†Department of Crop Science, University of Padova, Agripolis, Via Romea 16,

35020 Legnaro, Padova, Italy (e-mail: gbarcac@agripolis.unipd.it).

‡Institute of Plant Breeding, University of Perugia, Borgo XX Giugno 74,

06100 Perugia, Italy (e-mail: veronesi@unipg.it).

Alfalfa has a dynamic reproductive system. Whereas the predominating gametes are mostly normal, restitutional 2n gametes (i.e. gametes with the somatic chromosome number) occur at a low frequency in the vast majority of diploid alfalfa plants. The 2n gametes are usually functional in fertilization events that involve tetraploid forms of cultivated alfalfa. The widespread occurrence of 2n gametes in the Medicago sativa-coerulea-falcata complex provided early support for the concept that gene flow from diploid to tetraploid occurs continuously and naturally via 2n pollen. It also suggested that gene exchange from tetraploid to diploid alfalfa via 2n eggs had potential use in assisted breeding programs. Investigations on the reproductive system provided insights into the types of meiotic alteration responsible for the production of 2n gametes. Research on wild diploid relatives of cultivated alfalfa (M. sativa ssp. sativa L., 2n=4x=32) revealed that the formation of 2n gametes is due to cytological alterations genetically equivalent to first division restitution (FDR) and second division restitution (SDR). In particular, 2n pollen formation was shown to be due to the disorientation of spindles at metaphase II or abnormal cytokinesis, whereas 2n egg production was mainly associated with the absence of cytokinesis after telophase II, but the omission of the first and second meiotic divisions was also documented. For breeding purposes on alfalfa, theoretical data showed that 2n gametes of the FDR type are more advantageous than those obtained by SDR mechanisms for transferring parental heterozygosity and retaining epistatic interactions. The use of diploid meiotic mutants that produce 2n gametes is now recognized as one of the main methods available for exploiting heterosis and introgressing wild germplasm traits in cultivated alfalfa via 4x-2x crosses and reciprocals. The high levels of heterosis in terms of forage yield in tetraploid alfalfa hybrids produced through unreduced gametes was the impetus that propelled efforts to select diploid 2n pollen and 2n egg producers for use in sexual polyploidization schemes. Results obtained in alfalfa demonstrated that 2n pollen and 2n egg producers can be effectively used for carrying out unilateral (USP) and bilateral sexual polyploidization (BSP). They also suggested that plants which produce 2n eggs and 2n pollen could be used for direct gene exchange from wild diploid relatives into cultivated alfalfa by means of 2x-4x and 4x-2x crosses. Although experimental data have shown that forage yield improvement is attainable when plants are sexually tetraploidized, problems related to plant fertility remain unexplained. Despite the lack of data on seed production, the well-documented positive effect of sexual tetraploidization on forage yield seems to be accompanied by a worsening of fertility characteristics. In fact, scaling up the ploidy level by means of 2n gamete led to USP plants with low fertility while the BSP process yielded virtually self-sterile and highly cross-sterile plants. It is likely that as alfalfa breeding programs become more sophisticated, future varieties may be genetically uniform because of the selection of inbreds for the constitution of hybrids or the discover and utilization of apomixis. Apomictic reproduction has the potential of cloning plants through seed and it provides a unique opportunity for developing superior tetraploid cultivars with permanently fixed heterosis. Apomixis as a whole has not been detected in the M. sativa-coerulea-falcata complex, but features of apomixis have been documented. The formation of unreduced eggs through diplospory has been found in a diploid alfalfa mutant, PG-F9, while the induction of haploid parthenogenesis in tetraploid alfalfa has been widely exploited to obtain CADL populations. A main goal in alfalfa breeding could be that of combining the components of functional apomixis in cultivated alfalfa stocks. Because apomixis was mainly associated with hybridity and polyploidy, the diplosporic mutation was introgressed at the tetraploid level in order to eventually induce somatic parthenogenesis through wide-crosses with unrelated diploid materials. Although certain progeny plants displayed matromorphy, morphological and molecular progeny tests showed that in the vast majority of DTA plants diplospory is not naturally associated with parthenogenesis. In the future, the efficiency of alfalfa breeding programs based on the use of reproductive mutants could be improved by direct selection at the genotype level assisted by molecular markers, such as RFLPs and PCR-based markers. Suitable DNA markers and detailed linkage maps of alfalfa mutants should help the discover of apomictic mutants and address basic genetic questions such as the extent of genomic recombination in polyploid hybrids and the effect of sexual polyploidization on heterosis. Molecular markers were recently used in alfalfa for studying the inheritance of 2n gamete formation and identifying polymorphisms associated to genes involved in meiotic abnormalities. Molecular tagging of 2n egg and 2n pollen formation should not only explain the genetic control and regulation of these traits, but may also be an essential forward step to marker-assisted selection of 2n gamete producers and highly-efficient realization of USP and BSP breeding schemes.

