Heterosis and AFLP Marker Diversity Among Nine Alfalfa Germplasms.

 

 

 

            Ian M. Ray, Armando Segovia-Lerma, Lee W. Murray, and M. Shaun Townsend

 

I.M. Ray and A. Segovia-Lerma, Dep. of Agronomy and Horticulture, and L.W. Murray, University Statistics Center, New Mexico State University, Las Cruces, NM 88003. M.S. Townsend, USDA-ARS, Corvallis, OR 97331. Research support provided by the New Mexico Agric. Exp. Station and a grant from the Southwest Consortium on Plant Genetics and Water Resources (SWC-98-01). * Corresponding author (iaray@nmsu.edu).

            ABSTRACT

Heterotic groups and patterns are of fundamental importance in hybrid breeding. A systematic approach to characterize heterotic responses among nine alfalfa germplasms, that represent most of the genetic diversity in U.S. cultivars, was conducted to provide information to more knowledgeably manipulate heterosis among these populations. Heterotic responses among the germplasms was determined by evaluating forage yield of the nine germplasms and their 36 diallel hybrids in seeded plots that were harvested five times in each of two years. Commercially acceptable yields were obtained from some hybrids of unimproved parents, where at least one parent was adapted to the study environment. Favorable allelic combinations influencing forage yield were obtained using some materials generally considered to be unadapted to the irrigated southwestern U.S. Variation among crosses was attributed primarily to GCA effects, however, SCA effects were also significant. Significant heterosis was detected among the germplasms and relative midparent heterosis ranged from -21% to 55%. Selection of populations with the highest GCA effects, within two broad dormancy categories (i.e., dormant and mostly nondormant), identified all parents for five of the six highest yielding hybrids. The data indicated that unimproved parents possessing good general performance may be useful in commercial breeding programs. Additional opportunities also exist to capitalize on heterosis associated with specific crosses. Analysis of genetic relationships between populations, based on AFLP profiles of bulked DNAs from genotypes within each germplasm, produced essentially two clusters with one resembling M. sativa ssp. sativa and the other M. sativa ssp. falcata. Pairwise genetic distances between germplasms ranged from 0.09 to 0.24. Genetic distance was not a useful predictor of hybrid performance because the association between these traits was apparently influenced by differing frequencies of favorable alleles in the populations under evaluation. Additive genetic effects were much more strongly correlated with yield than were measures of marker heterozygosity suggesting that field evaluation of hybrids is needed to identify high yielding alfalfa hybrid combinations.

Heterosis describes the deviation of a hybrid relative to the mean of its parents. This phenomenon is of fundamental importance in commercial hybrid development. Optimum use of heterosis requires crossing between genetically unrelated germplasm pools, which consist of genotypes that display similar combining ability and heterotic response when crossed with other genetically distinct groups. Heterotic groups can be developed based on hybrid and parental performance, in addition to maintenance of large genetic variance, and low inbreeding depression in the source materials (Hill, 1983; Melchinger and Gumber, 1998). Hybrid and parent performance can be accurately determined under replicated field evaluation. Diallel analyses including parents are useful for such evaluations when a few populations are involved. With many uncharacterized populations, other mating strategies must be used which provide a good estimate of parental general combining ability (GCA). Estimation of population genetic variances is very resource intensive because large numbers of families must be evaluated within each population. Alternatively, DNA markers can be used to measure genetic diversity within populations. These markers may also be used to group populations based on their genetic similarities (Melchinger and Gumber, 1998). Practical application of molecular markers to genetic studies of heterozygous, heterogeneous populations, however, have been limited because large numbers of individuals need to be examined. Bulking DNA, from individual genotypes within populations, may provide an efficient approach to study genetic relationships among heterogeneous populations. The restriction fragment length polymorphism (RFLP) and random amplified polymorphic DNA (RAPD) patterns obtained from bulked DNA samples in alfalfa have generally represented consensus patterns for the common markers seen for individual lines (Kidwell et al., 1994a; Yu and Pauls, 1993b). Lastly, minimizing inbreeding depression in source populations can most easily be achieved by maintaining reasonable effective population sizes during cycles of selection. Alternatively, high levels of heterosis may potentially be achieved using partially inbred sources that are mated in double crosses or double-double crosses.

Studies examining relationships between genetic distance among parents and hybrid heterosis have provided conflicting results. Moll et al. (1962, 1965) first reported that crosses between maize parents from different geographic areas tended to produce hybrids with greater heterosis than parents from similar geographic origins. A decrease in heterosis was observed, however, when levels of genetic diversity were very high. Bonierbale et al. (1993) reported similar results in autotetraploid potato. Some studies in alfalfa and maize detected positive associations between DNA marker diversity and hybrid yield (Kidwell et al., 1994b, 1994c; Smith et al., 1990, Osborn et al., 1998). Most studies, however, reported low correlations between DNA marker distance and hybrid performance (Lee et al., 1989, Godshalk et al., 1990; Melchinger et al., 1990; Martin et al., 1995). These variable observations, in conjunction with simulation studies of Bernardo (1992), suggest that associations between hybrid yield and genetic distance are influenced by the frequency of favorable/unfavorable alleles present in the populations under evaluation.

