Breeding for Aluminum Tolerance

Joe Bouton, Mary Sledge, and Wayne Parrott

Department of Crop and Soil Sciences

University of Georgia, Athens, GA 30602

Previously, we wrote a paper summarizing the state of aluminum (Al) tolerance in alfalfa and other forage legumes (Bouton and Parrott, 1997). This current paper will serve as a summary and update of that paper.

Acid, Al-Toxic Soils

Soil acidity is common to soils where rainfall is high enough to leach appreciable amounts of exchangeable bases from the soil surface layers (Brady, 1974). This leaching effectively removes the buffering capacity of the soil and causes a drop in pH. Leaching also encourages acidity by percolating organic acids derived from naturally decomposing organic matter into the soil profile and replacing the bases, which are then removed by the drainage water. Therefore, in many areas of the world, acidity is the most widespread and discussed property of soils.

Under very acid conditions, aluminum (Al) becomes soluble in soil and is present in the toxic Al³⁺ or Al(OH)²⁺ forms (Brady, 1974). These then become adsorbed, even in preference to hydrogen ions, to clay minerals, with the adsorbed Al present in equilibrium with the Al ions in the soil solution. The latter also contribute to overall soil acidity. When soil pH is moved toward neutrality with liming, the toxicity of Al is suppressed by changing to less toxic forms such as Al(OH)⁰. Al can also complex with phosphates, sulfates, or organic acids, such as tartrates or citrates, at any pH and be converted to nontoxic forms.

Al toxicity occurs by definition when the ratio of extractable Al (found in the toxic forms at low pH) to extractable Al plus exchangeable Ca, Mg, and K is greater than 60% within 50 cm of the soil surface. Based on this definition, Al toxicity is estimated to be present in 56% of the soils in the humid tropics (Buol and Eswaran, 1993). Some general level of Al tolerance will be necessary in most crops if these extensive areas are to be brought into some level of productivity.

The most common effect of Al on plant growth is the reduction of root elongation and proliferation, thereby leading to poor water and nutrient extraction (Buol and Eswaran, 1993). Even where liming is practiced, subsoils remain acid and Al-toxic, reducing rooting and the plant's ability to further extract water and nutrients.

Conventional Approaches in Breeding for Aluminum Tolerance

It is clear that genetic diversity for Al tolerance is present within alfalfa (Devine et al., 1990; Mugwira and Haque, 1993; Baligar et al., 1993; Campbell et al., 1989). The genetic tolerance to Al in alfalfa is primarily manifested as general combining ability (Campbell et al., 1994), indicating the potential is present to use phenotypic recurrent selection to enhance Al tolerance. In fact, such a strategy has been attempted, (Campbell et al., 1988; Bouton and Sumner, 1983; Devine et al., 1976; Hartel and Bouton, 1989; 1991), though the technique is not without its limitations. Besides the difficulties inherent in correctly identifying genotypes with Al tolerance, progress from selection in an autotetraploid such as alfalfa will always be slower than in a diploid. In addition, the heterosis and inbreeding depression manifested by autotetraploids can help mask the presence or absence of genes for Al tolerance; i.e,. a highly heterotic genotype may not be as susceptible to Al toxicity, even in the absence of specific genes for Al tolerance (Campbell et al., 1988), while inbreeding depression in an otherwise Al-tolerant genotype would prevent the presence of the Al tolerance genes from being identified.

We have for several years pursued a plant breeding program whose objective was to develop alfalfa germplasms tolerant to these acid, Al-toxic soils. Parents for production of experimental synthetic germplasms were screened and identified in acid, Al-toxic soil, and subjected to recurrent selection. This strategy achieved some success in improving alfalfa's tolerance to acid soils, but also demonstrated that greater levels of tolerance will be needed to make the crop's performance an economic success (Bouton et al. 1981; Bouton and Sumner, 1983; Bouton et al. 1986;Brooks et al. 1982;Bouton and Radcliffe, 1989; Hartel and Bouton, 1989;1991). Nevertheless, the resulting GA-AT germplasm was the most Al-tolerant tetraploid germplasm developed during that time.

