Recent advances in genetics of salt tolerance in tomato

Transcription

1 Plant Cell, Tissue and Organ Culture 76: , Kluwer Academic Publishers. Printed in the Netherlands. 101 Review of Plant Biotechnology and Applied Genetics Recent advances in genetics of salt tolerance in tomato M. R. Foolad Department of Horticulture, The Pennsylvania State University, University Park, PA 16802, USA ( requests for offprints; Fax: (814) ; mrf5@psu.edu) Received 27 September 2002; accepted in revised form 23 June 2003 Key words: breeding, gene mapping, genetic engineering, genetic transformation, quantitative trait loci (QTLs), salt stress, salt tolerance, stress tolerance, transgenic plants Abstract Salinity is an important environmental constraint to crop productivity in arid and semi-arid regions of the world. Most crop plants, including tomato, Lycopersicon esculentum Mill., are sensitive to salinity throughout the ontogeny of the plant. Despite considerable research on salinity in plants, there are only a few instances where salttolerant cultivars have been developed. This is due in part to the complexity of the trait. A plant s response to salt stress is modulated by many physiological and agronomical characteristics, which may be controlled by the actions of several to many genes whose expressions are influenced by various environmental factors. In addition, salinity tolerance is a developmentally regulated, stage-specific phenomenon; tolerance at one stage of plant development is often not correlated with tolerance at other stages. Specific ontogenic stages should be evaluated separately for the assessment of tolerance and the identification, characterization, and utilization of useful genetic components. In tomato, genetic resources for salt tolerance have been identified largely within the related wild species, and considerable efforts have been made to characterize the genetic controls of tolerance at various developmental stages. For example, the inheritance of several tolerance-related traits has been determined and quantitative trait loci (QTLs) associated with tolerance at individual developmental stages have been identified and characterized. It has been determined that at each stage salt tolerance is largely controlled by a few QTLs with major effects and several QTLs with smaller effects. Different QTLs have been identified at different developmental stages, suggesting the absence of genetic relationships among stages in tolerance to salinity. Furthermore, it has been determined that in addition to QTLs which are population specific, several QTLs for salt tolerance are conserved across populations and species. Research is currently underway to develop tomatoes with improved salt tolerance throughout the ontogeny of the plant by pyramiding QTLs through marker-assisted selection (MAS). Transgenic approaches also have been employed to gain a better understanding of the genetics of salt tolerance and to develop tomatoes with improved tolerance. For example, transgenic tomatoes with overexpression of a single-gene-controlled vacuolar Na + /H + antiport protein, transferred from Arabidopsis thaliana, have exhibited a high level of salt tolerance under greenhouse conditions. Although transgenic plants are yet to be examined for field salt tolerance and salt-tolerant tomatoes are yet to be developed by MAS, the recent genetic advances suggest a good prospect for developing commercial cultivars of tomato with enhanced salt tolerance in near future. Abbreviations: BC backcross; DW dry weight; FW fresh weight; G E genotype environment; h 2 heritability; PI plant introduction; QTLs quantitative trait loci; RGA resistance gene analog; RILs recombinant inbred lines; T50 time to 50% germination; TI tolerance index; TI G germination tolerance index; TI VS vegetative stage tolerance index

2 102 Introduction Salinity and crop production Salinity is an increasingly important environmental constraint to crop production worldwide (Ghassemi et al, 1995). Regardless of the cause (ion toxicity, water deficit, and/or nutritional imbalance), high salinity in the root zone severely impedes normal plant growth and development, resulting in reduced crop productivity or crop failure. A saline soil is generally defined as one in which the electrical conductivity (EC) of the saturation extract in the root zone exceeds 4 dsm 1 at 25 C and has an exchangeable sodium percentage of 15. Of the total 14 billion ha of land available on earth 6.5 billion ha are estimated to be arid and semi-arid, of which about 1 billion are saline soils. Furthermore, it is estimated that about 20% of cultivated lands and 33% of irrigated agricultural lands worldwide are afflicted by high salinity (Ghassemi et al., 1995; Szabolcs, 1992). Moreover, the salinized areas are increasing at a rate of 10% annually; low precipitation, high surface evaporation, weathering of native rocks, irrigation with saline water, and poor cultural practices are among the major contributors to the increasing soil salinity. Two major approaches have been proposed and employed to minimize the deleterious effects of high soil/water salinity in agriculture (Epstein et al., 1980). First, a technological approach of implementing large engineering schemes for reclamation, drainage and irrigation with high-quality water. Although this approach has been effective in some areas, the associated costs are high and it often provides only a temporary solution to the problem. The second approach, which must be implemented simultaneously along with the first to achieve sustainable crop production in the presence of excessive salts, entails biological strategies focused upon the exploitation or development of plants capable of tolerating high levels of salts. This approach includes diversifying cropping systems to include crops that are known to be salt tolerant (e.g., by crop substitution), exploiting wild or feral species that are adapted to saline environments (e.g., by domestication), and genetically modifying domesticated crops by breeding and selection for improved salt tolerance (ST). Breeding for salt-tolerant genotypes that can grow more efficiently than the conventional varieties under saline conditions, however, depends on a profound understanding of the physiology, genetics, and molecular