Alfalfa has a dynamic reproductive system. Whereas normal gametes largely predominate, restitutional 2n gametes (i.e. gametes with the somatic chromosome number) often occur and are functional in fertilization events. The Medicago sativa-coerulea-falcata complex consists of several subspecies, both diploids (2n=2x=16) and tetraploids (2n=4x=32), which are interfertile (Lesins and Lesins, 1979) and characterized by the same karyotype (Quiros and Bauchan, 1988). Some M. sativa ssp. falcata accessions and M. sativa ssp. coerulea are diploids, but other M. sativa ssp. falcata accessions, as well as cultivated M. sativa ssp. sativa, are tetraploids. In this complex, combinations of n and 2n gametes generate new sexual polyploids that explain evolutionary processes allowing germplasm transfer and cultivar improvement. In fact, the formation of 2n pollen and 2n eggs in alfalfa has been investigated both for evolution studies (Stanford et al., 1972; Harlan and de Wet, 1975) and breeding programs (Bingham, 1980; Hermsen, 1984a,b; Veilleux, 1985; Veronesi et al., 1986; Barcaccia et al., 1997a).

The widespread occurrence of 2n gametes in M. sativa ssp. coerulea and ssp. falcata provided early support for the possibility of continuous naturally-occurring gene flow from diploid to tetraploid alfalfa via 2n pollen and let also envisage the potentials of gene exchange from tetraploid to diploid alfalfa via 2n eggs in man-assisted breeding programs. Distinct from the relatively rare occurrence of 2n gametes, certain plants are able to produce a high frequency of 2n gametes under genetic control. The use of diploid meiotic mutants that produce 2n gametes is now recognized as one of the main methods available for exploiting heterosis and introgressing wild germplasm traits in cultivated alfalfa (Bingham, 1980; McCoy and Rowe, 1986; Veronesi et al., 1986; Barcaccia et al., 1995).

Research on diploid alfalfa revealed that the formation of 2n gametes is due to cytological alterations genetically equivalent to first division restitution (FDR) and second division restitution (SDR) (Vorsa and Bingham, 1979; McCoy, 1982; Pfeiffer and Bingham, 1983; Tavoletti et al., 1991a; Tavoletti, 1994; Barcaccia et al., 1997b). For breeding purposes on polysomic polyploids such as cultivated alfalfa, 2n gametes of the FDR type are considered more advantageous than those obtained by SDR mechanisms for transferring parental heterozygosity and retaining epistatic interactions in progeny by means of sexual polyploidization (Bingham, 1980. Sexual polyploidization is unilateral (USP) or bilateral (BSP) when one or both 2n gametes contribute to the somatic chromosome number, respectively.

Investigation of the reproductive system had initially to provide information on the types of meiotic alteration responsible for the production of 2n pollen grains and 2n eggs. It was, therefore, important the availability of specific cytoembryological methodologies, such as those based on stain-clearing techniques, to analyze all steps of sporogenesis and gametogenesis within both anthers and ovules (Stelly et al., 1984; Tavoletti et al., 1991a, Barcaccia et al., 1996).

