Ploidy Manipulation for Seedless Cultivar Breeding

The selection of triploid lines has been, and remains a very interesting way to develop seedless cultivars. Indeed triploidy is gener- ally associated with both male and female sterility. Thus, most of the trees under fi eld evaluation present these characters, and an effi cient selection can be done for other traits. Moreover, the larger fruit size associ- ated with triploidy should correct the reduc- tion of fruit size generally observed in seedless mutants of seedy varieties. The more famous citrus triploid is ‘Tahiti lime’. This natural triploid produces large seedless fruits while its diploid kin ‘Mexican lime’ produces small seedy fruits. Such pro- grammes are being conducted in several

countries (Starrantino, 1992; Khan et al., 1996; Ollitrault et al., 1998a; Grosser et al., 2000; Guo et al., 2000; Chandler et al., 2001; Navarro et al., 2004; Froelicher et al., 2005; Handaji et al., 2005; Wakana et al., 2005; Wu et al., 2005), and new triploid cultivars have been recently released in Italy (Starrantino, 1999; Russo et al., 2004), in California (Anonymous, 2002) and in Japan (Yoshida et al., 2003; Tokunaga et al., 2005).

Natural polyploids

Polyploid germplasm

Diploidy is the general rule in Citrus and its related genera with the basic chromosome number x = 9 (Krug, 1943). However, some polyploid genotypes were detected early on in citrus germplasm. Longley (1925) was the fi rst formally to identify a tetraploid wild form: the Hong Kong kumquat (Fortunella Hindsii Swing.). Triploid Tahiti lime (Jackson and Sherman, 1975), tetraploid strains of Poncirus trifoliata (Iwasaki, 1943), allotetraploid Clausena excavata Burm. F. (Froelicher et al., 2000), tetraploid Clausena harmandiana and hexaploid Glycosmis pentaphylla are other examples of natural polyploidy found in the germplasm of the Aurantioideae sub- family (Froelicher, 1999).

Mechanisms of formation of polyploid citrus

TRIPLOIDS. The occurrence of spontaneous triploid seedlings was reported many years ago. Lapin (1937) mentioned about 4% triploidy in hybrid seedlings of Citrus limon crossed with eight other diploid species and varieties. Frost and Soost (1968) reported that more than 5% of 1200 hybrids from diploid parents grown at Riverside were probably triploids, of which 20 were proved by chromosome counts.

Studies have shown that most of the spontaneous triploids arising from diploid parents are found in small and abnormal seeds of monoembryonic female parents (Esen and Soost, 1971, 1973). The rates of

spontaneous triploid formation vary among cultivars. Rates of 1 and 6–7% for clemen- tine and King mandarin, respectively, have been observed in California and Sicily (Geraci et al., 1975). ‘Wilking’ mandarin also shows a high rate (14.6%) of triploid seedlings (Soost, 1987).

Triploid embryos have also been observed in immature seeds of polyembry- onic cultivars, with frequencies between 0 and 8% in C. deliciosa clones (cv ‘Tardiva de Ciaculli’ and ‘Avana’) and between 0 and 11.5% in different C. limon cultivars (Geraci, 1978). The rate of triploid embryos is very high in several sweet orange cultivars. Oiyama et al. (1980) found between 26 and 30% small seeds and between 8 and 33% triploid seedlings in sweet orange cultivars and intraspecifi c hybrids of sweet orange. A large percentage of triploid embryos has also been found in presumed interspecifi c hybrids of sweet orange such as ‘Ortanique’ tangor (25%) (Wakana et al., 1981), Temple tangor (6.8%) and ‘Sugeka’ orangelo (23%) (Esen and Soost, 1971). Therefore, it appears that there is genetic control of triploid embryo formation as well as environmental factors because the same cultivar may pro- duce different numbers of triploid hybrids in different years or different areas of produc- tion. Colder conditions seems to be favourable for triploid induction (our unpub- lished data). With the exception of the report concerning sweet orange (Oiyama et al., 1980), it appears that the nucellar embryos present in the seeds are more developed than the triploid ones and considerably limit the probability of triploid seedling recovery from polyembryonic species (Wakana et al., 1981). Cytogenetic studies showed that triploid embryos are associated with penta- ploid endosperm, which is a strong indica- tion that triploid hybrids result from the fertilization of unreduced ovules by normal haploid pollen (Esen and Soost, 1977; Esen et al., 1979; Wakana et al., 1981). This ploidy ratio of 5/3 between endosperm and zygotic embryos is generally considered by citrus breeders as the origin of precocious abortion of endosperm development and subsequent overdevelopment of embryos

found in small citrus seeds. It should also occur for monoembryonic and polyembry- onic cultivars where the triploid embryo is associated with diploid nucellar embryos (Wakana et al., 1981). However, another hypothesis should be considered: the endosperm balance number (EBN) theory proposed in the early 1980s (Johnston and Hanneman, 1980) to explain the basis of normal development of seeds in interspe- cifi c and interploidy crosses in the potato. According to this hypothesis, each species has a genome-specific effective ploidy (called the EBN) which must be in a 2/1 maternal to paternal ratio in the endosperm for normal development. The EBN, which can be different from the actual ploidy level, functions in an additive way during ploidy manipulation. This hypothesis, ini- tially developed for potato, has also been applied in several other plants species such as Trifolium, Lycopersicum, Avena, Datura and Impatiens (Carputo et al., 1999).

