is diploid (2n=16), self-pollinating, has a low DNA content per haploid genome (~465 Mbp), is suitable for cultivation in the laboratory (size, growth habit, flowering, etc.) and has a rapid lifecycle. Furthermore, several studies on M. truncatula have contributed to an expanded genetic and genomics tools (Tang et al., 2014).  It is also an important forage crop species in several countries especially Australia.

In recent years, molecular mechanisms for tolerance to drought stress in plants have become an active area of investigation and many genes regulated by drought stress have been reported in a variety of plants (Sehgal et al., 2012; Thudi et al., 2014; Xu et al., 2014). Very little is known about the physiological and molecular mechanisms by which the model legume M. truncatula responds to drought stress. In most of the previous experiments plants were terminated at the vegetative stage (Badri et al., 2011; Wang, 2014; Zhang et al., 2014). Furthermore, these previous studies have used only one population of RILs to analyze the genetic determinants for tolerance to drought stress. Therefore, it is important to analyze the genetic bases of tolerance to this constraint by using more than one population of RILs in addition to studying their responses at both the vegetative and reproductive stages.

The objective of this study was to evaluate the effects of water deficit on photosynthetic and growth charac-teristics in parental genotypes of RILs of M. truncatula at flowering and end of lifecycle stages. This as well as similar investigations can lead to the selection of more than one appropriate RIL population useful for the analysis of genetic determinism to drought.  Furthermore, knowledge obtained by studying M. truncatula can be transferred to other legumes.

AWC = 100(AFW-ADW)/AFW

RWC = 100(RFW-RDW)/RFW

Where FWL, AFW and RWC are the fresh weights of leaves, aerial part and roots, respectively, while LDW, ADW and RDW are the dry weights of leaves, aerial part and roots, respectively. The aerial part is the leaves plus stems.

The leaf water potential (Ψh) was measured using a pressure chamber (Soil Moisture Equipment, Santa Barbara, CA, USA) (Scholander et al., 1965).

Photosynthesis measurements

Gas exchange measurements (net CO₂ assimilation rate (A), transpiration rate (E), stomatal conductance (gs), and instantaneous water-use efficiency (WUEi) as the ratio A/E were done on mature leaves using a CO₂ and H₂O infrared gas analyzer (Li-Cor 6200, Li-Cor Nebraska, USA). Measurements were performed between 10 and 12 am on the leaves of plants subjected to various treatments as described by Hessini et al. (2013).

Statistical analyses

A three-way analysis of variance was used to test for line, drought treatment differences, and line x treatment interaction effects. Only characters that showed a significant line x treatment interaction were used for further analysis. Means were compared using Duncan's multiple range test at 5%. Correlations between parameters were calculated using Pearson’s correlation method. The significance level of associations between morphological traits and photosynthetic parameters was set to 0.05, and adjusted for multiple comparisons by Bonferroni corrections. All analyses were performed using SPSS software (SPSS Inc. Released  2007  SPSS for Windows, Version 16.0. Chicago, SPSS Inc.). Drought response index (DRI) is considered as an indicator of tolerance to water deficit (Chen et al., 2007) and is estimated as the ratio between the values ​​observed under water deficit conditions and those obtained in the control treatment.

Broad-sense heritability (H²) for each trait was estimated as:

Where, σ²_g is the genetic variance observed between the lines and σ²_e is the environmental variance corresponding to the residual error between the eight replicates of the same genotype (=line).

Drought response indices (DRI) for measured traits in lines of M. truncatula grown under 50 and 30% of field capacity were subjected to principal component analysis (PCA). This analysis was carried out using the XLSTAT software version 7.5 (Addinsoft, 2007, New York, USA).

In 30% FC, the first component accounted for 29.58% of the total variation between studied lines while the most contributing  parameters  are   the   root   fresh   and   dry weights, the number of pods and the harvest index. The second component explained 24.47% of the total variation among the 10 lines while the most contributing parameters are the photosynthetic rate, the transpiration rate and the stomata conductance. The third component accounted for 17.26% of the total variation among lines with the number and area of leaves being the most contributing characters. The scatter diagram of studied lines defined by the three first principal components shows three groups (Figure 2). The first group consisted of Jemalong A17 while the second group was formed by DZA45.5 and F83005.5, and the third group was composed by the remaining lines. The lines DZA45.5 and Jemalong A17 are negatively associated with PC1 while TN1.21  and  TN8.3  are  positively  associated  with PC1(Table 5). DZA45.5, TN8.20 and TN8.3 are positively correlated with PC2 while DZA315.15, F83005.5 are negatively associated with PC2. Jemalong A17 and TN6.18 are positively associated with PC3 while DZA45.5 and F83005.5 are negatively correlated with PC3. Overall, our results showed that similarities between lines of M. truncatula were dependent on the treatment effects (50 and 30% FC).

Lines of M. truncatula are potentially subjected to different selection pressures (Badri et al., 2007; Lazrek et al., 2009) in their natural areas of distribution therefore variability is expected in response to water deficit in the different genotypes used in this study. Lines may have developed an adaptive response to drought that can be exploited later to identify the genetic bases of tolerance to drought. The analysis of 27 morph-physiological characters showed significant variation among the 10 lines of M. truncatula. The treatment factor explains the maximum of variation of measured traits. Out of the 27 characters, 14 were explained by the interaction treatment x line indicating an involvement of these parameters in the adaptation of lines to water stress.

