Diversity and Evolution of Salt Tolerance in the Genus Vigna

Breeding salt tolerant plants is difficult without utilizing a diversity of wild crop relatives. Since the genus Vigna (family Fabaceae) is comprised of many wild relatives adapted to various environmental conditions, we evaluated the salt tolerance of 69 accessions of this genus, including that of wild and domesticated accessions originating from Asia, Africa, Oceania, and South America. We grew plants under 50 mM and 200 mM NaCl for two weeks and then measured the biomass, relative quantum yield of photosystem II, leaf Na+ concentrations, and leaf K+ concentrations. The accessions were clustered into four groups: the most tolerant, tolerant, moderately susceptible, and susceptible. From the most tolerant group, we selected six accessions, all of which were wild accessions adapted to coastal environments, as promising sources of salt tolerance because of their consistently high relative shoot biomass and relative quantum yield. Interestingly, variations in leaf Na+ concentration were observed between the accessions in the most tolerant group, suggesting different mechanisms were responsible for their salt tolerance. Phylogenetic analysis with nuclear DNA sequences revealed that salt tolerance had evolved independently at least four times in the genus Vigna, within a relatively short period. The findings suggested that simple genetic changes in a few genes might have greatly affected salt tolerances. The elucidation of genetic mechanisms of salt tolerances in the selected accessions may contribute to improving the poor salt tolerance in legume crops.


Introduction
Soil salinity is one of the major factors responsible for losses in agricultural production. Saltaffected lands cover approximately 10% of the global land surface and nearly 50% of the irrigated cropland worldwide [1]. Although screening for salt tolerance has been extensively performed on staple food crops, in order to develop salt tolerant crop varieties [2], few accessions have been identified [3][4][5]. A major problem is the low diversity of salt tolerance in domesticated accessions, especially in legume crops. In the family Fabaceae, the majority of salt tolerant or halophytic species are limited to shrubs or woody plants [6], which are genetically divergent from crop species, and are not suitable for agricultural use.
One solution to overcome the difficulties associated with salt tolerance breeding is the use of wild crop relatives. Although wild ancestors of crops are expected to have adaptability to unfavorable conditions associated with their original habitats, domesticated crops may have lost such adaptability during the domestication process, given that abiotic stress tolerance is not necessary under protective cultivations. Therefore, the greater diversity in salinity tolerance of wild genetic resources should be harnessed. However, in the majority of cases little attention has been paid to the wild crop relatives because of insufficient genetic and phenotypic information [7], although successful screening and analysis have been performed for some species [8,9]. Recently, the intensive use of wild relatives for crop improvement has been proposed, with support of plant biotechnology [10]. This trend will promote the exploitation and use of wild crop relatives that have yet to be evaluated.
One good example to demonstrate the potential of wild crop relatives is the genus Vigna, family Fabaceae. The genus Vigna is comprised of over 100 species, including nine that contain domesticated or semi-domesticated accessions. They are distributed across tropical and subtropical regions, including coastal regions [11]. In a previous study, we identified two salt tolerant accessions (ID 17 and 18 in this study) from wild germplasm, which are cross compatible with azuki bean (V. angularis), an important Vigna crop in East Asia [12]. In another study, genetic analysis on a beach species V. marina (ID66) identified a strong QTL for salt tolerance [13]. In addition, the majority of the species are diploid, with relatively small genome sizes (500-600 Mb), and are capable of some interspecific crosses [14]. As such, the wild Vigna accessions are valuable genetic resources for gene isolation and cross breeding via straightforward genetic approaches.
However, to effectively harness the salt tolerance of genus Vigna, a systematic evaluation of various accessions across various species is necessary. To date, several salt tolerance mechanisms such as Na + exclusion, increasing antioxidant ability, osmotic regulation, and changes in hydraulic conductance have been studied in Vigna species [15][16][17][18][19][20][21], but these studies were limited to domestic varieties. We evaluated the salt tolerance in a total of 69 accessions, including wild and domesticated accessions collected from Asia, Africa, Oceania, and South America. We measured plant responses to different intensities of salt stress (50 and 200 mM NaCl) under a high transpiration environment in a greenhouse. The objectives of this study were to not only identify salt tolerant accessions but also estimate the mechanisms of tolerance by measuring photosystem activity, biomass, leaf Na + accumulation, and K + /Na + ratios. We also estimated the divergence time of accessions by phylogenetic analysis. Based on the results, we classified the accessions into four groups (the most tolerant, tolerant, moderately susceptible, and susceptible) and selected six wild accessions as promising resources of salt tolerance. We also found that mechanisms of salt tolerance were diverse in the tolerant accessions and that salt tolerance had independently occurred several times during speciation, within a relatively short time. As such, the information on salt tolerance provided here will promote the use of Vigna genetic resources for crop improvement, cloning genes of salt tolerance, and understanding salt tolerance evolution. Table 1 shows all 69 accessions in the genus Vigna used in this study, which included 15 domesticated and 54 wild accessions selected from wide ranges of their original habitats. Eight species contained both domesticated and wild accessions. Accessions in the subgenus Ceratotropis (taxonomic sections Angulares, Ceratotropis, and Aconitifoliae) represent the Asian Vigna, whereas those in the subgenus Plectrotropis (section Plectrotropis) and the subgenus pool. We set the moderate salt stress at 50 mM NaCl for two weeks according to our previous study in which sensitive accessions severely wilted within two weeks even at 50 mM NaCl concentration [12]. We also set the severe stress condition of 200 mM NaCl for two weeks so that we could evaluate the differences in more tolerant accessions. As biological replication, two pots (two plants) per accession were used for each stress treatment. After the stress treatment, chlorophyll fluorescence, leaf Na + and K + concentrations, and shoot biomass were measured to evaluate the effects of salt stress. The uppermost fully expanded leaves of all plants were measured for chlorophyll fluorescence and leaf Na + and K + concentrations. The dry weight of the remaining shoot of each plant was measured after incubation at 80°C for 48 h.