The GA-AT germplasm was found to possess better acid soil and Al tolerance than the most appropriate checks as determined by plant growth and nitrogen fixing capacity when grown in both greenhouse (Table 1; Hartel and Bouton, 1989) and field conditions (Table 2; Hartel and Bouton, 1991). This tolerance was also seen at the cellular level, as callus derived from an array of its genotypes was found to show less depression when grown in an Al-toxic medium than callus from the check (Table 3; Parrott and Bouton, 1990). It has likewise shown good field performance in South Africa when tested in unlimed soil (Table 4) and is currently being further tested in that country for possible commercialization (N. Miles, 1997, unpublished data). Finally, the USDA core collection of plant introductions was tested against GA-AT, and none of the tetraploid subspecies (spp. sativa or spp. falcata) demonstrated acid soil and Al tolerance significantly (p<0.05) better than that of GA-AT (Bouton, 1996), indicating no additional sources of tolerance are currently available within the primary cultivated, tetraploid alfalfa gene pool. With this screening, however, GA-AT was established as the standard against which everything will be judged for acid soil tolerance as reported in National Plant Germplasm System's GRIN Network (e.g. Germplasm Retrieval and Information Network). Therefore, because GA-AT provides the scientific community with a good, stable check germplasm, it has also been shared with other scientists throughout the world.

Table 1. Response of two alfalfa germplasms when tested in greenhouse pots containing an unlimed Appling sandy loam soil (pH=4.6; Al=32 µg g^-1) and inoculated with Rhizobium meliloti strain 59 (from Hartel and Bouton, 1989).

†Check was a germplasm derived from a composite of the same parental material as GA-AT but not subjected to acid soil selection.

Table 2. Dry matter yield of two alfalfa germplasms when tested in field plots containing an unlimed Wedowee sandy loam soil (pH=5.0) and inoculated with Rhizobium meliloti strain 59 (from Hartel and Bouton, 1991).

†Means followed by the same letter within a column are not significantly different based on a LSD (p<0.05) performed on square root transformed data in 1989. In 1988, non transformed data were used.

¶ Check was a germplasm derived from a composite of the same parental material as GA-AT but not subjected to acid soil selection.

Table 3. Average growth of callus from two alfalfa germplasms when grown on an acid Blayde medium (pH=4.0) containing aluminum (added as AlCl₃ï6H₂O at 96.58 mg L^-1) as a percentage of callus growth on Blades medium (pH=5.8; Al=0) for an 8- wk period (from Parrott and Bouton, 1990).

†Mean+S.E.

¶Check was a germplasm derived from a composite of the same parental material as GA-AT but not subjected to acid soil selection.

Table 4. Annual dry matter yield of four alfalfa cultivars when tested in an unlimed Oxisol (pH=4.3, Al saturation=15%) at Cedara Research Station, Republic of South Africa for two years (from Neil Miles, 1994, unpublished).

From the impressive response of GA-AT in acid, Al toxic soils (Tables 1- 4), it would be appropriate to ask why was it not being commercialized as a tolerant cultivar? The answer lies in the fact that the annual dry matter yields achieved in acid soils (2176 to 2819 kg ha^-1) is still very small compared to our average yields of >10,000 kg ha^- in conventionally limed plots. Therefore, for the farmer, liming to increase yield five fold is still a very economical undertaking.

By definition, tolerance (from a biotic stress standpoint) is defined as an ability to grow in the stress with little economical loss of yield. An 80% reduction in yield is clearly greater than can be reasonably considered tolerance, so tolerance better than that of GA-AT is essential if an acid-soil-tolerant alfalfa cultivar is to be a commercial success. Since no sources of tolerance were apparent in the primary gene pool of PIs (Bouton, 1996), we have moved the breeding effort into looking for tolerance genes in 1) the secondary gene pool (other related subspecies) and 2) other species or organisms and using transgenics to insert them. A screen of related Medicago subspecies identified PI 464724, a diploid accession of M. sativa ssp. coerulea which is highly resistant to Al-toxicity.

Use of Biotechnology in Breeding for Aluminum Tolerance

As mentioned previously, breeding programs for Al tolerance in alfalfa face several difficulties, and thus have yet to advance beyond the initial steps of germplasm screening and initial selection. Besides the limitations listed previously, the type of assay used to detect the presence of Al tolerance can also be crucial, and the best assay for this purpose has yet to be determined. Traits such as Al tolerance have large environmental interactions, and thus are difficult to detect accurately in the absence of sufficient replication. Given the heterozygous nature of many forage legumes, adequate replication of individual genotypes becomes very difficult, if not impossible.