The main cytological mechanisms responsible for 2n pollen and 2n egg formation in alfalfa are schematically illustrated in Figure 1.

Early investigations on 2x M. sativa ssp. falcata-4x M. sativa ssp. sativa crosses and reciprocal showed that fertile tetraploid hybrid progenies, rather than triploids, could be obtained (Ledingham, 1940; Nilan, 1951; Armstrong, 1954; Cleveland and Stanford, 1959; McLennan et al., 1966; Bingham, 1968). That 2n gamete production could have been involved in the exchange of germplasm from diploid relatives to cultivated alfalfa was reported by Nilan (1951) and by Graber (1953). A low frequency of both 2n eggs and 2n pollen occurred, and was documented for the first time, in a 2x haploid plant of cultivated alfalfa by Clement and Stanford (1961). The abnormal cytokinesis after telophase II of microsporogenesis was found to be responsible for 2n pollen production. The formation of 2n gametes was later documented in 2x-4x and 4x-2x crosses carried out to yield 4x hybrids from interploid matings (Stanford et al., 1972). Four diploid clones of alfalfa, able to set seeds when used as male parents in 4x-2x crosses, were identified among 2x haploids of cultivated 4x alfalfa (Bingham and McCoy, 1979) and then analyzed by Vorsa and Bingham (1979). Fertility in 4x-2x crosses was found to be due to the production of 2n pollen from the diploid parent. The cytological mechanism of 2n pollen formation was the disorientation of spindles at metaphase II of microsporogenesis in up to 38% of pollen mother cells. Disoriented spindles were basically parallel to each other and resulted in formation of dyads and occasionally triads as a consequence of abnormal cytokinesis. Dyads developed into 2n pollen grains having a genetic constitution similar to that expected after FDR. The cytological mechanism of 2n egg formation in alfalfa was studied for the first time by Pfeiffer and Bingham (1983) using the diploid clone HY6. This selection was obtained from a cross of diploids M. sativa ssp. falcata and clone W31 (Bingham and McCoy, 1979). Developmental sequences in the formation of n and 2n eggs were the same through anaphase II. After this phase, in 2n egg formation cytokinesis occurred only in the micropylar dyad, not in the chalazal one. Micropylar megaspores disintegrated leaving a functional 2n megaspore of the SDR type at the chalazal end.

Because in alfalfa is effective a triploid block, which operates in interploid matings eliminating almost all triploid embryos due to endosperm imbalances, the number of seeds produced per flower pollinated (i.e., seed set) in 2x-4x crosses and reciprocals gives a measure of the frequency of 2n gametes produced by the diploids on the male or female sides, respectively. After this finding, several 2n pollen and 2n egg producers were selected within diploid natural populations of M. sativa ssp. coerulea and M. sativa ssp. falcata on the basis of seed set values with tetraploid parental counterparts (Veronesi et al., 1986).

In addition to seed set in interploid crosses, other traits were looked for rapidly discriminate 2n from n gametes. As far as male gametes, traits associated with ploidy level like the difference in pollen germ pore number typical of potato or the distinct morphology of pollen shape typical of red clover are not clearly detectable in alfalfa. Although the continuous distribution of pollen size, a mathematical approach based on the evaluation of pollen diameter proved to be effective for the estimation of 2n pollen frequency (Veronesi et al., 1988; Tondini et al., 1993). For instance, the 2n pollen diameter of diploids is comparable to that of n pollen of tetraploids and bigger than that of n pollen of diploids (on average, 38.0 µm vs. 33.2 µm). Concerning female gametes, the nucleolus diameter appeared to be a good cell marker to discriminate restitutional 2n from normal n megaspores and megagametophytes (Tavoletti et al., 1994; Barcaccia et al., 1997b). In particular, the nucleolus diameter of diplosporic cells was on average 1.6-fold (2.40 µm vs. 1.51 µm) that of meiotic ones (Barcaccia et al., 1999a). However, only the concurrent examination of nucleolus size, integument growth and cell appearance can give a reliable estimation of 2n megaspore formation in alfalfa. Examples of frequency distribution of pollen grain diameter and megaspore nucleolus diameter in 2n pollen and 2n egg producers compared to normal diploid and tetraploid plants are given in Figure 2.

