In the present study, the F1
population between amphimictic (94Mo13, 94Mo49 and 94Mo50) and apomictic (KaD2) diploids and the BC1
population between 94Mo49 and apomictic triploid F1
plants were used to evaluate the mode of inheritance of apomixis. Because the parental diploid plants belong to the same species, production of these populations was accomplished without concern for hybrid sterility attributable to interspecific or intergeneric crosses. There was no clear segregation of apomixis components in the F1
population either at the diploid or triploid level in this study. All of the diploid F1
plants characterized were either completely or highly eusporous and completely syngamic, while all the triploid ones were very highly diplosporous and highly parthenogenetic (). If the genes responsible for apomixis in Chinese chive are dominant, as reported in many other plant taxa (Nogler 1984b
, Noyes and Rieseberg 2000
, Ozias-Akins et al. 1998
, van Dijk et al. 1999
), no segregation of the components of apomixis would be expected to appear in the triploids, because the unreduced diploid pollen grains should have a genotype identical to that of the apomictic parent KaD2. In contrast, it is notable that neither diplospory nor parthenogenesis was observed in the diploid F1
plants. In other words, KaD2 was unable to transmit its high level of either diplospory or parthenogenesis via monoploid pollen, while it could transmit both components of apomixis via diploid pollen. These results mean that KaD2 is heterozygous for the dominant genes responsible for a high level of apomixis (both diplospory and parthenogenesis). Furthermore, severe gametophytic competition exists among monoploid pollen grains with and without the apomixis genes in KaD2.
Regardless of plant taxa, apomictic diploid plants have not yet been obtained from crossing experiments because of the lack of transmission of apomixis via monoploid gametophytes. Ozias-Akins et al. (1998)
postulated that a gametic lethal factor would be linked to apomixis genes in Pennisetum squamulatum.
interpreted the results of crossing experiments in Ranunculus
as indicating that a dominant apomixis gene expresses pleiotropic, recessive lethal activity, when the gene is carried by monoploid pollen grains. The present results in Chinese chive may be explained either by close linkage of apomixis genes in KaD2 to recessive lethal factors (either as genes or segmental chromosome deficiencies) or by a pleiotropic, recessive lethal effect of the genes.
Gametophytic competition also seemed to exist between diploid and monoploid pollen grains from KaD2. Although the percentage of pollen mother cells showing endoreduplication was only 9% in KaD2 (Kojima and Nagato 1997
), the ratio of 3× to 2× F1
plants was as high as 7 : 1 (). To explain this segregation distortion by only a lethal effect associated with apomixis genes, more than six such loci should be assumed:
[frequency of endoreduplicational pollen mother cells] : ([frequency of meiotic pollen mother cells] × [expected rate of monoploid microspores carrying non-apomixis alleles at 6 loci]) = 0.09 : ([1 − 0.09] × [0.5]6) < observed rate of 3× to 2× F1 plants (7 : 1).
This assumption is however not valid because the analysis of BC1
offspring indicates that there are two major genes responsible for apomixis in Chinese chive as described below. Instead, it is likely that the poor viability of monoploid pollen grains is caused not only by some lethal factors associated with apomixis genes but also by other deleterious recessive mutations accumulated in a heterozygous state throughout the genome of apomicts, by the mechanism known as Muller’s ratchet (1932
). Such accumulation of deleterious mutations is also inferred from the fact that pollen grains with 13 or fewer chromosomes inherited from triploid apomictic F1
plants had a distinct competitive disadvantage compared to those with 14 to 16 chromosomes.