biology of plant ST. While the general effect of salinity is a reduction in plant growth in almost all plant species, ST is commonly defined as the inherent ability of the plant to withstand the effects of salinity with insignificant adverse effects on plant productivity. Considerable research has been conducted and substantial information is obtained regarding the physiological, metabolic and cellular aspects of plant ST. These aspects, however, are not the subject of this paper, as they have been reviewed extensively by other investigators (Greenway and Munns, 1980; Levitt, 1980b; Cheeseman, 1988; Munns, 1993; Jacoby, 1994; Zhu et al., 1997; Flowers, 1999; Bohnert and Shen, 1999; Storey and Walker, 1999; Hasegawa et al., 2000; Zhu, 2001). In contrast, efforts to characterize the genetic controls of ST at the whole plant, cellular, or molecular level are more recent and far from complete (Blum, 1988; Grover et al., 1999; Jaiwal et al., 1997; Foolad, 1997; Jain and Selvaraj, 1997; Ashraf, 1994; Shannon, 1997; Winicov, 1998; Borsani et al., 2001). Accumulating evidence, however, suggests that plants respond to salt stress (SS) by displaying complex, quantitative traits, which involve the functions of many genes and physiological mechanisms whose expression are influenced by numerous environmental factors (Levitt, 1980a; Blum, 1988; Chaubey and Senadhira, 1994; Richards, 1996). The quantification of plant s ST poses serious difficulties. Direct selection under field conditions is not always possible, because uncontrollable environmental factors adversely affect the precision and repeatability of such trials (Richards, 1996). There is no reliable field screening technique that can be used in subsequent generations of selection and breeding. Furthermore, ST appears to be a developmentally regulated, stage-specific phenomenon; tolerance at one stage of plant development is often not correlated with tolerance at other stages (Greenway and Munns, 1980; Maas, 1986; Jones and Qualset, 1984; Shannon, 1985; Johnson et al., 1992; Foolad and Lin, 1997a; Pearen et al., 1997; Foolad, 1999). Specific ontogenetic stages, including seed germination and emergence, seedling survival and growth, and vegetative growth and reproduction, may have to be evaluated separately for the assessment of tolerance and the identification, characterization and utilization of useful genetic components. Partitioning of the tol-

3 103 erance into its component traits related to ontogenic stages would facilitate a better understanding of the genetic basis of tolerance and the development of salttolerant genotypes. This approach has recently been employed in tomato to characterize the genetic controls of ST at different plant developmental stages, as discussed in this review. Recent advances in molecular genetic techniques, including genetic transformation, marker mapping and quantitative trait loci (QTLs) analysis, have contributed significantly to a better understanding of the genetic, physiological and biochemical bases of plant ST and have facilitated the development of plants with enhanced ST. For example, significant progress has been made in the identification of genes, enzymes or compounds with significant effects on plant ST at the cell or organismal level (Shen et al., 1997; Winicov, 1998; Apse et al., 1999; Bohnert and Shen, 1999; Allakhverdiev et al., 1999; Grover et al., 1999). Manipulation of the expression or production of such genes, enzymes, or compounds through transgenic approaches have resulted in the development of plants with enhanced ST in different plant species (Thomas et al., 1995; Xu et al., 1996; Lilius et al., 1996; Apse et al., 1999; Zhang and Blumwald, 2001; Serrano et al., 1999). Molecular markers technology also has allowed the identification, characterization, and comparison of QTLs with significant effects on plant ST during different stages of plant development (Mano and Takeda, 1997; Ellis et al., 1997; Forster et al., 1997; Foolad et al., 1998; Foolad, 1999; Foolad and Chen, 1999; Foolad et al., 2001). This review focuses solely on the recent advances in genetics of ST in tomato, a model plant, with emphasis on the potential and limitation of different approaches for developing tomatoes with field tolerance to salinity. Tomato and salinity The cultivated tomato, Lycopersicon esculentum Mill., a fruit that is almost universally treated as a vegetable and a perennial plant that is almost universally cultivated as an annual, is a widely-grown crop plant and the focus of a large agricultural industry. Although a tropical plant, originating and domesticating in South and Central America (Rick, 1978), tomato is grown in almost every corner of the world, from the tropics to within a few degrees of the Arctic Circle. In spite of its global distribution, a major portion of the world tomato production is concentrated in a rather limited number of warm and dry areas, in particular regions around the Mediterranean Sea (including the Middle East), southern and western parts of the US and Mexico. Although such areas generally provide optimal climates for tomato production, a high level of salinity frequently encountered in the soil or in the irrigation water poses serious constraints to tomato production. For example, of the 4 million hectares of irrigated cropland in California, a major tomato producing state in the US, 1.2 are affected by salinity (Backlund and Hoppes, 1984) and the salinity problems are increasing. Therefore, to maintain the current level of productivity, or to increase the total production by using marginal saline lands, tomato cultivars with enhanced ST are needed. Most commercial cultivars of tomato are moderately sensitive to salinity at all stages of plant development, including seed germination, vegetative growth and reproduction, and, as a result, their economic yield is substantially reduced under SS (Jones et al., 1988; Maas, 1986; Bolarin et al., 1993). Although there is limited variation for ST within the cultivated species, there are several wild species within Lycopersicon that represent a potential source of useful genes for ST breeding (Rick, 1979). Attempts to identify gene resources for ST in tomato were first made by Lyon (1941) who proposed the improvement of tomato ST by introgression