There is evidence that techniques such as magnetic resonance imaging may be refined to the point where they will be useful in identifying aluminum-resistant genotypes of alfalfa (Campbell, 1999). In the mean time, biotechnology has been considered as a way to help solve the problems of development and enhancement of useful genetic variation for Al toxicity. The use of biotechnology, namely cell culture for screening and/or creation of somaclonal variation, assymetric hybridization, genetic transformation, and marker-assisted selection, have been considered for crop improvement programs in regions where the magnitude to these soil-related problems severely limits production and where the value of the crop justifies the investment of resources.

Cell culture techniques

The ability to define a distinct chemical basis for Al toxicity allows one to design a cell culture screen which may be more definitive than using the problem soil itself. This is because interactions with other soil chemical variables or environmental conditions can mask expression of true tolerance genes, as explained earlier.

Therefore, cell culture techniques have been proposed as an effective tool for identification, creation, and/or selection of useful genetic variation for problem soils. However, to be truly useful, the mechanism of Al tolerance at the cellular level must be the same as that at the whole-plant level. For this discussion, three ways to use cell culture will be considered: 1) as a method of finding useful genes through somaclonal variation, 2) as a bioassay to identify tolerant genotypes by screening cells taken directly from the plant, and 3) as a facilitator of wide hybridization for the interspecific transfer of genes from tolerant species.

Somaclonal variation

Somaclonal variation can be visualized as simply a form of mutation breeding which might offer hope for plant improvement (Larkin and Scowcroft, 1981). Any time plants are regenerated from somatic cells via cell culture, many show evidence of new genetic variability. This variability may be epigenetic and transient, and therefore not useful for crop improvement, as it is not transmitted through meiosis. However, stable genetic changes are common, and if useful, can be of interest to plant breeders. These genetic changes are usually due to any of the following (Evans, 1989): polyploidy, single gene mutations, cytoplasmic gene changes, chromosome rearrangements, mitotic crossing over, and activation of transposable elements. The best application of somaclonal variation to conventional plant breeding lies in introducing the best available cultivars into cell culture and selecting among regenerated plants or their progeny for the desired changes. As of yet, the only report whereby this approach has been successful is in sorghum (Duncan et al., 1995).

Cellular selection

A major limitation to the somaclonal variation approach is that mutations in culture occur at random, and there is no mechanism to select for desired mutations. The desired trait can be recovered with greater efficiency by adding a selection agent to the culture medium. The strategy of selecting cultured cells for mutations conferring Al resistance was pioneered by Meredith (1978), who found that the tolerance at the cellular level corresponded to that of the whole-plant level, and by Ojima and Ohara (1982), who regenerated an Al-tolerant carrot following in vitro selection. A medium was later designed which adequately simulated the acid, Al-toxic conditions found in soil (Conner and Meredith, 1985b) by lowering the pH to 4.0, lowering the Ca and PO₄ concentrations, and adding 600 µM of aluminum. This medium was used to select for Al-tolerant cell lines of Nicotiana plumbaginifolia Viv. Resistance in regenerated plants was conferred by a single, dominant mutation (Conner and Meredith, 1985a; 1985c). A similar approach has been used to recover Al-tolerant regenerants of rice (Jan et al., 1997).

Campbell et al. (1988) reported that cell selection coupled with plant regeneration may be a viable strategy to breeding for Al tolerance in alfalfa. To support this conclusion, Kamp-Glass et al., (1993) modified Schenk and Hildebrant's medium by reducing the concentration of Ca to 11.1 mg l^-1, the eliminating EDTA, reducing Fe to 1.5 mg l ^-1, increasing NH₄ by adding 1.65 g l^-1 of NH₄NO₃, and supplying 150 µM l^-1 of Al. Calli were started from 'Arc', 'Regen Y', and Saranac. The medium produced higher initiation rates and callus growth than that reported by Conner and Meredith (1985a). Saranac was the most tolerant cultivar, with 62% of explants producing callus on the Al-containing medium, as compared to 84% on the medium without Al. By contrast, Arc formed callus from 50 and 79% of explants on medium with and without Al, respectively, while the corresponding figures for Regen Y were 20 and 73%. At last report, 79 regenerated plants were showing some Al tolerance in preliminary field studies, but no information quantifying this resistance is available. Likewise, no information is available as to the heritability of this resistance. This last piece of information is vital, as putative Al tolerance acquired by potato, Solanum tuberosum L., plants after cell selection in vitro has been found to disappear in the absence of selection (Wersuhn et al., 1994).