             Heterosis plays a key role in maximizing alfalfa forage yield. Various studies indicate that higher alfalfa yields result from the accumulation of favorable alleles through breeding and selection (Pfeiffer and Bingham, 1983a, 1983b; Woodfield and Bingham, 1995), and potentially through maximizing complementary gene interaction (Bingham et al., 1994; Bingham, 1998). Complementary gene interaction refers to nonallelic interaction (i.e., epistasis) where dominant alleles at heterozygous loci compliment each other by masking recessive alleles at respective loci. One approach to maximize heterosis and complementary gene interactions in alfalfa is to intermate genetically diverse populations. Bingham (1983) proposed that controlled crosses between source populations could result in hybrid progeny with nearly twice the frequency of maximally heterozygous loci and greater heterosis compared to randomly mated populations. Similar results should also be obtained with respect to complementary gene interactions. Furthermore, if selection is practiced within populations prior to hybridization, the opportunity exists for the breeder to exploit many forms of epistasis.

Much of the genetic diversity present in North American alfalfa cultivars traces to nine geographically distinct germplasm sources: African, Chilean, Flemish, Indian, Ladak, M.falcata, M. varia, Peruvian, and Turkistan (Barnes et al., 1977). Determining heterotic response patterns among these populations could provide information to better manipulate heterosis in cultivated alfalfa. Evaluation of DNA marker diversity among germplasms may be useful for classifying populations into genetically similar groups, as well as, providing information useful for predicting heterosis. The objectives of this research were: (i) to describe heterotic responses among diallel hybrids of nine alfalfa germplasms, (ii) to determine if amplified fragment length polymorphisms (AFLPs) from genomic DNA bulks could be used to describe genetic relationships among the nine germplasms, and (iii) to determine if AFLP marker based genetic relationships among the germplasms were associated with heterotic response.

            MATERIALS AND METHODS

Nine alfalfa germplasms were synthesized using two "relic" seed sources (1940-1960 era) that represented each germplasm. Seed sources were obtained from breeding programs in California, New Mexico, Washington, and Wisconsin and were grouped according to their respective germplasm designation (Table 1). Each germplasm was represented by 30 plants (i.e., 15 plants per source) in order to account for genetic heterogeneity within the seed sources. The germplasms were crossed by hand in a half-diallel. Hybrid populations were generated by crossing each plant within a germplasm to one other plant in each of the other eight germplasms. Parental populations were synthesized by randomly intercrossing all genotypes within a given germplasm. An equal number of seed, within each cross from each plant, was bulked to form balanced composite populations for each cross.

Field performance of the nine parents, 36 hybrid populations, and four check cultivars were evaluated using three-row 1.5-m plots that were seeded at a rate of 300 seed plot^-1. Control plots of the cultivar Dona Ana, were established between experimental plots to minimize interplot interactions. Three replicates of this study were planted near Las Cruces, NM, in randomized complete blocks. Forage yield was collected over 5 harvests each in 1997 and 1998. Data were collected when the Flemish and Turkistan germplasms (possessing moderate maturation rates) reached 10% bloom. Analyses were based on mean yield over harvests within each year. Analyses of hybrid performance, parent heterosis, specific heterosis, general combining ability (GCA), and specific combining ability (SCA) were determined according to Method II and Method III of Gardner and Eberhart (1966). Midparent heterosis (MPH) was calculated as the difference between hybrid yield and the parental mean (midparent value = MPV) expressed in percentage relative to the MPV. High parent heterosis (HPH) was determined as the difference between hybrid yield and its highest yielding parent expressed in percentage relative to the high parent yield.

Total genomic DNA was isolated (Yu and Pauls, 1993a) from leaf tissue collected from the 30 genotypes representing each germplasm. Equal quantities of DNA from each genotype were pooled within their respective germplasm. Bulked DNAs of each germplasm were subjected to AFLP (Perkin Elmer Applied Biosystems, Foster City, CA) analysis using ten fluorescently labeled primer combinations according to the procedures of Thomas et al. (1995) and Vos et al. (1995). The DNA fragments were resolved on a 5% denaturing polyacrylamide gel using an ABI 377 DNA sequencer. Fragment data were analyzed using GeneScan software. Polymorphic DNA fragments were scored as either present (1) or absent (0) in each DNA bulk. The AFLP data were subjected to cluster analysis using the Numerical Taxonomy and Multivariate Analysis System (NTSYS; Rohlf, 1993) to produce a matrix of genetic similarity values based on Jaccard's coefficient of similarity (J). Similarity estimates were converted to genetic distance (GD) according to Swofford and Olson (1990). The genetic distance matrix was used to create a dendogram using the unweighted pair-group arithmetic average method (UPGMA).