of genes from L. pimpinellifolium (Jusl.) Mill., a closely-related wild species. Later investigations resulted in the identification of salt-tolerant accessions within tomato wild species L. pimpinellifolium, L. peruvianum (L.) Mill., L. cheesmanii Riley, L. hirsutum Humb. and Bonpl., and L. pennellii (Corr.) D Arcy (Tal, 1971; Tal and Shannon, 1983; Phills et al., 1979; Rush and Epstein, 1976; Jones, 1986b; Sarg et al., 1993; Foolad and Lin, 1997b). During the past few decades, significant progress has been made in genetic characterization of ST in tomato. A significant dicovery has been that ST in tomato is a developmentally regulated, stage-specific phenomenon; tolerance at one stage of plant development is often not correlated with tolerance at other developmental stages (Jones and Qualset, 1984; Shannon et al., 1987; Asins et al., 1993; Foolad and Lin, 1997a; Foolad, 1999). This is similar to findings in other plant species (Ashraf and McNeilly, 1988; Johnson et al., 1992). Furthermore, it has been demonstrated that tomato ST is generally increased with plant age (Bolarin et al., 1993), similar to responses reported for barley (Hordeum spp.), corn (Zea mays L.), rice (Oryza sativa L.), and wheat (Triticum spp.) (Maas,

4 ). Thus, it was proposed that salinity tolerance at each developmental stage in tomato be evaluated and genetically characterized separately (Jones and Qualset, 1984). Here, the recent findings on genetics of ST in tomato are reviewed and the potential for improving tomato ST using different genetic approaches is discussed. Genetics of salt tolerance during seed germination and seedling emergence Background information and inheritance. The commercial cultivars of tomato are most vulnerable to salinity at the seed germination and early seedling growth stages (Cook, 1979; Jones, 1986a; Maas, 1986; Foolad and Jones, 1991). At these stages, tomato exhibits sensitivity even to low concentrations of salts ( 75 mm NaCl) (Jones, 1986b; Foolad and Lin, 1997b; Cuartero and Fernandez-Munoz, 1999). Surface soils, however, may have salinities several fold that of the subsoil, presenting a serious problem at the germination stage. High salinity may cause a delay in the onset, a reduction in the rate and the final percentage germination, and an increase in the dispersion of germination events. This sensitivity has important biological and applied significance. The costly operations of greenhouse seedling production and transplantation into the field have encouraged many tomato producers to grow direct-seeded crops. Furthermore, the dependence on mechanization in modern cultivation systems and the use of costly hybrid seed requires rapid, uniform and complete germination. Improving the uniformity and rapidity of seed germination responses under saline conditions would contribute significantly to the efficiency of stand establishment in tomato. Genetic resources for ST during seed germination in tomato have been identified within the cultivated and related wild species, including L. pennellii, L. pimpinellifolium, andl. peruvianum (Jones, 1986b; Foolad and Lin, 1997b; Cuartero and Fernandez- Munoz, 1999). Salt-tolerant accessions have been used to investigate the genetic basis of tolerance during seed germination. For example, a generation means analysis, using the parental lines and reciprocal filial and backcross populations (total of 10 generations) of a cross between a salt-sensitive tomato line (UCT5) and a salt-tolerant L. esculentum accession (PI174263), determined that the ability of tomato seed to germinate rapidly under SS was genetically controlled with a narrow-sense heritability (h 2 ) of 0.75 (Foolad and Jones, 1991). A parent-offspring regression analyses, using F 2 :F 3 and F 3 :F 4 progeny of the same cross, provided similar estimates of h 2 for this trait (Foolad and Jones, 1992). Both studies indicated that tomato ST during seed germination was genetically controlled with additivity being the major genetic component and suggested that this trait could be improved by directional phenotypic selection. However, the effectiveness of phenotypic selection to improve tomato ST during seed germination was subsequently examined and verified in an analysis of response to selection in F 2,F 3 and F 4 progeny populations of the same UCT5 PI cross (Foolad, 1996b). Together these studies indicated that ST during seed germination in tomato was controlled by genes with additive effects and could be improved by directional phenotypic selection. However, the number of controlling genes or magnitudes of their effects could not be determined in these studies. QTL mapping. A better approach to understanding the genetic basis of, and to improving the efficiency of selection for, complex traits is to discover genetic markers that are associated with genes or QTLs controlling the traits. Molecular markers technology can facilitate determination of the number, chromosomal location, and individual and interactive effects of QTLs that control a trait. Following their identification, useful QTLs can be introgressed into desirable genetic backgrounds by a process known as markerassisted selection (MAS). Furthermore, because several traits and mechanisms are usually involved in plant ST, pyramiding traits of interest via MAS may be an effective approach to substantial improvement in plant ST. Moreover, such approach may facilitate pyramiding of tolerance components from different resources. Recently, several QTL mapping studies determined the number, genomic location, and individual effects of QTLs affecting ST during seed germination in tomato. For example, using an F 2 population (N = 2500) of a cross between a salt-sensitive tomato line (UCT5) and a salt-tolerant L. pennellii accession (LA716) and a selective genotyping approach (Darvasi and Soller, 1992) [also known as trait-based analysis (Lebowitz et al., 1987) or distributional extreme analysis (Lander and Botstein, 1989)], Foolad and Jones (1993) identified five QTLs on chromosomes 1, 3, 7, 8, and 12 with significant effects on this trait. The validity of these QTLs was subsequently examined in several studies using different interspecific populations of tomato derived from L. esculentum L. pennellii and L. esculentum L. pimpinellifolium