The high levels of heterosis in terms of forage yield theoretically achievable in tetraploid alfalfa hybrids produced through unreduced gametes (Bingham, 1980) prompted efforts to select diploid 2n pollen and 2n egg producers for use in sexual polyploidization schemes.

Nine mutants producing 2n pollen at high frequencies, six of M. sativa ssp. falcata and three of M. sativa ssp. coerulea, were selected after two cycles of phenotypic recurrent selection (Tavoletti et al., 1991b) started using 2n pollen producers previously found by Veronesi et al. (1988). These mutants were characterized to produce on average 43% (ranging from 14 to 83%) of 2n pollen, mainly by FDR mechanisms (Barcaccia et al., 1995; 1997b; 1998) and were then largely employed in sexual polyploidization schemes. Furthermore, an experimental population produced by E.T. Bingham at University of Wisconsin, Madison, USA, was screened for the presence of reproductive mutants producing 2n gametes. This population was obtained by crossing the diploid clone HY6, which produced a high frequency of 2n eggs, with a diploid clone of Cultivated Alfalfa at Diploid Level, CADL (Bingham and McCoy, 1979). Among the segregants of this cross-combination, five diploid clones producing 4n pollen and 2n eggs were selected (Veronesi et al., 1990; Mariani et al., 1993). An additional diploid clone producing both male and female 2n gametes was isolated by Tavoletti et al. (1991a). In this plant about 19% SDR 2n eggs and 64% FDR 2n pollen were produced as consequences of the absence of cytokinesis in the chalazal dyad after telophase II and parallel or nearly parallel spindles at metaphase II, respectively. Furthermore, a diploid plant of M. sativa ssp. falcata, named PG-F9, selected for its ability to produce high frequencies of 2n eggs (Veronesi et al., 1988), was later characterized for the occurrence of distinct pathways of megasporogenesis. This mutant has been shown to produce between 55% and 70% of 2n megaspores (Tavoletti, 1994) depending on environmental conditions of growth (Barcaccia et al., 1997b). Cytological investigations of PG-F9 ovules, conducted by means of sectioning and stain-clearing techniques, revealed that the most frequent anomalies of megasporogenesis were failure of cytokinesis at the end of meiosis (76%) and diplosporic apomeiosis (24%), which resulted in 2n functional megaspores (Barcaccia et al., 1996; 1997b). Half tetrad analysis based on multiple restriction length fragment polymorphisms confirmed that the vast majority of PG-F9 megaspore mother cells underwent SDR (94%) and revealed that a low frequency (6%) of diplosporic 2n eggs was genetically equivalent to FDR (Tavoletti et al., 1996a). The discrepancy between cytological and molecular data indicated that some 2n megaspores could not develop into functional 2n embryo sacs, but both cytological and molecular analyses have independently provided evidence that unreduced eggs, through diplosporic mechanisms, would occur in alfalfa (Tavoletti et al., 1996a; Barcaccia et al., 1997b). Consistent expression of diplospory in a specific environment later provided a unique opportunity for verifying the occurrence of parthenogenesis, and thus of apomixis, in this mutant (Barcaccia et al., 1997c; 1999a).