Pairwise genetic distances between the nine germplasms, based on the AFLP analyses, were tested for associations with field performance of the 36 hybrid populations. Correlations between these traits and GCA, MPV, MPH, and HPH were also determined. Multiple linear regression was used to test the association between hybrid forage yield (dependent variable) and GD in combination with GCA (both independent variables). The correlation was calculated between the predicted values obtained for forage yield from the regression and actual forage yield. M. falcata and its hybrids strongly biased associations between GD and forage yield, therefore, trait associations were reexamined by excluding M. falcata hybrids from the analyses. The influence of favorable allele frequency on associations between hybrid yield and GD were examined using correlation coefficients between the two traits obtained from the "half-sib" families/hybrids derived from each parent germplasm. Correlations for eight of the germplasms were based on n = 7 hybrids (M. falcata hybrids excluded for reasons stated above). Correlations between GD and yield within M. falcata were based on n = 8 hybrids.

            RESULTS AND DISCUSSION

Field Performance and Heterotic Patterns of Diallel Hybrids

Significant differences in forage yields were detected for all main effects using Analysis II (Table 2). With the exception of average heterosis, these effects also demonstrated significant interactions with years. The magnitudes of the interaction mean squares, however, were substantially smaller than the mean squares of the main effects. The results of Analysis III indicated that variation among crosses was attributed primarily to GCA effects, however, SCA effects were also significant (Table 3). This behavior would be expected if additive effects were of major importance. The significance of SCA effects in both Analysis II and III suggest that opportunities may exist to capitalize on heterosis associated with specific crosses.

            Forage yields of African and Chilean were highest while M. falcata, Ladak, Flemish, and M. varia were lowest (Table 4). Ten of the 36 hybrids from these unimproved parents produced forage yields comparable to the cultivars Wilson, Commercial 1, Commercial 2, and Dona Ana, which yielded 1.98, 1.75, 1.70 and 1.67 kg/plot, respectively. These data suggest that commercially acceptable yields can be obtained using unimproved parents. Each of these 10 hybrids were derived from crosses involving either Peruvian or Chilean as parents. Three hybrids, Peruvian x African, Peruvian x Chilean, and Chilean x Flemish, yielded similarly to Wilson, the best check cultivar. Forage yields of the three remaining checks were similar to each other and to hybrids derived from crossing Chilean to Flemish, Turkistan, Indian, Ladak, and M. varia, as well as, crossing Peruvian to Turkistan, Flemish, and African. The largest positive SCA estimates (data not shown) were associated with the African x M. falcata, African x Peruvian, and Chilean x Flemish hybrids. Two of these three hybrids ranked #1 and #3 for yield. The vigor shown in the Chilean x Flemish, Chilean x Ladak, Chilean x M. varia, and Peruvian x Ladak hybrids indicate that important alleles/allelic combinations influencing forage yield can be obtained using materials generally considered to be unadapted to the irrigated southwest.

Individual GCA effects were positive for African, Chilean, Peruvian, and Turkistan and were negative for M. falcata, M. varia, Ladak, Indian, and Flemish (Table 4). Relative midparent heterosis ranged from a low of -21% (MPH=79%) in the Indian x Ladak hybrid, to 55% (MPH=155%) in the Chilean x M. falcata hybrid. M. falcata hybrids performed poorly but demonstrated the highest average MPH because M. falcata yielded poorly as a parent resulting in low MPVs. M. falcata's contribution to crosses was negative (GCA=-0.38) limiting its usefulness as a parent in our study environment. Conversely, Peruvian showed high levels of heterosis in all crosses, the highest cross-mean performance, and possessed a GCA effect comparable to that of Chilean. Although yields of Peruvian per se were moderate, these data suggest that it may have value as a parent for breeding purposes. Townsend and Henning (1995) and Barnes et al. (1977) reported that Peruvian germplasm has been utilized in only a small percentage of U.S alfalfa varieties and its contribution to those varieties is small (<10%). The strong performance of the Chilean hybrids was expected in our study environment based on previous experience (Melton, 1999, pers. comm.; Ray et al., 1998).