Quantum yield of photosystem II
The effective quantum yield of photosystem II (F PSII ) was measured with a chlorophyll fluorometer (Mini-PAM, Walz Gmbh, Effeltrich, Germany), from 10:00 am through 2:00 pm. Steady-state (F s ) and maximum (F' m ) yields were obtained with a leaf clip holder under irradiance of approximately 1,000 μmol photon m -2 s -1 PPFD. F PSII was calculated according to a previous study [22], in which Because of differences in the meteorological environments, especially in solar radiation and air humidity between the experiments, the absolute values of the quantum yield were variable between the first and second experiments (S1 Appendix). In the second experiment, the solar radiation was lower and air humidity was higher than in the first experiment. Therefore, relative quantum yields (RQYs: the quantum yields in the stressed condition divided by those in the control) were used to evaluate salt tolerance.

Cluster analysis
Euclidean distances computed from the average relative quantum yield of photosystem II, under the 50 mM NaCl and 200 mM NaCl conditions, were subjected to hierarchical clustering to identify the patterns of salt response in the 69 accessions. Cluster analyses were performed by Ward's method [23]. To compare all the evaluated traits between the different groups revealed by cluster analysis, analysis of variance (ANOVA) was performed with Tukey's multiple range test, using the groups and salt treatments as the factors.

Na + and K + concentrations
Leaf Na + and K + concentrations were determined using a method from a previous study [24] with some modifications. Three leaf pieces of 7 mm diameter were collected from each plant in a 2 ml tube. Following dry weight determination, samples were ground and suspended in 1 ml of 1 M ammonium acetate and shaken for 1 h. Homogenates were then centrifuged for 5 min at 7,000 ×g. The supernatant was measured for Na + and K + concentrations with a LAQUAtwin Na + meter (Horiba, Kyoto, Japan) and LAQUAtwin K + meter (Horiba, Kyoto, Japan), respectively, calibrated with standard solutions. The concentrations were then calculated to represent mmol g -1 dry weight.