Positive correlations between flowering time and growth traits showed that late flowering lines have well-developed vegetative organs. The same observation was noted in chickpea (Soltani et al., 2006) and populations of M. laciniata and M. truncatula (Badri et al., 2007). Indeed, it has been shown that the genotype of plants, temperature and photoperiod factors affect flowering time in chickpea (Berger et al., 2006; Soltani et al., 2006) and rice (Ouk et al., 2006). For some accessions of M. truncatula, vernalization is a significant factor in reducing time to flowering when days are long and temperatures are moderate (e.g. 16 h/8 h and 21-25°C) (Chabaud et al., 2006). Under greenhouse condition, with seasonal changes in day-length and temperature, the effect of vernalization on time to flowering is secondary to the photoperiodic response. Annual Medicago grow best under long days and warm temperatures (De Ruiter and Taylor, 1979) and on neutral to alkaline soils (Nichols et al., 2010). Early flowering is an adaptation to escape adverse conditions, as observed in populations of chickpea in India (Berger et al., 2006). This adaptive characteristic allows lines of M. truncatula to survive in unfavorable regions for growth by restricting their development to short rainy periods of the year. This result is consistent with results observed in Algerian populations of M. truncatula (Si Ziani and Abdelguerfi​, 1995), where flowering time was correlated with altitude and annual  rainfall  of collection sites. A high growth rate can potentially compensate for low plant vigor at flowering, but this reduces their tolerance to drought (Mitchell-Olds, 1996).Our results showed that DZA315.16 line allocates more effort to root growth under both irrigation regimes (50 and 30% FC). The importance of the root system in the acquisition of water has long been recognized (Lynch et al., 2014). The development of the root system increases the water absorption and maintains the osmotic pressure. An increase in the root system under water stress has been reported in sunflower (Tahir et al., 2002), and in Sesuvium portulacastrum (Slama et al., 2007). Water deficit reduces the growth of leaves more compared to roots because the growth of roots and leaves exhibits different sensitivities to reduced water availability (Verslues et al., 2006). This implies a reallocation of carbon (C) to the roots, even if the rate of net assimilation of C is reduced due to the low diffusion of CO₂ caused by the limited stomata opening under impeding water loss (Cornic, 2000). The water deficit potentially increased availability of C for roots (Hummel et al., 2010).

In the current study, the reduction of growth is associated with a decrease in the number of leaves. Strong positive correlations were recorded between the biomass of whole plant and number of leaves (NL) for studied lines under control treatment and drought stress. Indeed, plants reduce water loss by restricting their leaf area transpiration.

The decrease of plants growth under water deficit is also due to an inhibition of photosynthesis (Cornic, 2000). In this study, lines of M. truncatula showed a decrease in net CO₂ assimilation under water deficit, which is the highest value recorded for F83005.05. This decrease is mainly due to the closure of stomata and fewer leaves and/or leaf area as part of a strategy adopted by plants to reduce their transpiring surface. The reduction in the number as well as the biomass of leaves suggest that net CO₂ assimilation (A) per unit of leaf area is higher in plants subjected to drought stress than those grown in non-limiting conditions. Accordingly, Slama et al. (2007) reported a decrease of number and size of leaves and an increase in net CO₂ assimilation in leaf growth rate per unit under water deficit in S. portulacastrum. This reduction saves water for the plant and is therefore considered as a criterion for adaptation to drought. Our results showed negative correlations between net CO₂ assimilation (A) and flowering time (FLOR), indicating that plant growth was affected under water deficit due to reduced capacity of photosynthetic organs (that is, the amount of assimilates). This reduction can be harmful, from an agronomic point of view, because it is irreversible when the water conditions become favorable after droughts (Flexas and Medrano, 2002). According to Chaves et al. (2003), most plants tend to increase their water use efficiency (WUE) when they are subjected to moderate drought stress. This increase is due to a non-linear relationship between carbon assimilation and stomata conductance. Stomata conductance is one of the main factors affecting photosynthesis of plants (Medrano et al., 2002). It appears that plants use efficient mechanisms to limit the loss of water before the inhibition photosynthesis.

High to moderate values ​​of heritability (H²) were registered for most measured parameters in the control treatment, 50 and 30% FC, indicating that much of the variation of these traits is under genetic control. The spatial structure of studied lines in the three first principal component based on their responses to water deficit in 50 and 30% FC showed that lines were differently affected  by  drought  stress   (Figure   2).   In   30%   FC, Jemalong A17 exhibited the lowest reductions for FWL, NPOD, W100P, and HI while it showed the highest reduction for LA. Hence, all populations of recombinant inbred lines derived from crosses between Jemalong A17 and the remaining lines (TN1.21, TN1.11, A20, DZA315, DZA45 and F83) (Figure 1) are useful for the identification of the genetic determinants for tolerance to water deficit in M. truncatula.

Overall, the variability of measured parameters was mainly explained by the treatment factor. High to moderate broad-sense heritability (H²) were registered for most traits under control treatment and drought stress. The high heritability of most traits indicates that genetic factors are most likely controlling these characteristics even in water deficit conditions. Further genetic studies will be required to dissect the specific genetic factors that are responsible for these traits. Most of the correlations between the characters are positive under the three water regimes. The tolerance of lines to water deficit seems to depend, in particular, on their aptitude to maintain a good photosynthetic activity. Knowledge obtained by studying this model plant can be transferred to other crop and forage legumes.

The authors have not declared any conflict of interests.

The authors thank Ken Moriuchi and Marshall Keyster for the helpful comments on the manuscript, and Kamel Hessini and Fethia Zribi for the photosynthesis measurements. They are grateful to Mohamed Elarbi Aouani for the useful discussion, Mohamed Salhi and Najah Ben Cheikh for assistance in the greenhouse. This work was supported by the Tunisian Ministry of Higher Education and Scientific Research (LR10 CBBC 02), and the FP6 European Grain Legume Integrated Project.