            These data suggest that parent selection should be based on hybrid performance, parental GCA effects, and SCA estimates of specific hybrids. Practically speaking, hybrid performance is crucial for commercial forage production. The relatively low ratio of s²_SCA to s²_GCA reported in our study and elsewhere (Hill, 1983) indicated that early testing should be effective in alfalfa. It should also be possible to identify and select parent populations of superior synthetic hybrids based mainly on their GCA effects. The significance of SCA effects in our study suggests that further improvements in yield can be made by identifying specific high yielding crosses among good general combiners. Given the resources required for diallel analyses it is difficult to envision using this approach, or any other controlled mating design, to evaluate large numbers of parents such as exist in the alfalfa core collection (n = 200) or in perennial Medicago accessions (n = 2400) within the National Plant Germplasm System (Basigalup et al., 1995; Diwan et al., 1995). Evaluation of topcross or polycross progeny or parents per se, however, would provide alternative approaches for identifying useful parents in these collections based primarily on their GCA. Evaluation of polycross or topcross progeny would identify the most promising parents from a group of size n with only n crosses instead of n(n-1)/2 crosses (e.g., half diallel). These two approaches, however, still require considerable effort and time. Once parents with good combining ability are identified the breeder has at least two options. They may choose to pool/intermate the parents to form one large gene pool, which is then subjected to recurrent selection and hybridization between elite genotypes. This approach increases additive genetic variation within the population and allows the breeder to select for additive effects and additive x additive epistasis. Synthetics produced by the pooling approach, however, will have fewer maximally heterozygous loci and fewer complementary gene interactions. Alternatively, the breeder could practice recurrent selection within selected populations followed by hybridization between elite genotypes from different populations for commercial cultivar development. This approach allows the breeder to select for favorable additive alleles, capitalize on all forms of epistasis (including complementary gene interactions), and generate maximally heterozygous synthetics.

The usefulness of this approach can be partially demonstrated using our data. Due to the influence of fall dormancy on forage yield we divided the nine germplasms into two dormancy categories involving dormant (i.e., M. falcata, Ladak, M. varia, Turkistan, and Flemish) and relatively nondormant materials (i.e., Chilean, Peruvian, Indian, and African). Selection of two populations with the best GCA within a given dormancy category identified Turkistan and Flemish as the best dormant combiners and Peruvian and Chilean as the best nondormant combiners. The five hybrids from these four parents, excluding the Flemish x Turkistan (i.e., unadapted x unadapted) cross, were five of the six highest yielding hybrids. Extending this analysis to multiple environments would help to identify which of the nine germplasms have the greatest breeding potential for various geographic regions in the U.S. and elsewhere. Our data indicate that only nine topcross/polycross populations would need to be evaluated at each location rather than their 36 diallel hybrids. Evaluation of parents would also be useful although somewhat less informative. A similar type of approach could potentially be used to systematically evaluate unimproved accessions in the National Plant Germplasm System for their usefulness as parents in commercial breeding programs. Accessions demonstrating positive GCA effects for a given environment(s) could be mated in partial diallels to determine heterotic patterns for classifying them into combining ability/heterotic groups. Populations within a heterotic group could be pooled or divided into a subset of pools for intrapopulation improvement programs. Heterotic groups would not be considered as closed populations, but could be broadened continuously by introgressing unique germplasm (with similar heterotic patterns) to ensure medium and long-term gains from selection. Collaboration among public and private organizations would be needed to achieve this goal.

Estimates of Genetic Diversity Among the Nine Germplasms

Variation of RFLPs and RAPDs among and within Medicago species and subspecies has demonstrated significant genetic diversity at both the diploid and tetraploid levels (Brummer et al., 1991, 1995; Echt et al., 1992). Kidwell et al., (1994a) characterized RFLP profiles of 12 individual genotypes within each of nine germplasms that were similar to those used in our study. In general, the UPGMA cluster based on 358 AFLPs from bulked DNAs of the nine germplasms (Fig. 1) agreed with the RFLP results reported by Kidwell et al. (1994a); at least for those populations whose genotypes clustered reasonably well in their study. Both studies essentially classified the nine germplasms into two main clusters of M. sativa ssp. sativa and M. sativa ssp. falcata. Principle component analysis results reported by Kidwell et al. (1994a) provided additional information suggesting that the Peruvian germplasm was genetically distinct from the other seven populations within the ssp. sativa complex. Pair-wise genetic diversity estimates among the eight germplasms comprising the M. sativa ssp. sativa complex were remarkably narrow and ranged from 0.09 to 0.12 (Table 5). Pair-wise genetic distances between M. sativa ssp. falcata and the other 8 germplasms were greater and averaged 0.24. Our GD estimates were 54 to 75% (avg. = 70%) less than those reported by Kidwell et al. (1994a). These differences likely resulted from the DNA bulking process where less common fragments, that were present in the patterns of individual genotypes, may have been lost through a dilution effect. Yu and Pauls (1993b) reported that GD based on pooled DNA samples from seven individuals averaged 43% less (range 38 to 44%) than that obtained based on individual genotypes.