Phylogenetic analysis
Phylogenetic trees of all 69 accessions were constructed with the sequence data of the internal transcribed spacer (ITS) region of nuclear ribosomal DNA with Phaseolus vulgaris (plant accession number: JP219310, sequence DDBJ number: LC082303) as an out-group. For all the accessions, sequence data [25] were downloaded from the DNA data bank of Japan (DDBJ: www.ddbj.nig.ac.jp, see Table 1 for accession numbers). Phylogenetic analysis was performed by the Neighbor-Joining method with 1,000 bootstraps using MEGA 6.0 software [26]. From the sequence data of all accessions, positions containing insertion/deletion and missing data were eliminated, and a total of 463 bases were used as the final dataset. Evolutionary divergence time was calculated, based on the estimated divergence time between Vigna and Phaseolus, as 8 million years ago [27].

Relative quantum yields
We evaluated salt tolerance of the 69 accessions using RQY (see Materials and Methods); because the quantum yield of photosystem II is commonly used as an indicator of plant response to environmental stresses [28]. The following cluster analysis clearly classified the accessions into four groups (Fig 1). Photos of the representative accessions of each group are shown in Fig 2. The RQY of each accession in each experiment is shown in S1 Appendix.
Group I included the 14 accessions that showed the highest RQYs in salt-stressed conditions (Fig 1) and no visual salt damage, even in 200 mM NaCl (Fig 2, S1 Fig). The average RQYs of the accessions were higher than 0.9 in both 50 mM and 200 mM NaCl (Table 2). In this group, RQYs in 200 mM NaCl were repeatedly high in the wild accessions, except V. vexillata var. ovata (ID55) (Fig 3A). In contrast, the RQYs in 200 mM NaCl were not consistent between the experimental replicates in all three domesticated accessions including tuber cowpea (V. vexillata var. macrosperma: ID56), cowpea (V. unguiculata: ID60), and yardlong bean (V. unguiculata subsp. sesquipedalis: ID61) (Fig 3A). Particularly in the first experiment, the RQYs declined to 0.3-0.4 in these three domesticated accessions. We ranked this group as "the most tolerant. " Group II included 13 accessions, where the average RQY was 0.80 in 50 mM NaCl and 0.47 in 200 mM NaCl (Table 2). Leaves became slightly yellowish in 50 mM NaCl and partially wilted, except V. vexillata (ID51 and ID53) (Fig 2, S1 Fig). This group included four domesticated accessions such as black gram (V. mungo; ID29), cowpea (ID62 and ID63), and bambara groundnut (V. subterranea; ID69). We ranked this group as "tolerant. " Group III included 21 accessions, where the average RQY (0.84) and visual leaf damages were similar to those of Group II in 50 mM NaCl (Table 2, Fig 2, S1 Fig). In 200 mM NaCl, however, the RQY severely decreased (0.08 in average) and many leaves were wilted (Table 2, Fig 2, S1 Fig). We ranked this group as "moderately susceptible. " Group IV included the remaining 21 accessions, where the average RQY decreased to 0.55 in 50 mM NaCl ( Table 2). The leaves of these accessions became yellowish and partially wilted in 50 mM NaCl and completely wilted under 200 mM NaCl (Fig 2). We ranked this group as "susceptible. "