Germplasm has been transferred on several occasions to cultivated alfalfa from wild diploid relatives of the M. sativa-coerulea-falcata complex using 2n gametes (McCoy and Bingham, 1988; Bingham, 1990; Barcaccia et al., 1999a). M. sativa ssp. falcata rather than M. sativa ssp. coerulea has played an important role in contributing genetic diversity and specific genes in alfalfa breeding. For instance, several historically important North American alfalfa cultivars were products of natural hybridization between diploid forms of M. sativa ssp. falcata and tetraploid M. sativa ssp. sativa. This wild germplasm source has contributed winter hardiness, disease resistance and hybrid vigor to cultivated alfalfa (Barnes et al., 1977). In some cases, it is known that interploid hybridization mediated by 2n gametes occurred either naturally or assisted by man. Most of the early alfalfa cultivars with adaptation to cold North American regions (e.g., ‘Grimm’, ‘Ladak’, ‘Cossack’ and ‘Rambler’) contained M. sativa ssp. falcata genes as well as benchmark cultivars like ‘Narragansett’, ‘Iroquois’, and ‘Vernal’ that have subsequently contributed to many modern cultivars (Barnes et al., 1977; Bingham et al., 1991).

The presence in alfalfa of an effective triploid block facilitated the introgression of new traits from wild diploid germplasm into cultivated tetraploid accessions by USP via either 4x-2x crosses (Bingham, 1968; McCoy and Rowe, 1986; Motzo et al., 1994; Barcaccia et al., 1995) or 2x-4x crosses (Bingham, 1990; Barcaccia et al., 1999a). Owing to the greater difficulties encountered in obtaining tetraploids by BSP, there are few reports on alfalfa tetraploid progenies from 2x-2x crosses (Barcaccia et al., 1995b). Performances in terms of forage yield or related traits as assessed in USP and BSP progenies are summarized in Table 1.

The potential breeding value of 2n gametes from diploid forms of alfalfa was first tested by McCoy and Rowe (1986) by comparing single-cross hybrids produced via 2n=2x gametes from diploid clones versus n=2x somatic chromosome-doubled tetraploid counterparts. Forage yield comparisons of progenies demonstrated a significant yield increase of the hybrid progeny from FDR 2n gametes from the diploids over the hybrid progeny from n gametes from the chromosome doubled tetraploids. The yield gain ranged from a 12% to a 32% increase. Theoretical comparisons indicated that the 2n gametes from diploids would have 12.5 to 50% more heterozygous loci, on average, than n gametes derived from somatic doubling. Thus, the experimental results confirmed the importance of heterozygosity on alfalfa performance and also demonstrated that selected diploid 2n pollen producers can be used in 4x-2x crosses to transfer the selected genotype relatively intact at the tetraploid level. It seemed that unreduced gametes can really offer a unique method for producing highly heterotic alfalfa hybrids.

An introgression program in tetraploid cultivated alfalfa from diploid M. sativa ssp. coerulea and M. sativa ssp. falcata mediated by male 2n gametes was carried out to transfer wild germplasm traits into cultivated Italian varieties (Motzo et al., 1994). Tetraploid hybrid plants from 4x-2x crosses characterized by M. sativa ssp. sativa cytoplasm and a percentage of diploid germplasm ranging from 25 to 75% were produced. Some of these hybrid plants showed dry matter yields 100% higher than that of diploid parents and higher than the average dry matter yield of cultivated alfalfa plants.

A similar program of sexual polyploidization was carried out in alfalfa by Barcaccia et al. (1995) using plants from wild diploid materials which produced male or female 2n gametes in order to assemble and test new tetraploid genotypes. Sixteen progenies from 2x-4x and 2x-2x crosses were examined with a combination of morphological, cytological and molecular analyses. The chromosome counts revealed diploid, tetraploid and aneuploid plants (Fig. 3). Plants with B chromosomes were also detected. Morphological traits such as leaf and stem size and plant height markedly increased in both USP and BSP progenies. The leaf area of the plants was a useful characteristic for distinguishing tetraploid from diploid plants obtained by USP or BSP. Leaf shape and leaf margin were not correlated with ploidy levels. Aneuploid plants with additional chromosomes displayed obovate or elliptic leaves which differed markedly from the range of forms typical of diploid and tetraploid alfalfa plants. Results demonstrated that 2n pollen and 2n egg producers can be effectively used for carrying out BSP and that the obtaining of tetraploids from 2x-2x crosses depends on parental genotypes, 2n gamete rates being similar.