5 105 crosses (Foolad et al., 1997; Foolad and Chen, 1998; Foolad et al., 1998) (M.R. Foolad, unpubl. data). The latter studies verified most of the QTLs identified in the original study, and further detected a few new QTLs on chromosomes 2 and 9. The overall results supported the suggestion that ST during seed germination in tomato was a quantitative trait, controlled by more than one gene. However, all of the studies demonstrated that this trait was controlled by a few QTLs with major effects in addition to several QTLs with smaller effects. Furthermore, there were little or no epistatic interaction effects among the identified QTLs. These findings are promising and suggest a good prospect for improving tomato ST during seed germination via MAS, as discussed below. Comparisons of the QTLs identified for ST in different interspecific populations of tomato, including those derived from L. esculentum L. pennellii and L. esculentum L. pimpinellifolium crosses, indicated that some QTLs were conserved across species whereas others were species-specific (Foolad et al., 1997; Foolad and Chen, 1998; Foolad et al., 1998) (M.R. Foolad, unpubl. data). However, because in the life of most breeding projects often more than one gene resource is utilized, it should be possible to incorporate and pyramid QTLs from different resources via MAS. Comparisons of the QTLs identified in different populations of the same cross, for example those detected in BC 1 S 1 and RIL populations of a L. esculentum L. pimpinellifolium cross, indicated that most QTLs were stable across populations/generations, suggesting their utility for MAS. Furthermore, results from the various studies indicate that, in comparison with ST at later stages of plant development in tomato (discussed below), ST during seed germination is less complex and less influenced by environmental factors, and potentially can be improved by either phenotypic selection or MAS. However, because ST genes are often found within the wild species, a combination of phenotypic and marker-assisted selections might be the most effective approach. Tolerance to different levels of stress. In many saline soils the concentration of salts may vary across the soil horizon, ranging from low to moderate and high (Richards and Dennett, 1980). A successful cultivar for production under saline conditions is one which can tolerate a wide range of SS and which also performs well in the absence of stress. Selection and breeding for ST under a wider range of salinity stress, however, may not always be doable in a breeding program. Therefore, it is important to examine plants responses to different levels of SS and determine if there is any relationship among tolerances to different stress levels. Selection might be more effective or useful under specific stress levels. Several studies have examined such relationships during seed germination in tomato. For example, evaluation of 56 tomato genotypes at different salinity levels, including 75 mm (low), 150 mm (intermediate), and 200 mm (high) NaCl, indicated that genotypes that germinated rapidly at the low salt concentration also germinated generally rapidly at the moderate and high salt concentrations (Foolad and Lin, 1997b). A phenotypic correlation coefficient of r = 0.90 (p < 0.01) was observed between germination at 75 and 150 mm salts, suggesting the involvement of same genetic controls for germination under different SS levels. In a later study, this suggestion was examined and verified genetically by analyses of response and correlated response to selection for ST (Foolad, 1996b). Selections were made separately under 100 (low), 150 (intermediate), and 200 mm (high) salts and progeny responses were examined at all three SS levels. It was determined that selection for rapid seed germination at any of the three stress levels resulted in progeny with enhanced germination rate at all three stress levels, indicating the presence genetic relationships between ST at different stress levels (Foolad, 1996b). The results were also consistent with the findings that in F 2 populations of a cross between a salt-tolerant accession (LA716) of L. pennellii and a salt-sensitive tomato cultivar (UCT5) identical QTLs contributed to ST during seed germination at different SS levels (Foolad and Jones, 1993). The combined results suggest the involvement of the same genes with tolerance to salinity at different stress levels. Thus, to develop tomato cultivars with improved ST during seed germination at diverse SS levels, it is sufficient to conduct selections and breeding at a single SS level. However, because the rate of tomato seed germination at a moderate level of SS (e.g., 150 mm salt) exhibits high correlations with those both at a low (100 mm salt) and a high SS level (200 mm salt), preferably selections to improve ST should be made at an intermediate stress level (Foolad, 1996b). Physiological genetics of ST during seed germination. Although QTLs for ST during seed germination in tomato have been identified, their genetic nature or the physiological mechanisms which they modulate have

6 106 not been determined. Such resolution requires the isolation and functional analysis of QTLs. However, based on the current physiological knowledge of ST during seed germination, some speculations can been made as to the role of QTLs. ST during seed germination is a measure of the seed s ability to withstand the effects of high concentrations of salts in the medium. Excessive salt depresses the external water potential, making water less available to the seed. Slower seed germination under stress compared to nonstress conditions, however, could be due to osmotic and/or ionic effects of the saline medium. Physiological investigations to distinguish between these two types of effect have been very limited. However, accumulating evidence in different crop species suggests that low water potential of the external medium, rather than ion toxicity effects, is the major limiting factor to germination under SS (Kaufman, 1969; Ungar, 1978; Haigh and Barlow, 1987; Bliss et al., 1986; Bradford, 1995), although few reports have suggested otherwise (Choudhuri, 1968; Younis and Hatata, 1971; Redmann, 1974). In a recent investigation, germination responses of eight tomato genotypes, including