Shoot biomass
After two weeks of salt treatment, we collected plant shoots and measured the dry weight. Since the absolute shoot biomass in the control condition largely differed between the accessions (Fig 4C), the effects of salt stresses were evaluated according to their relative values ( Fig  4A and 4B). As expected, the effect of salt stress on RSB was much more severe in 200 mM NaCl than in 50 mM. Interestingly, many accessions with large shoot biomass were categorized in Groups I and II (Fig 4C).
Although the overall variation in RSB was smaller compared to RQY, the average RSB of Group I was significantly higher than other groups in both 50 mM and 200 mM NaCl (Table 2). In 50 mM NaCl, many accessions in this group showed RSBs of almost 0.9 and the average was 0.83 ( Fig 4A, Table 2). In this salt condition, V. riukiuensis (ID18) grew better than  Table 2. Averages of relative quantum yield, relative shoot biomass, leaf Na + concentration, and K + /Na + ratio for each group.

Relative shoot biomass
Leaf Na + concentration (mmol g -1 DW) in the control (Fig 4A). In 200 mM NaCl, however, the variation was larger and ranged from 0. 27 Fig 3B). In Groups II-IV, the average RSBs were not significantly different from each other ( Table 2). However, the RSBs of some accessions in Groups II and III were comparable to those of Group I accessions. In particular, V. subterranea (ID69), a domesticated accession in Group II, showed the third highest RSB, even exceeding most of the accessions in Group I ( Fig  4B). In 50 mM NaCl, accessions of V. indica (ID44) and V. aridicola (ID49) in Group III showed particularly high RSBs (Fig 4A).

Leaf Na + concentration
Plant salt tolerance is considered to involve either avoidance (excluder type) or tolerance (includer type) mechanisms [15]. The "excluder type" plants keeps Na + ions out of internal tissues, whereas "includer type" plants isolate Na + ions in vacuoles [15]. To determine whether the accessions were excluders or includers, we measured Na + concentrations in the topmost fully expanded leaves. In the control condition, leaf Na + concentrations were less than 0.5 mmol g -1 in all accessions ( Fig 5A) and were not significantly different between the groups ( Table 2).
In 50 mM NaCl, Group I accessions accumulated lower amounts of Na + in their leaves, not exceeding 0.8 mmol g -1 DW (Fig 5B). In Group II, similar to Group I, most of the accessions showed Na + concentrations < 1.0 mmol g -1 DW (Fig 5B). However, some accessions (V. mungo: ID29 and V. stipulacea: ID36) exceeded 1.0 mmol g -1 DW (S1 Appendix). In Groups   III and IV, Na + concentration were highly varied, ranging from 0.1 to 2.0 mmol g -1 DW (S1 Appendix), but the averages were significantly higher than Group I averages (Table 2).

Potassium/sodium ratio
We also measured leaf K + concentration to calculate the K + /Na + ratio, which is often used as an important indicator of salt tolerance [29].
In the control condition, K + /Na + ratios varied little between both the accessions and the groups (Fig 6A). In 50 mM NaCl, K + /Na + ratios were reduced to < 0.6 in many accessions ( Fig  6B). Most accessions with higher K + /Na + ratios belonged to the Groups I or II, but two accessions (V. nepalensis: ID5 and V. hirtella: ID14) belonged to Group IV (S1 Appendix). In 200 mM NaCl, no accessions showed K + /Na + values > 0.6 ( Fig 6C).

Geographical distribution of accessions
Of the 69 accessions, we were able to locate collection sites of 52 (Fig 7). Accessions of Groups I and II were mainly collected in coastal or near-coastal areas such as beaches (Fig 8A, 8B and 8D), and coastal cliffs (Fig 8C). Exceptions to this were V. vexillata var. vexillata (ID51), V. unguiculata (ID60) and V. luteola (ID65) which originated from inland areas of Africa. Accessions of Group III were mainly found in inland areas, although some of them were found in coastal areas. Accessions of Group IV were all found in inland areas. Accessions of Groups I-III were widely distributed across Asia, Africa, Oceania, and South America. In contrast, accessions of group IV originated only in Asia.