Bingham (1990) recovered tetraploidy starting from a diploid M. sativa ssp. falcata synthetic (2x WISFAL-1) through five 2x-4x backcrosses using 2n eggs of the recurrent parent and a 4x M. sativa ssp. sativa as the nonrecurrent parent. The tetraploid form, designated 4x WISFAL-1, contained 98% M. sativa ssp. falcata germplasm and 2% M. sativa ssp. sativa. 4x WISFAL-1 showed all the characteristics of 2x WISFAL-1 except that it has larger vegetative and reproductive organs. The morphology (leaf size, flower size and plant height) and productivity of 4x WISFAL-1 was further investigated by comparing it with four diploid M. sativa ssp. falcata accessions, two near-isogenic M. sativa ssp. sativa populations (W2xiso-1 and W4xiso-1) and three standard alfalfa cultivars. 4x WISFAL-1 produced nearly 30% more dry matter than 2x WISFAL-1 and also showed a faster regrowth. Only one tetraploid cultivar had a significantly higher forage yield than 4x WISFAL-1 (Bingham, pers. comm.). Because of this newly-constituted tetraploid selection can be crossed naturally with cultivated alfalfa, it can be used directly for gene transfer and cultivated germplasm improvement.

The identification of tetraploid plants is a fundamental step in this type of alfalfa breeding. As already above reported, ploidy levels can be identified on the basis of phenotypic traits, such as plant vigor and pollen size. Although these characteristics give an indication of the ploidy level of progenies, they do not permit the effective chromosome number to be ascertained. Cytological and cytometric analysis allow the effective tetraploid status of plants to be verified and the aneuploids to be detected (Barcaccia et al. 1995; 1999a). Since aneuploids are undesirable in programs of sexual polyploidization, their identification is essential. At the same time, aneuploid series could be useful for locating genes on chromosomes and for research on genic dosages (McCoy and Echt, 1992).

On the whole, experimental data have shown that forage yield improvement is attainable when plants are sexually tetraploidized, while information on plant fertility is needed. In sexually tetraploidized alfalfa populations, not only forage yield but also plant fertility is a major concern. If it is known that forage yield is determinant when sexual polyploidization schemes are used for maximizing or retaining favorable allelic and non-allelic gene interactions in the progeny, then it is equally true that plant fertility is essential when sexually tetraploidized plants are employed as bridges for introgressing valuable traits from wild relatives into cultivated germplasm or for developing new cultivars.

In this area, a study was recently conducted with the aim of evaluating the fertility of tetraploid plants obtained from diploid mutants that produced 2n gametes via BSP and USP schemes (Barcaccia et al., 1998). Controlled matings between selected plants from BSP and USP were carried out according to a complete diallel design. The level of male and female fertility of each plant was estimated within full-sib, half-sib and non-inbred crosses. Crosses with unrelated self-fertile and male-sterile testers were also performed. Results are reported in Table 2. Cross-fertility was generally much higher for USP than BSP plants (on average, 0.3 vs. 0.03 seeds per flower pollinated). Both male and female fertility were inversely related to the inbreeding level of cross combinations. Female fertility was restored in both BSP and USP groups when plants were crossed with unrelated tetraploid testers. Male fertility also increased in USP plants but remained rather low in the BSP plants. Scaling up the ploidy level by means of 2n gamete resulted in tetraploid plants with low fertility. In particular, the BSP process yielded virtually self-sterile and highly cross-sterile plants.

In conclusion, the well documented positive effect of sexual tetraploidization on forage yield does not seem to be active for fertility characters. Thus, the low fertility of sexually tetraploidized plants makes it difficult at the moment to envisage their direct use in cultivation. However, since fertility is enhanced in crosses with unrelated standard tetraploid testers, their direct use for the introgression of wild traits into cultivated alfalfa seems feasible (Barcaccia et al., 1998; 1999a).