Influence of Average Favorable Allele Frequency on Associations between Yield & Genetic Distance

Simulation results of Bernardo (1992) and empirical evidence (Moll, 1962, 1965; Bonierbale et al., 1993) indicate that associations between general marker heterozygosity and either hybrid performance or GCA may be influenced by allele frequencies in the tester lines. Bernardo (1992) predicted positive associations between hybrid performance and GD if lines with low frequencies of favorable alleles were crossed to lines with higher frequencies of favorable alleles. However, this association became negative if lines with high frequencies of favorable alleles were crossed to lines with lower frequencies of favorable alleles. We attempted to determine the influence of average favorable allele frequencies on associations between hybrid yield and GD in alfalfa by examining associations between hybrid yield and GD within crosses made to each germplasm (Table 6). Populations that were poorly adapted to our study environment (i.e., M. varia, Flemish, Ladak, and M. falcata) demonstrated positive correlations between hybrid yield and GD; in agreement with the results of Bernardo (1992). Populations that were better adapted to our environment (i.e., Chilean, African, Peruvian, and Turkistan) demonstrated negative correlations between hybrid yield and GD; also in general agreement with the results of Bernardo (1992). These data support the premise that hybrid performance and heterosis are not merely a function of genetic distance but also depend on the adaptation of the parents. The associations between GD and hybrid yield are biologically reasonable since adapted parents are more likely to be genetically similar to each other, and their hybrids should be relatively high yielding. Conversely, adapted and unadapted parents are more likely to be genetically diverse and their hybrids should be lower yielding on average.

Predicting Hybrid Performance

Associations between some traits were strongly influenced by M. falcata. Correlation analyses, therefore, were conducted using two data sets that contained and excluded M. falcata hybrids (Table 7). In both analyses forage yield was positively associated with GCA, MPV, and HPH. GCA was also positively associated with MPV and HPH in both analyses. The positive association between MPV and hybrid yield indicated that superior parents tended to produce higher yielding hybrids. Hybrid yield, however, was more highly correlated with GCA than MPV indicating that progeny performance was a better predictor of hybrid yield than parental performance per se. A classic example of this was demonstrated by the Peruvian population which possessed intermediate yields, per se, but exhibited a GCA value as high as that of the best parent, Chilean. The superiority of GCA over parental performance was further supported by the observation that GCA was positively associated with HPH and MPH, while MPV showed no association with either HPH or MPH (M. falcata excluded from analysis). The lack of associations between MPV and either MPH or HPH indicated that heterosis per se was not influenced by parental performance. When M. falcata hybrids were included in the analysis the positive association between MPV and HPH and the negative association between MPV and MPH primarily reflected that M. falcata hybrids tended to have both low MPV and HPH values, but high MPH values. When M. falcata hybrids were excluded from the analyses MPH was positively associated with yield, GCA and HPH. These associations were not significant when M. falcata hybrids were included because M. falcata hybrids possessed high levels of MPH but relatively low values for yield, GCA and HPH. Genetic distance was not associated with any trait/parameter when M. falcata was excluded from the analysis. Large pairwise GDs between M. falcata and the other eight populations, in conjunction with the poor performance of M. falcata and its hybrids, resulted in negative correlations between GD and forage yield, HPH, MPV and GCA. M. falcata hybrids possessed the highest MPH values in the study resulting in a positive correlation between GD and MPH.

In our study, hybrid performance was influenced by both GCA and SCA effects. Therefore, we tested the relationship between hybrid yield and both GCA and GD (as a potential indirect estimator of SCA) using multiple linear regression. The correlation coefficient between yield predicted from this regression model and actual yield (r=0.93; P<0.01; n=36) was similar to that obtained between hybrid yield and GCA alone (r=0.91;P<0.01; n=36). Similar results were obtained when M.falcata hybrids were excluded from the analysis (n=27; data not shown). In potato and oilseed rape additive genetic effects were also reported to be more strongly correlated with yield components than were measures of marker heterozygosity (Bonnierbale et al., 1993; Diers et al., 1996). These results suggest that genetic distance estimates alone do not consistently identify high yielding hybrid combinations. This is reasonable since total genome heterozygosity is less important in its influence on heterosis, per se, than heterozygosity at specific loci controlling traits of interest.

The results of our study indicate that commercially acceptable yields were obtained from some hybrids of unimproved parents, where at least one parent was adapted to the study environment. Selection of parents based on GCA effects was very effective in identifying parents of the highest yielding hybrids. Unimproved parents possessing good general performance (e.g., GCA effects), therefore, could potentially be used to immediately widen cultivated alfalfa's gene base without sacrificing forage yield. The significance of SCA effects indicated that additional opportunities exist to capitalize on heterosis using specific crosses. Analysis of AFLP profiles based on bulked DNAs from genotypes within each germplasm was a useful approach to determine genetic relationships between populations. Genetic distance based on AFLP markers was not a useful predictor of hybrid performance, however, because associations between hybrid yield and GD were apparently influenced by allele frequencies of the populations under evaluation. Additive genetic effects were more strongly correlated with yield than were measures of marker heterozygosity suggesting that field evaluation of hybrid performance is needed to identify high yielding alfalfa hybrid combinations.