Divergence time
To reveal the evolutionary origins of salt tolerance in Group I accessions, we constructed a phylogenetic tree from rDNA-ITS sequences and estimated divergence time of accessions [27]. The phylogenetic relationship agreed with the relationship described in Takahashi et al. [25]. As shown in Fig 9, we color-coded each accession to indicate the group of salt tolerance (see Fig 1). The phylogenetic tree clearly shows that the African accessions (subgenus Vigna and Plectrotropis) were often in the tolerant groups (I or II) whereas Asian accessions were mostly in the susceptible groups (III or IV).
More importantly, however, the accessions of Group I were present in all taxonomic sections, except section Ceratotropis. In addition, the tolerant accessions (Groups I or II) often neighbored susceptible accessions (Groups III or IV). One typical example was the case of V. nakashimae (ID17), V. riukiuensis (ID18) and V. minima (ID19), where the former two (tolerant) had diverged from the latter (susceptible) about 120,000 years ago. Another example was the case of V. trilobata (ID46 and ID47), where the tolerant accession diverged from the Salt Tolerance in Genus Vigna susceptible accession only approximately 60,000 years ago (Fig 9). V. vexillata (ID54) also diverged from the common ancestor of other accessions (ID51-53) about 60,000 years ago.

Discussion
In this study, we obtained phenotypic information of several wild Vigna accessions regarding salt tolerance. With the information gathered, we were able to identify several salt-tolerant accessions and elucidate the evolutionary origins of salt tolerance.

Salt tolerant accessions in the genus Vigna
Here we selected six accessions as the most salt-tolerant accessions. The selected accessions were V. riukiuensis (ID18), V. trilobata (ID47), V. vexillata (ID54), V. marina subsp. oblonga (ID66), V. luteola (ID67), and V. marina (ID68), because they showed higher RQYs and RSBs in both the first and second experiments (Fig 3). Unlike in our previous study [12], we discarded V. nakashimae (ID17), which we had previously selected as one of the most tolerant accessions in Asian Vigna [12]. In the last study, we screened only for visual salt damage whereas, in this study, we considered the RSB. Here V. nakashimae (ID17) was consistently high in the RQY (Fig 3A), which strongly correlated with visual damage (Fig 2, S1 Fig), but not in the RSB (Figs 3 and 4).
We consider that these accessions could serve as promising sources of salt tolerance for breeding purposes, because, at least some of them are cross compatible with domesticated species. Cross compatible examples are; V. riukiuensis (ID 18) × azuki bean (V. angularis var. angularis) and V. vexillata (ID 54) × tuber cowpea (V. vexillata var. macrosperma) [30]. Even wider hybridization is also possible if assisted with embryo rescue [31].
In addition, these accessions could also serve as genetic materials to clone genes involved in salt tolerance. V. trilobata is the best example of this, as it has an intraspecific variation in salt tolerance (Figs 1 and 9). If we cross ID47 and ID46 of this species, we will be able to develop a population with minimal genetic divergence, except in salt tolerance. In another of our previous studies [13], we demonstrated the strength of this strategy using a mapping population derived from V. marina subsp. oblonga (ID66) and V. luteola (ID65). Although this study ranked V. luteola (ID65) in Group I (Fig 1), it does not survive in 350 mM NaCl while V. marina subsp. oblonga (ID66) does not show any symptoms of wilting [13]. Using this approach, we detected a strong QTL that could solely explain 50% of phenotypic variance between the two [13].
Moreover, since the selected accessions are already well adapted to salt-affected environments, it would be interesting to de novo domesticate them into new crops, as we recently proposed [32].