Bioassays

Cell culture procedures can be adapted to be used as a bioassay. In the previous section, the objective was the preferential recovery of desired mutations which might occur during the cell culture process. In a bioassay, the objective is simply to identify genotypic variation that might already be present in a population of plants. Cells are taken from each plant to be screened, grown as callus, this callus replicated and screened on the appropriate medium containing toxic levels of Al, and compared with replicates on normal media. The plants are therefore assayed for their cellular tolerance to the problem. Obviously, it is critical that the cellular tolerance translate to whole plant tolerance. However, these bioassays are nondestructive to the plants because only some of their cells are isolated, grown, and assayed. Secondly, bioassays overcome the problems associated with somaclonal variation, such as unwanted genetic and epigenetic changes in the regenerants. The tolerant plants can then be identified and used as parents to produce elite germplasm.

When using low pH, Al-toxic medium as a bioassay to screen alfalfa, Parrott and Bouton (1990) were able to distinguish between tolerant and intolerant germplasm (Table 3), and suggested that a similar assay could be used to identify Al-tolerant parents in a plant breeding program. Furthermore, the callus assay suggested that acid resistance and Al resistance might be different phenomena. Later observations with root tips have confirmed this initial observation (Yokomota and Ohima, 1995).

In a follow-up study, Dall'Agnol et al. (1996) investigated the effectiveness of conventional whole-plant screening in soil to screening via a bioassay in cell culture by comparing the acid soil and Al tolerance of the alfalfa germplasms developed by each method as well as with combinations of both methods. Although good germplasms were developed when a cell culture bioassay was used either alone or in tandem with whole-plant screening in soil, whole plant selection in unlimed soil was found to be the fastest and most cost-effective method for producing a germplasm tolerant of an acid soil containing toxic levels of Al.

Asymmetric hybridization

Legumes already exist which are acid-tolerant, such as sericea lespedeza (Joost et al., 1986). This species is very tolerant of acid, Al-toxic soils and is planted in many areas in the southeastern United States to vegetate unlimed roadsides, but is not used widely as a forage legume because of poor palatability and an inability to be intensively grazed (Hoveland and Donnelly, 1985). However, lespedeza is not cross compatible with other forage legumes. The recovery of fertile hybrids via protoplast fusion is probably not possible due to its phylogenetic distance from the other widely used forage legumes (Negrutiu et al., 1989). Such problems with genomic incompatibilities can potentially be avoided through the use of asymmetric hybridization.

In asymmetric fusion, as first attempted by Gupta et al. (1984), the chromosomes in protoplasts of the donor species with the desired trait are broken into small pieces through the use of ionizing radiation. Irradiated protoplasts are then fused with protoplasts of the recipient species. Chromosome or smaller DNA segments from the donor species, small enough to prevent genomic incompatibilities, become integrated into the recipient prior to regeneration. In legumes, this technique was used (Kihara et al., 1983) in an attempt to transfer chromosome segments from soybean (Glycine max [L.] Merr.) into birdsfoot trefoil. However, all soybean chromosomes were lost prior to plant regeneration. Most recently, Li et al. (1993) used this technique to transfer DNA from sainfoin (Onobrychis viciifolia Scop.) into alfalfa, in an attempt to transfer the capacity for tannin production from sainfoin. A total of 43 out of 158 regenerated alfalfa plants contained sainfoin DNA. As a technique, asymmetric hybridization can conceivably be used to transfer traits under complex genetic control across species barriers. The main limitation is that the DNA fragments which get transferred do so at random. The coupling of this technique with marker-assisted selection, as discussed below, could help overcome this limitation.

Molecular Markers

We constructed and published the first map in diploid alfalfa based on RFLP markers (Brummer et al., 1993). Other maps have also been constructed in diploid alfalfa (Echt et al., 1994; Kiss et al. 1993). DNA markers are also being successfully employed in alfalfa improvement in four areas: genetic linkage mapping, germplasm characterization, heterosis and inbreeding, and genome introgression (Osborn et al., 1998). Finally, markers have been used to characterize North American germplasm sources (Kidwell et al., 1994a), to study genetic diversity and its relation to forage yield (Kidwell et al., 1994b and 1994c), and to determine the feasibility of introducing traits from unimproved germplasm into cultivated alfalfa (McCoy and Echt, 1993).