            LITERATURE CITED

Basigalup, D.H., D.K. Barnes, and R.E. Stucker. 1995. Development of a core collection for perennial Medicago plant introductions. Crop Sci. 35:1163-1168.

Bingham, E.T., R.W. Groose, D.R. Woodfield, and K.K. Kidwell. 1994. Complementary gene interactions in alfalfa are greater in autotetraploids than diploids. Crop Sci. 34:823-829.

Bingham, E.T. 1983. Maximizing hybrid vigour in tetraploid alfalfa. p. 130-143. In: Nugent and O’Connor (ed.) Better crops for food. Pitman, London.

Bonierbale, M.W., R.L. Plaisted, and S.D. Tanksley, 1993. A test of the maximum heterozygosity hypothesis using molecular markers in tetraploid potatoes. Theor. Appl. Genet. 86:481-491.

Brummer, E.C., J.H. Bouton, and G. Kochert. 1995. Analysis of annual Medicago species using RAPD markers. Genome 38:362-367.

Brummer, E.C., J.H. Bouton, and G. Kochert. 1993. RFLP variation in diploid and tetraploid alfalfa. Theor. Appl. Genet. 86:329-332.

Brummer, E.C., G. Kochert, and J.H. Bouton. 1991. RFLP variation in diploid and tetraploid alfalfa. Theor. Appl. Genet. 83:89-96.

Diers, B.W., P.B.E. McVetty, and T.C. Osborn. 1996. Relationship between heterosis and genetic distance based on restriction fragment length polymorphism markers in oilseed rape (Brassica napus L.) Crop Sci. 36:79-83.

Diwan, N., M.S. McIntosh, G.R. Bauchan. 1995. Methods of developing a core collection of annual Medicago species. Theor. Appl. Genet. 90:755-761.

Echt, C.S., K.K. Kidwell, S.J. Knapp, T.C. Osborn, and T.J. McCoy. 1993. Linkage mapping in diploid alfalfa (Medicago sativa). Genome 37:61-71.

Echt, C.S., L.A. Erdahl, and T.J. McCoy. 1992. Genetic segregation of random amplified polymorphic DNA in diploid cultivated alfalfa. Genome 35:84-87.

Gardner, C.O. and S.A. Eberhart. 1966. Analysis and interpretation of the variety cross diallel and related populations. Biometrics 22:439-452.

Godshalk, E.B., M. Lee, and K.R. Lamkey. 1990. Relationship of restriction fragment length polymorphisms to single-cross hybrid performance of maize. Theor. Appl. Genet. 80:272-280.

Havey, M.J. 1998. Molecular analyses and heterosis in the vegetables: can we breed them like maize? p. 109-116. In: K.R. Lamkey and J.E. Staub (ed.) Concepts and Breeding of Heterosis in Crop Plants. CSSA Special Publication 25. CSSA, Madison, WI.

Hill, R.R. 1983. Heterosis in population crosses of alfalfa. Crop Sci. 23:48-50.

Kidwell, K.K., D.F. Austin, and T.C. Osborn. 1994a. RFLP evaluation of nine Medicago accessions representing the original germplasm sources for North American alfalfa cultivars. Crop Sci. 34:230-236.

Kidwell, K.K, D.R. Woodfield, E.T. Bingham, and T.C. Osborn. 1994b. Molecular marker diversity and yield of isogenic 2x and 4x single-crosses of alfalfa. Crop Sci. 34:784-788.

Kidwell, K.K., E.T. Bingham, D.R. Woodfield, and T.C. Osborn. 1994c. Relationships among genetic distance, forage yield, and heterozygosity in isogenic diploid and tetraploid alfalfa populations. Theor. Appl. Genet. 89:323-328.

Lee, M., E.B. Godshalk, K.R. Lamkey, W.L. Woodman. 1989. Association of restriction fragment length polymorphisms among maize inbreds with agronomic performance of their crosses. Crop Sci. 29:1067-1071.

Martin, J.M., L.E. Talbert, S.P. Lanning, and N.K. Blake. 1995. Hybrid performance in wheat as related to parental diversity. Crop Sci. 35:104-108.

Melchinger, A.E. and R.K. Gumber. 1998. Overview of heterosis and heterotic groups in agronomic crops. p.29-44. In: K.R. Lamkey and J.E. Staub (ed.) Concepts and Breeding of Heterosis in Crop Plants. CSSA Special Publication 25. CSSA, Madison, WI.

Melchinger, A.E., M. Lee, K.R. Lamkey, and W.L. Woodman. 1990. Genetic diversity for restriction fragment length polymorphisms: relation to genetic effects in maize inbreds. Crop Sci. 30:1033-1040.

Moll, R.H., W.S. Salhauana, and H.F. Robinson. 1962. Heterosis and genetic diversity in variety crosses of maize. Crop Sci 2: 197-198.