It is likely that as alfalfa breeding programs become more sophisticated, future varieties may well be hybrids or hybrid derivatives. If successful pollination control systems can be implemented there are several places where 2n gametes my have utility. A breeding scheme involving the union of 2n eggs and 2n pollen from different backgrounds was reported by McCoy (1992). It is based on the production of tetraploid hybrids from crossing FDR 2n egg-producing seed parents with FDR 2n pollen-producing pollen parents. A possible alternative could include the derivation of inbreds or partial inbred lines to be used as 2n female and 2n male gamete producers. This would result in highly heterotic tetraploid double hybrids obtained by crossing 2n gamete-producing diploid simple hybrids.

The detection of numerous recessive mutant genes which affect the pre-meiotic, meiotic and post-meiotic course of events, is evidence that sporogenesis is conditioned by a large number of genes, the majority of which are present in a dominant state (Kaul and Murthy, 1985). Mutants in which 2n female and male gamete formation is controlled by one, or a few, recessive major genes are known. Single locus mutations have been shown to determine 2n gamete production in several species, including corn (Rhoades and Dempsey, 1966), soybean (Kennel and Horner, 1985), red clover (Parrott and Smith, 1986) and potato (Mok and Peloquin, 1975; Jongedijk, 1991). In Medicago, McCoy (1982) detected a recessive gene rp for parallel spindles responsible for FDR 2n pollen production. Moreover, even the failure of the postmeiotic cytokinesis leading to the formation of jumbo (4n) pollen seems to be controlled by a single recessive gene, termed jp (McCoy and Smith, 1983). On the megasporogenesis side, Tavoletti et al. (1991a) and Calderini and Mariani (1997) hypothesized one or more recessive major genes for SDR 2n eggs.

Concerning the genetic control of 2n gamete formation, the absence of correlation between diplogynous and diplandrous gemete production in alfalfa was evidenced by Veronesi et al. (1986). A program of phenotypic recurrent selection carried out to verify the possibility of increasing the frequency of 2n gametes indicated that two different systems of major and minor genes control the production of 2n pollen and 2n eggs (Tavoletti et al., 1991b; Calderini and Mariani, 1997). Molecular tagging of 2n egg and 2n pollen formation can help explaining the genetic control of these traits and appears to be an essential step towards marker-assisted breeding and map-based cloning strategies aimed at manipulating reproductive mutants of alfalfa.

Genome mapping in alfalfa would help to elucidate the genetic inheritance and expression of genes involved in gamete formation and plant reproduction. For this purpose two linkage maps of the 2n egg mutant PG-F9 were constructed using a pseudo-testcross mapping strategy. The first includes 50 RFLP loci arranged in 8 major linkage groups (Tavoletti et al., 1996b) and the second 64 PCR (29 AFLP and 35 RAPD) and 3 RFLP loci (Barcaccia et al., 1999b). A map of the 4n pollen mutant H-25 was recently developed by Tavoletti et al. (Tavoletti, this Meeting) using RFLP and AFLP markers. This work led to the identification of genomic regions with a significant effect on multinucleated pollen formation in diploid alfalfa.

The inheritance of 2n eggs is actually investigated by our research team using three F₁ progenies obtained by crossing the diploid mutant PG-F9 with three plants of M. sativa ssp. coerulea. The genetic capacity for 2n egg production was assessed by hand-pollinating the F₁ progenies with tetraploid plants of M. sativa ssp. sativa. Seed set in 2x-4x crosses was used to discriminate n and 2n egg producers. F₁ plants that exhibited null or very low seed sets were classified as normal egg producers and plants with high seed sets as 2n egg producers. A bulked segregant analysis with RAPD, ISSR and AFLP markers was employed to identify a genetic linkage group related to the 2n egg trait using a two-ways pseudo-testcross strategy. This approach enabled us to detect a paternal ISSR marker of 610 bp, generated by primer (CA)₈-GC, located 9.8 cM from a major gene, name tne (two-n eggs), that putatively controls 2n eggs and a 30 % recombination genomic window surrounding the target locus. Seven additional RAPD and AFLP markers of maternal origin significantly co-segregated with the trait under investigation. The minimum number of quantitative trait loci controlling seed set in 2x-4x crosses was estimated by ANOVA and regression analysis. Four maternal and three paternal independent molecular markers significantly affected the trait. A map of the PG-F9 chromosome regions carrying the minor genes that determine the expression level of 2n eggs has been constructed using selected RAPD and AFLP markers. Two of these genes are linked to previously mapped RFLP loci belonging to groups 1 and 8. Molecular and genetic evidences support the involvement of one major gene and at least four minor genes (Barcaccia et al., unpublished results).