The diploid, Al-tolerant ssp. coerulea genotype mentioned above (now identified in our program as 724-25) was crossed with the coerulea parent (440501-2) from our original mapping population (Brummer et al. 1993), which was found to be an Al-sensitive, diploid genotype. Three F₁s were produced and self-pollinated to produce an F₂ population. A random selection of 104 F₂s representing all 3 hybrids was assayed and given an Al tolerance score by our cell culture screening procedure (Dall'Agnol, 1996; Parrott and Bouton, 1990). These F₂ plants showed a clear and distinct normal distribution of Al tolerance, with the highest frequency of F₂s being at the mid-parent mean, but also showing transgressive segregation for both higher sensitivity and higher tolerance than either parent (Sledge et al., 1996).

Alfalfa cDNAs that were used to construct our diploid alfalfa map (Brummer et al. 1993) were used to probe Southern blots of the population segregating for aluminum tolerance. The parents were screened with 146 cDNA probes, and 58 were polymorphic. These 58 cDNAs were mapped to nine linkage groups. The F₂ genotypic classes for each of the 58 RFLP loci were contrasted with means from the callus-growth bioassay, using analysis of variance (Sledge et al., 1996). Five markers, UGAc044, UGAc141, UGAc191, UGAc471, and UGAc502 are now felt to be associated with Al tolerance (M.K. Sledge, 1999, unpublished data). UGAc502 has the greatest effect, accounting for 18% of the variation in Al tolerance. In contrast, a similar approach in maize, using RFLP markers and bulked-segregant analyses, found only 2 loci responsible for aluminum tolerance (Sibov et al., 1999). Similarly, two loci have been identified in arabidopsis which give rise to Al resistance (Larsen et al., 1998).

Having molecular markers associated with Al tolerance in alfalfa is only half the problem. However, the transfer of these loci for Al resistance to the tetraploid level will be a challenge, and will likely require a combination of marker-assisted selection, analytical breeding, and/or 2n gametes.

Transgenics

Since variation for genes for Al tolerance seem to be in short supply in the current alfalfa germplasm, a potential approach would be the insertion of tolerance genes into elite germplasm of these species (Smith and Bhaskaran, 1993).

Little is known about the mechanism of Al toxicity in plants. Hypotheses include aluminum interactions within the root cell wall, Al disruption of the plasma membrane and related transport processes, and aluminum interactions with cytoplasmic components (Kochian, 1995). In order to tolerate aluminum, plants must either prevent Al uptake by their roots, or detoxify Al after it has been absorbed. Some cultivars of wheat, barley, peas and maize raise the pH of nutrient solutions, making Al insoluble and preventing its uptake by the roots. Some cultivars of wheat and cotton accumulate Al in their roots, but not in their shoots, indicating a possible detoxification system in the roots (Foy, 1988). Recently, using the tissue culture medium we designed (Parrott and Bouton, 1990), a gene for citrate synthase from Pseudomonas aeruginosa (Donald et al., 1989) was transformed into tobacco and papaya plants (de la Fuente et al., 1997). These transformed plants release fourfold more citrate from their roots than do control plants, and are more resistant to the effects of Al when germinated on Al-containing media. This demonstrates that organic acid excretion can be a mechanism of Al tolerance, and that it can be engineered transgenically.

Two additional genes for Al resistance have been mapped in Arabidopsis, and will undoubtedly be cloned, thus making more genes available for engineering Al tolerance in plants. The first of these conditions for increased levels of excreted citrate, malate, and pyruvate (Larsen et al., 1998). Thus resistance obtained using this arabidopsis gene would be similar to that obtained using the citrate synthase gene from Pseudomonas aeruginosa. The second gene increases H⁺ influx in the presence of Al, which leads to an increase in the pH of the rhizosphere and the precipitation of Al out of solution (Degenhardt et al., 1998). Hence transformation with this gene could lead to an Al tolerance like that of the cereal grains mentioned in the preceding paragraph.