salt-tolerant and salt-sensitive accessions of L. esculentum and L. pimpinellifolium, were evaluated in iso-osmotic (water potential 700 kpa or 15 dsm 1 ) media of NaCl, MgCl 2,KCl,CaCl 2, sorbitol, sucrose, or mannitol (J.R. Hyman and M.R. Foolad, unpubl. data). The results confirmed the previous suggestion that the delay in the germination rate of tomato seed under SS was mainly due to osmotic effects rather than ion toxicity. The overall results, however, point to the direction that ST during seed germination in tomato is an adaptation to low water-potential and that there are little or no detrimental ionic effects during germination. Anatomically, the tomato seed is comprised of a seed coat (testa) that encloses a curved filiform embryo and an endosperm that practically fills the lumen of the seed not occupied by the embryo (Esau, 1953). For germination to occur, the hydraulic extension force of the embryo must exceed the opposing force of the seed coat and the living endosperm tissues at the placental end of the seed (Hegarty, 1978; Liptay and Schopfer, 1983; Bradford, 1986; Groot and Karsen, 1987). Embryo genotype was thus suggested to play a major role in determining germination rate in tomato seed (Liptay and Schopfer, 1983). According to this hypothesis, differences in salt sensitivity of tomato seeds during germination reside either in the osmotic pressure or turgor pressure of the germinating embryo. However, osmotic stress can also negatively affect seed imbibition, and thus retard (or prevent) weakening of the restrictive forces of the endosperm and testa, resulting in reduced rate (or inhibition) of germination (Liptay and Schopfer, 1983; Groot and Karsen, 1987; Dahal et al., 1990). Based on this hypothesis, endosperm plays a major role in determining the rate of tomato seed germination, and its contribution has been genetically supported (Foolad and Jones, 1991). Thus, the rate of tomato seed germination may be influenced by the physical, chemical, and genetic compositions of the embryo, endosperm and/or testa. The identified QTLs for ST during seed germination in tomato could therefore affect germination rate through the vigor of the germinating embryo, the variation in the thickness of the endosperm, the physical and permeability properties of the endosperm cell walls, the time of onset or rate of activity of enzymes which modify the properties of the endosperm cell wall, the release of gibberellin by the embryo (which is necessary for endosperm weakening), the base water potential required for seed germination, the hydrotime constant (Bradford, 1995), the rate of metabolic activities in the embryo or endosperm under osmotic stress, osmoregulation during germination, or any other physiological or metabolic processes which are essential for the initiation of germination. However, isolation, characterization and comparison of genes which facilitate rapid seed germination under SS is necessary to determine the actual functions of the identified QTLs. Nonetheless, irrespective of the physiological mechanism(s) of ST during seed germination in tomato, the identified QTLs could potentially be useful for improving tomato ST during seed germination using MAS and breeding, as discussed below. Genetics of salt tolerance during vegetative growth and reproduction Background information. For tomato production under saline conditions, ST during vegetative stage is more important than ST during seed germination because most tomato crops are established by seedling transplantation rather than direct seeding. ST during vegetative stage may also be more important than ST during reproduction (i.e., flowering and fruit set) because tomato tolerance of salinity generally increases with plant age, and plants are usually most tolerant at maturation (Bolarin et al., 1993). During the flowering and fruiting stages, tomato plants can generally withstand salt concentrations that can be fatal at the seedling stage. Furthermore, there is a positive

7 107 correlation between tomato yield and plant size during vegetative growth under SS (Bolarin et al., 1993; Pasternak et al., 1979), indicating the importance of ST during the vegetative stage. At low concentrations of salt (EC = 3 5 dsm 1 ), tomato growth is mainly restricted by nutritional imbalances because nutrients become the limiting factor under such conditions (Cuartero and Fernandez- Munoz, 1999). At moderate to high levels of salt (EC 6dSm 1 ), in addition to nutrient imbalances, osmotic effects and ion toxicities contribute to reduced growth. Most commercial cultivars of tomato are moderately sensitive to SS during vegetative stage (Maas, 1986; Tal and Shannon, 1983; Foolad and Lin, 1997b). However, phenotypic variation for ST has been identified within the cultivated tomato (Cuartero et al., 1992; Sarg et al., 1993; Foolad, 1997) and wild species L. peruvianum (Tal and Givish, 1973), L. pennellii (Dehan and Tal, 1978; Saranga et al., 1991; Perez- Alfocea et al., 1994; Cano et al., 1998), L. cheesmanii (Rush and Epstein, 1976; Asins et al., 1993), and L. pimpinellifolium (Bolarin et al., 1991; Cuartero et al., 1992; Asins et al., 1993; Foolad and Chen, 1999). Although much of this phenotypic variation has not been genetically verified, it can be potentially useful for developing tomatoes with improved ST. Limited research has been conducted to determine sources of variation or the genetics of ST during reproduction in tomato. For example, efforts to determine pollen viability or stigma receptivity, or the ability of the plant to produce flowers and set fruit under SS have been limited. This may be in part due to the presence of a generally higher level of ST during reproduction than earlier stages. (Grunberg et al., 1995) reported that salinity did not affect tomato pollen viability, though it did reduce the number of pollen grains per flower. Adams and Ho (1992) reported that increasing salinity to 10 dsm 1 did not significantly affect fruit set in tomato, which was reduced only at 15 dsm 1. In a recent investigation, 13 tomato accessions from three species were grown under saline (300 mm NaCl + 30 mm CaCl 2 ; equivalent to 28 dsm 1 )and control (no salt) conditions and their pollen production and in vitro pollen germination were examined (S. Prakash and