Moll, R.H., J.H. Lonquist, J.V. Fortuno, and E.C. Johnson. 1965. The relationship of heterosis and genetic divergence in maize. Genetics 52:139-144.

Osborn, T.C., D.J. Brouwer, and K.K. Kidwell. 1998. Exploiting genome differences for higher alfalfa forage yield. p. 95. In: Bouton and Bauchan (eds.). Proceedings of the North American Alfalfa Improvement Conference. Bozeman, MT.

Pfeiffer, T.W. and E.T. Bingham. 1983a. Comparisons of alfalfa sexual derivatives and tissue culture variants from the same genetic source. Theor. Appl. Genet. 67:263-266.

Pfeiffer, T.W. and E.T. Bingham. 1983b. Improvement of fertility and herbage yield by selection within two-allele populations of tetraploid alfalfa. Crop Sci. 23:633-636.

Ray, I.M., S.T. Townsend, and J.A. Henning. 1998. Variation for traits associated with yield, water-use efficiency, and canopy morphology among nine basic alfalfa germplasms. Crop Sci. 38:1386-1390.

Rohlf, F.J. 1993. NTSYS-pc: Numerical taxonomy and multivariate analysis system. Version 1.70 Exeter Software, Setauket, N.Y.

Smith, O.S., J.S.C. Smith, S.L. Bowen, R.A. Tenborg, and S.J. Wall. 1990. Similarities among a group of elite maize inbreds as measured by pedigree, F₁ grain yield, heterosis, and RFLPs. Theor. Appl. Genet. 80:833-840.

Swofford, D. L. and G.J. Olson. 1990. Phylogeny reconstruction. In: D.M. Hillis and C. Moritz (eds.) Molecular systematics. Sinauer Associates, Sunderland, MA.

 

Thomas, C.W., P. Vos, M. Zabeau, D.A. Jones, K.A. Norton, B.P, Chadwick, and J.D. Jones. 1995. Identification of amplified restriction fragment polymorphism (AFLP) markers tightly linked to the tomato Cf-9 gene for resistance to Cladosporium fulvum. Plant Journ. 8:785-794.

Townsend, M.S., J.A. Henning, and C.G. Currier. 1994. The alfalfa catalog software package. Agron. J. 86:337-339.

Vos, P., R. Hogers, M. Bleeker, M. Reijans, T. van de Lee, M. Hornes, A. Fritjers, J. Pot,. J. Pelerman, M. Kuiper, and M. Zabeau. 1995. AFLP: a new concept for DNA fingerprinting. Nucl. Acid Res.

Woodfield, D.R., and E.T. Bingham. 1995. Improvement in two-allele autotetraploid populations of alfalfa explained by accumulation of favorable alleles. Crop Sci. 35:988-994.

Yu, K.F. and K.P. Pauls. 1993a. Segregation of random amplified polymorphic DNA markers and strategies for molecular mapping in tetraploid alfalfa. Genome 36:844-851.

Yu, K.F. and K.P. Pauls. 1993b. Rapid estimation of genetic relatedness among heterogeneous populations of alfalfa by random amplification of bulked genomic DNA samples. Theor. Appl. Genet. 86:788-794.

Table 1. Accessions used to represent the nine basic alfalfa germplasms.

Germplasm Source	Representative Accessions (Era)	Origin of Seed Source
African	Moapa (pre 1956) Higazzi (pre 1956)	L. Teuber/C. Stanford, Univ. CA-Davis L. Teuber/C. Stanford, Univ. CA-Davis
Chilean	Arizona Chilean (1957) Chilean 21-5 (1950)	B. Melton, New Mexico State Univ. B. Melton, New Mexico State Univ.
Flemish	DuPuits (1960) Alfa (1960)	National Plant Germplasm System, Pullman, WA
Indian	FC23631 (pre 1960) FC32174 (pre 1960)	L. Teuber/C. Stanford, Univ. CA-Davis L. Teuber/C. Stanford, Univ. CA-Davis
Ladak	Ladak 1 (pre 1950) Ladak 2 (pre 1950)	E.T. Bingham, Univ. WI-Madison E.T. Bingham, Univ. WI-Madison
M. falcata	4x Wisfall (1993)	E.T. Bingham, Univ. WI-Madison
M. varia	Grimm (1960) Cossack (1951)	L. Teuber/C. Stanford, Univ. CA-Davis E.T. Bingham, Univ. WI-Madison
Peruvian	Hairy Peruvian (1959) Coastal Hairy Peruvian (1959)	B. Melton, New Mexico State Univ. B. Melton, New Mexico State Univ.
Turkistan	Turkistan-Samarkand. (pre 1950) Turkistan-Tokmak (pre 1950)	E.T. Bingham, Univ. WI-Madison E.T. Bingham, Univ. WI-Madison

Table 2. Mean squares for dry-matter yield (kg plot^-1) from alfalfa diallel hybrids

and their parents over 2 yr and one location using Analysis II of Gardner and Eberhart.