In the BC1
offspring, clear segregation was observed between the two components of apomixis, diplospory and parthenogenesis (). The widely separated bi-modal distributions of the degree of diplospory and parthenogenesis indicate that each apomixis component is regulated by one or a very few genes with a strong dominant effect. presents a clearer view of this model. Of the 52 BC1
plants characterized as highly diplosporous, 28 were highly parthenogenetic and the remaining 24 were completely syngamic. This result supports the view that a single dominant gene P
(parthenogenesis) plays a key role in parthenogenesis in Chinese chive (Nakazawa et al. 2006
), suggests that the P
gene is unlinked to the diplospory gene, and confirms that KaD2 is heterozygous for the P
gene. Similarly, 24 of the 49 completely syngamic BC1
plants were highly diplosporous and the other 25 were completely eusporous. This suggests that diplospory is probably controlled to a large extent by a single dominant gene D
(diplospory) in Chinese chive and that KaD2 is also heterozygous for the D
gene. It is therefore likely that the apomictic diploid KaD2, the Mongolian amphimictic diploid accessions and the apomictic triploid F1
plants have the genotypes DdPp
However, the segregation of apomixis components was apparently still distorted in the BC1 offspring. No BC1 plants exhibited eusporous parthenogenesis; all of the parthenogenetic BC1 plants subjected to the progeny test were diplosporous (). It is unlikely that dPp and ddPp male gametes from the triploid F1 plants are all lethal, because: 1) d and dd male gametes are unlikely to be lethal, as half of the syngamic BC1 plants were eusporous; 2) even if P has a pleiotropic recessive lethal effect, Pp male gametes may be highly competitive, as more than half of the diplosporous BC1 plants were parthenogenetic; and 3) the D locus does not appear to be linked to the P locus. Hence, a significant number of plants with the either ddPpp or dddPpp genotype must have been included in the BC1 population. The phenotype of such plants might be expressed as eusporous parthenogenetic with seed abortion or seedling lethality or as eusporous syngamic. The former explanation is, however, unlikely. Of the 36 BC1 plants that produced very few seeds or seedlings that could be evaluated only for the degree of parthenogenesis, only 18 showed a comparatively parthenogenetic phenotype upon observation of cleared ovules. If the Ppp genotype had expressed the parthenogenetic phenotype, even in combination with dd or ddd, BC1 plants with the ddPpp or dddPpp genotype could have comprised just a subgroup of these 18 plants. Instead, it seems more probable that the phenotype of ddPpp and dddPpp plants is expressed as eusporous syngamy. Some of the 25 eusporous syngamic BC1 plants and some of the 18 non-parthenogenetic plants with few seeds or progeny likely had the genotype ddPpp or dddPpp. The absence of phenotypically eusporous-parthenogenetic plants in the BC1 population can be explained by assuming that the presence of diplospory gene is a prerequisite for, or epistatic to, the expression of genes controlling parthenogenesis in Chinese chive. This means that the presence or absence of the P gene cannot be determined in eusporous offspring by the ovule-clearing method.
The apomictic diploid parent KaD2 originates from a tetraploid variety ‘Kaohsiung’ (Kojima and Nagato 1997
) through reduced embryo sac formation followed by parthenogenesis. The ‘Kaohsiung’ was assumed to be heterozygous at both D
loci. Because apomixis of Chinese chive is facultative (Kojima and Nagato 1991
), reduced embryo sacs as well as unreduced ones can be formed in apomictic plants. In the BC1
progeny, 28 plants showed the diplosporous parthenogenetic phenotype (). In these plants, not only unreduced female gametes but also reduced ones may be formed and some of the reduced female gametes except for haploid ones should have either or both apomixis genes, D
as does ‘Kaohsiung’. In contrast, in the 25 BC1
plants showing the eusporous syngamic phenotype, only reduced female gametes were formed and none show parthenogenesis. As mentioned above, plants with the genotype ddPpp
were eusporous syngamic. Therefore, it is inferred that the D
gene has some genetic function promoting the expression of the P
gene in female gametes in addition to inducing unreduced embryo sac formation.
The modes of inheritance of diplospory and parthenogenesis proposed here for Chinese chive are the same as those reported for Erigeron annuus.
By analyzing a segregating population of 130 F1
plants from a cross between E. strigosus
, a eusporous syngamic diploid species and E. annuus
, a diplosporous parthenogenetic triploid species, Noyes and Rieseberg (2000)
constructed an amplified fragment length polymorphism (AFLP) map of loci controlling diplospory and parthenogenesis. They concluded that genes controlling diplospory and parthenogenesis are unlinked, and that each component is influenced mainly by a single dominant gene. In addition, the gene controlling parthenogenesis was suggested to be silent in the absence of the gene controlling diplospory. Noyes (2006)
confirmed this hypothesis by reconstructing diplosporous parthenogenesis in the progenies from crosses between two highly diplosporous but non-parthenogenetic triploids lacking any of the four AFLP markers linked to parthenogenesis and a eusporous, phenotypically non-parthenogenetic hyper-diploid carrying all the four parthenogenesis-linked markers. Recently, Koltunow et al. (2011)
investigated the role of two individual dominant loci in Hieracium
, loss of apomeiosis (LOA) and loss of parthenogenesis (LOP), using mutants that had lost function of one or both loci. They found that loss of function in either LOA or LOP caused partial reversion to sexual reproduction, while loss of both LOA and LOP functions led to complete reversion to sexual production and concluded that sexual reproduction was the default reproductive mode, upon which apomixis was superimposed. Although mode of apomixis differs between Allium
(diplosporous) and Hieracium
(aposporous), apomixis might have evolved in these two species in a similar manner.
As demonstrated for the Antennaria
-type diplosporous apomixis in Erigeron
(Noyes and Rieseberg 2000
), the segregating heteroploid population produced in the present study will be useful for genetic mapping of genes controlling diplospory and parthenogenesis for Allium
-type diplosporous apomixis. Moreover, some hypotriploid apomicts and segregants in this population can be employed in future efforts to reconstruct apomictic diploids free from lethal factors. Such diploids, though not yet obtained in any plant species, will provide an elegant basis for further genetic studies of apomixis.