M.R. Foolad, unpubl. data). For most accessions, there was no significant reduction in pollen production (per flower) in response to SS. Furthermore, pollen from both salt-grown and control-grown plants was cultured at different salt concentrations, including 0, 0.2, 0.4 and 0.8% NaCl, and evaluated for percentage germination after 4 or 8 h of incubation. In all accessions, pollen germinability was decreased under salt compared to the control treatment, and the reduction was greater at higher (e.g., 0.8%) than lower (e.g., 0.2%) salt concentrations. However, in most accessions, in vitro pollen germinability of salt-grown plants was generally higher than the control-grown plants at high concentrations of salt (0.4 and 0.8% NaCl), suggesting that pollen ST in tomato was enhanced by growing plants under SS. The suggestion of a higher ST during reproduction in tomato, however, is in contrast with a report of the failure in producing fruit in several genotypes that grew adequately under saline conditions (Asins et al., 1993). The cultivated tomato is generally considered moderately sensitive to salinity, as measured by the reduction in fruit yield under SS compared to control conditions (Maas, 1990). Fruit yield of tomato starts decreasing when the EC of the saturated soil extract exceed 2.5 dsm 1 (Maas, 1990; Saranga et al., 1991), though there are reports of higher thresholds for yield reduction under SS in tomato (Adams, 1991). A 10% reduction in fruit yield is expected per additional dsm 1 beyond the threshold level (Saranga et al., 1991). The major cause of yield reduction in tomato under low to moderate levels of salinity (EC = 3 9 dsm 1 ) is the reduction in the average fruit size, and not a reduction in fruit number (van-ieperen, 1996). Approximately 10, 30, and 50% reduction in fruit size has been observed following irrigation with 5 6, 8, and 9 dsm 1 saline water, respectively (Cuartero and Fernandez-Munoz, 1999). Thus, small-fruited cultivars may be more successful than large-fruited cultivars for growing under low to moderate levels of SS; this is because the relative importance of reduction in fruit size is less in smaller fruits (Caro et al., 1991). At higher levels of salinity or prolonged exposure to salinity, however, a reduction in the total number of fruits per plant, mainly because of a reduction in the number of trusses per plant, is the major cause of yield reduction in tomato (van-ieperen, 1996; Cuartero and Fernandez-Munoz, 1999). Thus, under high levels of salts, significant yield reduction is expected in both large-fruited and small-fruited cultivars. Furthermore, because upper inflorescences are more sensitive to salinity (Cuartero and Fernandez-Munoz, 1999), when breeding for ST in tomato it is better to develop cultivars with compact plant type and short life cycle, which produce only 4 6 trusses before harvest. The potential of tomato wild species as sources of ST during reproduction has not been assessed critically. This is mainly because most of the wild acces-

8 108 sions are self-incompatible and/or produce very small fruit, which cannot be easily compared with those of the cultivated species. However, progenies from early or late generations derived from interspecific crosses have been used for such studies, as discussed below. Inheritance. Most of the earlier research on tomato ST during vegetative stage was focused mainly on the physiological responses to SS. During the past two decades, however, research to discern the genetic basis of ST and to facilitate the development of tomato cultivars with improved ST has been expanding. About 20 years ago, Epstein proposed the exploitation of gene resources within the wild species of tomato to increase ST of the commercial cultivars (Epstein et al., 1980). Subsequently, his group made hybridizations between a salt-tolerant accession (LA1401) of L. cheesmanii and a salt-sensitive tomato cultivar (Walter), and the resulting filial and backcross progeny were evaluated for ST under greenhouse conditions (Rush and Epstein, 1981a). They determined that ST of LA1401 was heritable, as selections in the F 2 and backcross populations produced progeny with improved ST compared to Walter. In another study, evaluation of BC 1 and BC 1 S 1 populations of a cross between a salt-sensitive tomato processing line (M82) and a salt-tolerant accession (LA716) of L. pennellii under field conditions suggested that total dry matter and total fruit yield under saline conditions, and dry matter under salt relative to control conditions, were good selection criteria for ST breeding in tomato; estimates of h 2 for these traits ranged between 0.3 and 0.45 (Saranga et al., 1992). Evaluation of F 2 progeny of a cross between a salt-sensitive tomato line and a salt-tolerant L. pimpinellifolium accession under SS suggested that total fruit yield and total fruit number were useful selection criteria for improving tomato ST; estimates of broad-sense h 2 for these two traits were 0.53 and 0.73, respectively (Asins et al., 1993). Furthermore, in a greenhouse hydroponics study, using parental and reciprocal filial and backcross generations of an intraspecific cross between a salt-sensitive tomato line (UCT5) and a salt-tolerant primitive cultivar (PI174263), Foolad (1996a) determined that growth under SS relative to control, the most widely used index in physiological investigation of ST in tomato and other plant species, was under additive genetic controls and could be an excellent selection criterion for improving tomato ST. In none of the latter three studies, however, was any empirical selection made to verify the suggestion that ST of tomato could be improved by directional phenotypic selection under saline conditions. Nonetheless, these and other studies (Bolarin, 1991; Foolad, 1996a) have indicated that shoot growth under salinity relative to control (i.e., relative growth under SS) is the best practical indicator of ST in tomato. It is generally agreed that direct selection for ST under field conditions is difficult because of the confounding effects of numerous environmental factors (Richards, 1983; Tal, 1985; Yeo and Flowers, 1990). A suggested approach to improve the efficiency of selection for ST has been the adoption of new selection criteria based on the genetic