Source	df	Mean square
Years	1	0.934**
Entries (E)	44	3.109**
Parents (P)	8	13.889**
Crosses (C)	36	0.714**
Avg. Heterosis (H)	1	5.957**
Parent Heterosis (PH)	8	1.512**
Specific Heterosis (SH)	27	0.390**
E x Y	44	1.072**
P x Y	8	0.368**
C x Y	36	0.129**
H x Y	1	0.010
PH x Y	8	0.104*
SH x Y	27	0.140**
Residual	90	0.045
CV %		7.2

Table 3. Mean squares for dry matter yield (kg plot^-1) from alfalfa diallel hybrids and their parents in each

of 2 years using Analysis III of Gardner and Eberhart.

Source	df	1997	1998
Entries	44	1.269**	2.012**
Parents	8	2.703**	3.134**
Parents vs. Crosses	1	3.239**	2.729**
Crosses	35	0.887**	1.736**
General (GCA)	8	3.300**	6.377**
Specific (SCA)	27	0.170*	0.361**
Residual	90	0.099	0.146

GCA, SCA: General and specific combining ability, respectively.

Table 4. Forage yields (kg plot^-1) of crosses (above diagonal) and parents (diagonal); relative heterosis and estimates of GCA effects (both below diagonal) from a diallel set of nine synthetics and crosses averaged over 1 location 2 years (10 harvests)

Synthetic	African	Chilean	Flemish	Indian	Ladak	M.falc.	M. varia	Peruvian	Turkistan	Mean
African	1.60	1.51	1.39	1.50	1.50	1.33	1.41	1.97	1.52	1.50
Chilean	94	1.60	1.78	1.61	1.59	1.36	1.55	1.95	1.64	1.62
Flemish	101	129	1.14	1.45	1.01	0.88	1.30	1.63	1.43	1.35
Indian	97	104	110	1.48	0.99	1.00	1.23	1.63	1.41	1.35
Ladak	113	120	93	79	1.04	0.91	1.05	1.54	1.30	1.23
M. falcata	151	155	135	122	152	0.16	0.78	1.16	1.02	1.05
M. varia	101	110	111	92	94	116	1.19	1.49	1.28	1.26
Peruvian	132	130	129	114	127	151	116	1.37	1.66	1.63
Turkistan	102	110	114	99	108	134	100	121	1.36	1.40
Mean	111	119	114	102	110	139	104	127	111
GCA effect	0.13	0.26	-0.04	-0.04	-0.18	-0.38	-0.14	0.27	0.01

 

Table 5. Pairwise genetic distances among 9 alfalfa germplasms based on 358 AFLPs.

	Chilean	Flemish	Indian	Ladak	M.falc.	M. varia	Peruvian	Turkistan
African	0.114	0.104	0.109	0.118	0.240	0.121	0.098	0.094
Chilean		0.104	0.090	0.110	0.242	0.105	0.097	0.108
Flemish			0.110	0.108	0.230	0.103	0.112	0.106
Indian				0.115	0.245	0.093	0.109	0.107
Ladak					0.226	0.104	0.122	0.109
M. falcata						0.233	0.243	0.228
M. varia							0.102	0.113
Peruvian								0.096

 

Table 6. Influence of "average favorable allele frequency" (AFAF) on correlations between hybrid forage yield and genetic distance.

Population	Mean Yield of Crosses	r	AFAF†
Peruvian	1.63	-0.58	+
Chilean	1.62	-0.47	+
African	1.50	-0.48	+
Turkistan	1.40	-0.66	+
Indian	1.35	-0.29	-
Flemish	1.35	0.09	-
M. varia	1.26	0.20	-
Ladak	1.23	0.50	-
M. falcata	1.05	0.62	-

^†AFAF value assigned based on GCA effect where GCA values greater than zero were assigned + and those less than zero were assigned -.

Table 7. Correlations between forage yield, general combining ability (GCA), midparent value (MPV), high (HPH) and midparent (MPH) heterosis, and genetic distance (GD) among 36 diallel hybrids (above diagonal) and 28 hybrids (below diagonal) of alfalfa.

Trait	Yield	GCA	MPV	HPH	MPH	GD
Yield	-	0.91**	0.78**	0.85**	0.03	-0.62**
GCA	0.87**	-	0.93**	0.74**	-0.30	-0.77**
MPV	0.63**	0.85**	-	0.60**	-0.58**	-0.89**
HPH	0.79**	0.57**	0.16	-	0.10	-0.63**
MPH	0.81**	0.49**	0.07	0.91**	-	0.67**
GD	-0.26	-0.27	-0.27	-0.26	-0.10	-

Figure 1. Phenogram of nine alfalfa germplasms based on UPGMA cluster analysis of 358 polymorphic fragments.