In cultivated alfalfa maximum heterosis may be expressed by a few elite individuals of the population but not by the entire population (Bingham, 1980). An approach to perpetuate elite individuals and preserve heterosis over generations could be the use of apomixis. Apomictic reproduction consists of two essential processes: apomeiosis (formation of embryo sacs without meiotic reduction) and parthenogenesis (development of embryos without fertilization) (Nogler, 1984). Thus, apomixis has the potential of cloning plants through seed and it provides a unique opportunity for developing superior tetraploid cultivars with permanently fixed heterosis.

Apomixis as a whole has not been detected in the genus Medicago, but components of apomixis have been documented. The formation of unreduced eggs through diplosporic meiosis in PG-F9 is an extremely interesting feature of apomixis (Tavoletti, 1994; Tavoletti et al., 1996a; Barcaccia et al., 1996, 1997b), as is the induction of haploid parthenogenesis in tetraploid lines of alfalfa (Bingham, 1971), which has been widely exploited to obtain populations of CADL (Bingham and McCoy, 1979).

A main goal in alfalfa breeding could be that of combining the components of functional apomixis in cultivated alfalfa stocks. Therefore, the introgression of the diplosporic apomeiosis mutation at the tetraploid level was attempted to get a new opportunity of eventually inducing somatic parthenogenesis through wide-crosses with unrelated diploid materials.

BSP schemes were adopted for introgressing the diplosporic mutation at the tetraploid level by crossing PG-F9 (employed as 2n egg donor) with 2n pollen producers of M. sativa ssp. coerulea (Barcaccia et al., 1998). Tetraploidized F₁ plants from BSP were then backcrossed (employed as n=2x pollen donor) to PG-F9 and the BC₁ progenies screened for level of ploidy, occurrence of diplospory and fertility in controlled matings with tetraploid testers of M. sativa ssp. sativa. Nine tetraploid BC₁ plants out of 15 recovered showed a degree of diplospory ranging from 5.04 to 40.96% (Table 3). These plants, named Diplosporic Tetraploidized Alfalfa (DTA), set a few seeds in wide-crosses performed with unrelated diploid testers evidencing a null or very low capacity for parthenogenesis (Barcaccia et al., 1999b).

In case asexual reproduction sensu lato took place in these diplosporic stocks, the realized degree of apomixis is lower than 1% and not useful for breeding purposes. The fact that reduced and unreduced eggs are produced by the same plant suggests that there might be a timing for the expression of the apomeiotic mutation during megasporogenesis that directs events towards either the normal or the displosporic pathway. Up to now, apomixis is believed to be controlled by a single, or a few tightly linked, dominant allele/s, that is/are responsible both for apomeiosis and parthenogenesis, thus directing the whole apomictic process (Grimanelli et al., 1998). The unreduced egg cell should have a built-in tendency to autonomous development if the mutation occurred in PG-F9 involves the genetic factor responsible for gametophytic apomixis; if not, a genetic independence for the two processes has to be supposed and more possibilities remain for sexual reproductive variation. In DTA apomeiosis does not seem to be naturally associated with parthenogenesis. The overall segregation ratio between BC₁ plants showing null or very low degree of diplospory and plants scoring a moderate or high d