knowledge of physiological processes which limit crop production under saline conditions (Tal, 1985; Yeo and Flowers, 1990; Munns et al., 2002). Physiological criteria which have been suggested or used as potential indicators of ST in tomato include tissue water potential, tissue ion content, K + /Na + ratio, osmoregulation, succulence, and water use efficiency (WUE). However, whether these physiological parameters are good indicators of ST in tomato, or if there is genetic variation in these responses, must be determined before the question of genetic controls can be addressed. Several laboratories have investigated the relationship between tomato ST and some of these physiological responses and have speculated on their utility as indirect selection criteria (Tal and Gavish, 1973; Tal et al., 1979; Noble and Rogers, 1992; Asins et al., 1993; Saranga et al., 1993; Foolad, 1997; Cuartero and Fernandez-Munoz, 1999). Some of these studies are briefly discussed below. Generally, there are two types of salinity-response mechanisms operating in plants; one which controls the entry of salt into the plant (salt exclusion or glycophytic response) (Greenway and Munns, 1980) and the other which controls the concentration of salt in the cytoplasm through compartmentalization, that is, sequestering salt in the vacuole of the cell (salt inclusion or halophytic response) (Greenway et al., 1981). Although most plants may compartmentalize ions to certain degrees, plants are often classified based on their predominant response to salinity. In glycophytes (nonhalophytes), osmotic adjustment is mainly accomplished by restriction of ion uptake (at the root or shoot level) and cellular synthesis of organic solutes (e.g., sugars and amino acids), which are used as osmotica. With increasing SS, however, the ion exclusion mechanism may fail and ionic stress and ensuing imbalances in metabolic processes may occur as a result of excessive ion intake (Lauchli, 1984). In halophytes, on the contrary, osmotic adjust-

9 109 ment is mainly achieved by uptaking inorganic ions from the soil and sequestering them in the cell vacuoles of the leaves or other plant organs (Flowers et al., 1977). Physiological studies have indicated that most of the salt-tolerant genotypes within the cultivated tomato and the closely-related wild species L. pimpinellifolium generally exhibit a glycophytic response to salinity (Caro et al., 1991; Cuartero et al., 1992; Bolarin et al., 1993; Perez-Alfocea et al., 1993; Foolad, 1997; Santa-Cruz et al., 1998). In contrast, salt-tolerant accessions within the tomato wild species L. pennellii,l. cheesmanii, and L. peruvianum generally exhibit a halophytic response (Sacher et al., 1983; Tal and Shannon, 1983; Bolarin et al., 1991; Perez-Alfocea et al., 1994). However, differential accumulation of ions in the leaf tissue has not always been identified as a major factor in determining tomato ST or salt sensitivity. For example, analysis of BC 1 and BC 1 S 1 populations of a cross between a salt-sensitive L. esculentum (M82) and a salt-tolerant L. pennellii (LA716) indicated that tissue ion content was not likely to provide an efficient selection criterion for ST, as no direct relationship was observed (Saranga et al., 1992). Also, no relationship was observed between tissue ions content and plant ST in BC 1 and BC 1 S 1 populations of a cross between a moderately salt-sensitive tomato breeding line (NC84173) and a salt-tolerant accession (LA722) of L. pimpinellifolium (Foolad and Chen, 1999). In contrast, analysis of the relationship between ST and leaf ion composition in the cultivated and three wild species of tomato prompted Saranga et al. (1993) to conclude that dry matter production under SS was positively correlated with K + /Na + ratio in the stem and negatively correlated with the Cl concentration in leaves and stems. The latter study suggested that tissue ion content and ion selectivity were good selection criteria for breeding for ST in tomato. Potassium selectivity over Na + was also reported as a good indicator of ST in a study of several genotypes of the cultivated and wild species of tomato (Cuartero et al., 1992). Further studies of wild species of tomato, including L. peruvianum (Tal, 1971), L. cheesmanii (Rush and Epstein, 1981b), and L. pimpinellifolium, L. hirsutum and L. pennellii (Bolarin et al., 1991), related elevated concentrations of Na + in the leaf tissue to the plant ST. Moreover, other studies suggested that the ability to regulate Na + concentration in the leaf tissue was more closely correlated with ST than Na + concentration per se (Sacher et al., 1983), and that the distribution of Na + in young and mature leaves was important part of such regulation (Shannon et al., 1987). The overall findings from different studies, however, indicate that tissue ion content per se may not be a universal indicator of ST across tomato genotypes, which is also consistent with findings in other crop species (Lauchli and Epstein, 1990; Noble and Rogers, 1992). It has been frequently reported that in salt-tolerant genotypes of the tomato which exhibit a glycophytic response to SS, as ion concentration increases beyond a threshold level the exclusion mechanism fails, and further increases in ion concentration in the root zone results in declining growth and gradual plant death (Perez-Alfocea et al., 1993; Foolad, 1997). Thus, salttolerant tomato genotypes with primarily exclusion mechanism may be only useful for cultivation in regions with low to moderate levels of salt. At higher salinity levels, genotypes that exhibit a halophytictype response may be more useful. Unfortunately, however, many salt-tolerant wild accessions of tomato that exhibit halophytic responses to SS often have the undesirable characteristic of sluggish growth (Foolad, 1996a; Tal, 1997). Although such accessions may survive high levels of salinity, they often grow extremely slowly under such conditions, a highly undesirable trait for tomato cultivation. Whether this negative association is due to pleiotropic effects of the same genes or linkages between different genes is unknown. Furthermore, it is yet to be determined whether this association can be eliminated by selections in segregating populations deriving from crosses between