Role of Rhizobium

Forage legumes are symbiotic organisms, and tolerance or sensitivity of the Rhizobium symbiont will confound the ability to detect tolerance or sensitivity in the legume host. The yield of alfalfa selected for acid tolerance was enhanced to a greater extent by inoculation with a more acid-tolerant strain of Rhizobium meliloti than with a strain posessing less acid tolerance (Hartel and Bouton, 1991). In addition, the host itself may differ in ability to nodulate under certain conditions. For example, genotypes of Medicago murex Willd., M. polymorpha L., and M. soleirolii Duby have been identified which have an outstanding capacity to nodulate in acid soils (Howieson and Ewing, 1989).

Conclusions

Thus far, there has not been enough genetic variation identified within tetraploid alfalfa germplasms to result in a commercially useful Al tolerant cultivar. Resistance to aluminum has been identified in diploid germplasm, but the exploitation of this resistance has remained difficult. The advent of marker-assisted selection should make it possible to better exploit the genetic variability existing within cultivated germplasm, and can assist with the introgression of resistance from diploid germplasm.

Mutagenesis, in the form of in vitro selection, and assymetric hybridization, still have the potential to increase the genetic variability available for Al resistance. However, judging from the lack of successful reports, one is forced to conclude that cell culture techniques and somaclonal variation have not achieved great success as a widely used plant breeding approach to achieve Al tolerance.

Transgenic approaches are emerging as the strategy with the greatest potential for success. Furthermore, a transgenic approach is not incompatible with a marker-assisted approach. It might be possible to achieve very high levels of Al resistance by pyramiding transgenes with genes tagged with molecular markers. In the end, the most successful strategy will almost certainly be one which combines traditional breeding and selection with some form of biotechnology.

REFERENCES

Baligar, V. C., Campbell, T. A. and Wright, R. J. (1993) Differential responses of alfalfa clones to Al-toxic acid soil. Journal of Plant Nutrition 16, 219-33.

Bouton, J.H. 1996. Screening the alfalfa core collection for acid soil tolerance. Crop Sci. 36:198-200.

Bouton, J. H. and Sumner, M. E. (1983) Alfalfa, Medicago sativa L., in highly weathered, acid soils. V. Field performance of alfalfa selected for acid tolerance. Plant and Soil 74, 431-6.

Bouton, J. H., M. E. Sumner, and J. E. Giddens. 1981. Alfalfa, Medicago sativa L., in highly weathered, acid soils. II. Yield and acetylene reduction of a plant germplasm and Rhizobium meliloti inoculum selected for tolerance to acid soil. Plant and Soil 60:205-211.

Bouton, J. H., M. E. Sumner, J. E. Hammel, and H. Shahandeh. 1986. Yield of an alfalfa germplasm selected for acid soil tolerance when grown in soil with modified subsoils. Crop Sci. 26:334-336.

Bouton, J. H., and D. E. Radcliffe. 1989. Effects of acid soil selection on agronomically important traits in alfalfa. p. 377-378. Proc. XVI International Grassland Congress, Nice, France, 4-11 October 1989.

Bouton, J.H., and W.A. Parrott. 1997. Salinity and aluminum. p. 203-226. In B.D. McKersie and D.C.W. Brown (eds.) Biotechnology and the Improvement of Forage Legumes. CAB International, Wallingford, UK.

Brady, N.C. (1974) The Nature and Property of Soils. Macmillian Publishing Co. Inc., New York.

Brooks, C. O., J. H. Bouton, and M. E. Sumner. 1982. Alfalfa, Medicago sativa L., in highly weathered, acid soils. III. The effects of seedling selection in an acid soil on alfalfa growth at varying levels of phosphorus and lime. Plant and Soil 65:27-33.

Brummer, E.C., Bouton, J.H., and Kochert, G. (1993) Development of an RFLP map in diploid alfalfa. Theoretical and Applied Genetics 86, 329-32.

Buol, S.W., and Eswaran, H. (1993) Assessment and Conquest of Poor Soils. In: Maranville, J.W. (ed.), Adaptation of Plants to Soil Stresses. Intsormil Publication No. 94-2, University of Nebraska, Lincoln. pp.17-27.

Campbell, T.A. (1999) Magnetic resonance imaging of absorbed aluminum in alfalfa roots. Journal of Plant Nutrition. 22, 827-834.

Campbell, T. A., Foy, C. D., McMurtrey III, J. E. and Elgin Jr., J. H. (1988) Selection in alfalfa for tolerance to toxic levels of aluminum. Canadian Journal of Plant Science 68, 743-53.