Mechanisms of salt tolerance
We observed two interesting features in the accessions of Group I. First, in 50 mM NaCl, all the accessions accumulated lower amounts of Na + than others (Fig 5B). Second, in 200 mM NaCl, they varied so greatly that the range of leaf Na + concentrations encompassed the lowest to the highest among the 69 accessions (Fig 5C). The first feature indicated that Group I accessions had better salt excluding ability than accessions in other groups. However, the second feature indicated they acquired different strategies to tolerate the severer salt stress. That is, the non-salt-accumulating accessions might be specialized to exclude salt, whereas the salt- accumulating accessions might have added salt-including abilities to isolate excessive Na + ions into vacuoles [29].
This argument agreed with our previous study, where we found V. nakashimae (ID17) and V. riukiuensis (ID18) to have different mechanisms of salt tolerance [12]. In that study, we showed the former prevented Na + uptake into leaves, whereas, the latter accumulated high volumes of Na + into leaves. In this study we found the same trend in ID17 and ID18 (1.1 mmol g -1 DW vs. 2.8 mmol g -1 DW) (S1 Appendix), further confirming the acquisition of different mechanisms of salt tolerance.
Given some of the includer-type accessions and excluder-type accessions are cross-compatible with each other, it may be possible to obtain a plant that is proficient in both abilities. It would be interesting to test whether such a plant could be even more tolerant against salt stress.
On the other hand, though K + /Na + ratio is often used as an important indicator of salt tolerance [29], we could not detect a sign of its contribution to salt tolerance in the genus Vigna. Several accessions showed higher K + /Na + ratios in 50 mM NaCl, but only V. nakashimae (ID 17) and V. vexillata (ID 53) retained high RQY (S1 Appendix). In addition, accessions with low leaf Na + concentrations did not always maintain K + /Na + ratios. In 200 mM NaCl, no accessions maintained K + /Na + ratios ( Fig 6). As such, we concluded that K + /Na + ratio is not a useful indicator for salt tolerance in the genus Vigna. This has also been reported as true for several subspecies of tetraploid wheat [33].

Evolution of salt tolerance
According to our evaluation of salt tolerance and phylogenetic analysis, we argued that the evolution of salt tolerance, in the genus Vigna, has independently occurred several times. Although the selected accessions were all wild accessions adapted to coastal environments (Figs 7 and 8), the phylogenetic tree indicated multiple evolutionary origins of salt tolerance. V. riukiuensis (ID18) belonging to the section Angulares, where most accessions were grouped into Groups III or IV. It had probably acquired salt tolerance after divergence from the ancestor of V. minima (ID19). Likewise, V. trilobata (ID47) had acquired salt tolerance after divergence from the ancestor of V. aridicola (ID41-43). For V. vexillata (ID54), salt tolerance originated after diverging from the susceptible accessions (ID51-53). The origin of salt tolerance in the accessions of V. marina and V. luteola was probably after the divergence from the section Catiang. As such, extreme salt tolerance has independently evolved at least four times in the genus Vigna. In addition, the phylogenetic positions of other Group I accessions suggested that salt tolerance could have evolved even more frequently. A good example is, again, the case of V. nakashimae (ID17) and V. riukiuensis (ID18). Since these two accessions are, as mentioned in the above section, different in their mechanisms of salt tolerance [12], the evolution of salt tolerance should have occurred after divergence of the two species (60,000 years ago) (Fig 9). Similarly, the origin of salt tolerance in V. vexillata (ID55-57) might be independent from that in ID54 because they formed a different clade (Fig 9). If we take Group II accessions into consideration, we could even postulate that salt tolerance could "often" evolve in the genus Vigna.
Moreover, we should note that many tolerant accessions diverged from the susceptible accessions only 50,000-200,000 years ago (Fig 9), indicating the salt tolerance can evolve within a relatively short time. As such, we expect salt tolerance in the genus Vigna might have occurred via simple genetic changes with large effects. This hypothesis is again supported by our previously study, where we found a strong QTL of salt tolerance from V. marina subsp. oblonga (ID66) [13].