slow-growing salt-tolerant wild accessions and fastgrowing salt-sensitive cultivated tomatoes. If it can, salt-tolerant wild accessions would be useful for ST breeding in tomato. Several studies in tomato and in other plant species have suggested that genes contributing to plant vigor are different from those conferring ST; this is a very desirable phenomenon (Forster et al., 1990; Foolad, 1996a). When breeding tomatoes for efficient production under saline conditions, genes for both plant vigor and ST are essential, which can be potentially incorporated from different resources. Research to determine the genetic basis of ion accumulation/exclusion in tomato plants grown under salinity has been very limited. Analysis of the parental, filial and backcross populations of a cross between a salt-sensitive tomato line (UCT5) and a salt-tolerant primitive cultivar (PI174263) indicated that growth under SS was positively correlated with leaf Ca 2+ content and negatively correlated with leaf Na + content (Foolad, 1997). Generation means analysis of these populations indicated that under SS,

10 110 accumulation of both Na + and Ca 2+ in the leaf was genetically controlled with additivity being the major genetic component. Thus, tissue ion concentration was suggested as a useful selection criterion when breeding for improved ST of tomato using PI as a genetic resource (Foolad, 1997). In spite of these studies, there is no consensus on what might be the best physiological or morphological characteristic(s) that should be used as indirect selection criteria when breeding for ST in tomato. Most likely a combination of different physiological characteristics, including absolute and relative DW of shoot and root, root length, leaf rolling, leaf water potential, leaf osmotic pressure, tissue ion content, K + /Na + ratio, osmoregulation, succulence, WUE, pollen viability and stigma receptivity, flower and fruit set, fruit weight, yield and plant survival, should be considered if salt-tolerant genotypes with commercial value are expected. This, by itself, indicates the complexity of ST and the need for identifying better approaches for plant evaluation and for characterizing genetic bases of tolerance components to facilitate breeding for ST. Recent advances in molecular marker technology, QTL mapping and MAS, and genetic transformation have provided some promising approaches, as briefly discussed below. QTL mapping. As alluded to before, molecular markers technology and QTL mapping have been very useful for discerning the genetic bases of complex inherited traits, including tolerance to abiotic stresses. Such technologies can facilitate determination of the number, chromosomal location, and individual and interactive effects of QTLs controlling tolerance-related traits at different developmental stages. The identified QTLs for various traits can be potentially transferred to desirable genetic backgrounds through a pyramiding approach using MAS. This can be a very effective approach to substantial improvement in plant ST, as described below. In tomato, significant progress has been made in mapping QTLs for ST during vegetative and reproductive stages, using various tolerance-related characteristics. In one study, for example, a BC 1 S 1 population of a cross between a moderately salt-sensitive tomato breeding line (NC84173) and a salt-tolerant accession of L. pimpinellifolium (LA722) was evaluated for ST during vegetative stage using an aerated hydroponics system (Foolad and Chen, 1999). The two parents were distinctly different in ST: while 80% of LA722 survived two weeks after the final salt concentration was reached, only 25% of NC84173 remained alive within that period. The BC 1 S 1 population exhibited a continuous variation for ST, with survival rates ranging from 9% to 94% across families and with a mean of 51%. Interval mapping identified five putative QTLs for ST, with individual effects ranging between 5.7 and 17.7% and with combined effects of 46% of the total phenotypic variation under SS. All QTLs had the positive QTL alleles from the salt-tolerant L. pimpinellifolium parent. The results supported the previous suggestion (Foolad, 1996a, 1997) that ST during vegetative stage in tomato was controlled by more than one gene. However, the involvement of only a few major QTLs, which accounted for a large portion of the total phenotypic variation, suggested the potential utility of MAS for improving tomato ST using LA722 as a gene resource. Analyses of leaf ions content (including Na +,K +,Mg 2+,Ca 2+,Cl, NO3,SO4 2 and PO4 3 )inthebc 1 S 1 population indicated the absence of a correlation between ST and tissue ions content in this population. Furthermore,despite the presence of significant variation among BC 1 S 1 families in concentration of the various ions, no major QTL was identified for tissue ions content under SS. Using a different BC 1 population of the same L. esculentum L. pimpinellifolium cross, a selective genotyping approach was employed to verify the previously identified QTLs and possibly identify new QTLs for ST during vegetative stage (Foolad et al., 2001). From a population of 792 BC 1 plants, 37 (4.7% of the total) that exhibited the highest ST were selected (referred to as the selected population ), grown to maturity and self-pollinated to produce BC 1 S 1 seeds. The 37 selected BC 1 S 1 families and 119 nonselected (random) BC 1 S 1 families were evaluated for ST and their performances were compared. The ST of the selected families was significantly higher than that of the unselected population. A realized h 2 of 0.46 was obtained for this trait, consistent with a previous estimate of h 2 obtained from another tomato population (Foolad, 1996a). The 37 selected BC 1 plants and the 119 nonselected BC 1 plants were subjected to RFLP analysis using 115 markers. A trait-based marker analysis (a.k.a. distributional extreme analysis) identified five QTLs for ST during vegetative growth in this tomato population (Foolad et al., 2001). Except for one, all QTLs had positive alleles from the salt-tolerant L. pimpinellifolium parent. Three of the five QTLs were at the same locations as those identified in the previous study (Foolad and Chen, 1999). Only one of the major QTLs that was identified in the previous study