Salt tolerance in Vigna crops
Our results showed that some Vigna crops were tolerant to salt stress, such as cowpea, bambara groundnut, tuber cowpea, and black gram. Of these crops, all were African Vigna, except black gram. Most of the Asian Vigna crops (azuki bean, rice bean, mung bean, and moth bean) were susceptible. The tolerant crops might be directly applied to cultivation in moderately saline soils.
The results agreed with the data shown in the Annex 1 (crop salt tolerance data) in FAO Corporate Document Repository, except black gram (V. mungo; ID29). In the Annex 1 Table, cowpea was classified as "Moderately Tolerant" while mung bean and black gram were categorized as "Susceptible, " based on shoot biomass and seed yield under long-term salt stress [34].
Unfortunately, we tested only one accession of wild cowpea (V. unguiculata var. dekindtiana; ID64), which was not more tolerant to salt stress than the domesticated cowpea. However, we expect there will be wild cowpea accessions with higher salt tolerance, given the broad intraspecific variation in cowpea varieties [35]. With such materials, it would be possible to further improve salt tolerance in domesticated cowpea.
In the section Angulares, none of the domesticated accessions and few of the wild accessions were tolerant to salt stress. As such, to improve the salt tolerance of azuki bean and rice bean, we might have no choice but to cross them with V. nakashimae (ID17) or V. riukiuensis (ID 18).

Salt tolerance vs. absolute shoot biomass
Since trade-offs between stress tolerance and growth rate have been reported [36], we expected salt tolerance would be found only in small accessions. This was true in our previous study, where we screened Asian Vigna accessions for salt tolerance and found only small accessions were tolerant against salt stress [12].
However, the results in this study suggested the salt tolerance might not severely affect growth. Firstly, Group I included accessions with the largest shoot biomass (ID60 and ID61) (Figs 1 and 4). This group also included accessions with relatively larger biomasses, such as V. vexillata var. macrosperma (ID56), V. luteola (ID 65 and ID67) and V. marina subsp. oblonga (ID66). Although the highest RSB was observed in one of the smallest accession (V. riukiuensis (ID18)), much larger accessions (V. marina (ID68) and V. subterranea (ID69)) showed almost the same RSBs in 200 mM NaCl (Fig 4). Moreover, the absolute biomass did not negatively correlate with salt tolerance, even between accessions with closer genetic distances, for example, within V. mungo (ID29 vs. ID30) or within V. unguiculata (ID60 and ID61 vs. ID62 and ID63). As such, the cost for salt tolerance may not be as destructive to plant growth as we had expected, or might even have little effect. We now consider that it would be possible to improve salt tolerance of crops without severely reducing growth rate.
However, we should be aware that we treated the plants with salt stress only for two weeks. This was because we had to evaluate the plants before the susceptible accessions had died. However, currently we have no information about grain yield of the tolerant accessions under salt stress. As such, it would be very important to evaluate the accessions in Groups I and II over a longer time period, until seed setting.

Conclusions
In this study, we evaluated several aspects of salt tolerance in various accessions of the genus Vigna and selected six accessions as the most tolerant accessions. These accessions were all wild accessions adapted to coastal habitats but were phylogenetically distinct from each other. This result indicated that extreme salt tolerance evolved at least four times during speciation in the genus Vigna. They showed diverse characteristics, especially in leaf Na + concentration, indicating the acquisition of different mechanisms of salt tolerance. In addition, because of the large differences in salt tolerance between the genetically close accessions, simple genetic changes might be responsible for salt tolerance evolution. Since most of the tolerant accessions are crossable to susceptible accessions, they can be donors of salt tolerance for breeding, or would be materials for isolating tolerance genes. It would also be interesting to de novo domesticate the wild salt tolerant accessions. Thus, from the above results, we demonstrated that the genus Vigna is a valuable genetic resource of salt tolerance.
Supporting Information S1 Appendix. Characteristics of shoot biomass, quantum yield of PSII, leaf Na + concentration, and K + /Na + ratio of the 69 accessions used in this study. (XLSX) S1 Fig. Images of the 69 accessions at 2