The frequency of occurrence of duplica- tion in the female gametes varies between less than 1% and more than 20% (Soost, 1987; Iwamasa and Nito, 1988) and it is sug- gested to be due to the abortion of the second meiotic division in the megaspore (Esen et al., 1979). This hypothesis is con- firmed for Clementine by a molecular marker analysis showing that less than 50% of maternal heterozygosity is transmitted by the 2 n ovules (Luro et al., 2000, 2004). However, recently Chen et al. (2007) pro- posed that 2N gametes of sweet orange result from fi rst division restitution. Very rare events of formation of triploid hybrids by fertilization of a haploid ovule by diploid pollen have also been demonstrated (Luro et al., 2000).

TETRAPLOIDS. Tetraploidization seems to occur frequently in Citrus polyembryonic genotypes. Cameron and Frost (1968) men- tion from the Riverside (California) experi- ments, that 2.5% of 3600 nucellar progeny from a broad range of genotypes were tetraploid. In Russia, Lapin (1937) also found tetraploid seedlings among eight Citrus species (from < 1% to 5.6%) and

Poncirus (4%). Russo and Torrisi (1951a) detected the tetraploid forms of C. auran- tium and C. limon among nucellar seedlings. For the rootstocks citrange Troyer and Carrizo, respective frequencies of 3 and 2.5% of tetraploid seedlings were found from a three year experiment (Hutchison and Barrett, 1981).

Chromosome doubling of nucellar stock seems to be the general rule for the generation of such tetraploid seedlings (Cameron and Frost, 1968). Indeed these tetraploid seedlings are homogenous and do not display traits of the pollen parents in controlled crosses. Isoenzymic studies of tetraploid seedlings of C. volkameriana have proved this (Ollitrault and Jacquemond, 1995). It seems that this dou- bling occurs repeatedly in the ovule tissues because tetraploid seedlings are found in fruit where seeds are mainly diploids (Cameron and Frost, 1968). Moreover, diploid seedlings can arise from the same seeds as the tetraploid seedlings (Hutchison and Barrett, 1981). From a systematic search of autotetraploids in a wide range of taxa, Barrett and Hutchison (1978) postu- lated that the ability to produce such autotetraploid seedlings is a variable genetic trait present in polyembryonic Citrus and relatives. They also propose that the rates of tetraploid seedlings are affected by environmental conditions. This asser- tion is clearly demonstrated by Hutchison and Barrett (1981) who showed that the rates of tetraploid seedlings of Troyer and Carrizo citranges vary among years and position of the fruit on the tree. In a recent work, Luis Navarro from IVIA (Spain) also found differences in the numbers of tetraploids in seedlings of several root- stocks coming from different places throughout the world, suggesting that colder conditions are favourable for sponta- neous tetraploidization in nucellar tissues (L. Navarro, personal comminication). Such an effect of cold on polyploidization events seems to be a general rule in both plants and animals (Ramsey and Schemske, 1998; Otto and Whitton, 2000).

Cases of chromosome doubling in

somatic tissue have been reported by Raghuvanshi (1962). If it occurs in meris- tem, it should lead to chimeric shoots and branches. However, very few tetraploid budsports have been identifi ed (Iwamasa and Nito, 1988) and these authors suggest that this is caused by unfavourable compe- tition between diploid and tetraploid cells in the meristem.

The formation of fully developed tetraploid seeds from diploid female ´ tetraploid male hybridization has been described (Tachikawa et al., 1961; Cameron and Soost, 1968; Esen and Soost, 1972). The authors have proposed that they originate from unreduced ovules fertilized by diploid pollen, leading to a good 3/2 ratio between endosperm and embryo. Considering the EBN theory, this also leads to an adequate 2/1 ratio of female EBN/male EBN in the endosperm and then normal endosperm and seed development.

In conclusion, two basic mechanisms are involved in spontaneous polyploidiza- tion in citrus:

● duplication of chromosome stocks in nucellar tissues that gave rise to autote- traploids

● 2 n gametes arising mostly from second division restitution during meiosis of the megaspore that produce triploid hybrids in diploid ´ diploid hybridization and allotetraploids in diploid ´ tetraploid hybridization.

Even if the rates of these natural poly- ploidization events are very high in Citrus compared with those proposed as general rules for plants by Ramsey and Schemske (1998) (autotetraploid plants formed at a rate of ~10–5 per individual per generation), it appears that polyploidy has played a minor role in citrus evolution. The lower vigour and fertility of most of the autote- traploids, and the lack of interest in them for human consumption due to thick peel and rough pulp, may have resulted in strong natural and human overselection of the autotetraploid nucellar plants. Moreover, the fact that autotetraploids are

mostly polyembryonic strongly limits the potential for evolution of the natural tetraploid gene pool. Indeed they could principally act as male parents for monoembryonic diploids, with unfavour- able competition with haploid pollen. Moreover, in the case of fertilization with diploid pollen, most of the seeds will pres- ent an abnormal development; thus a low probability of germination in natural condi- tions will result in triploids with very low fertility. For the same reason, very few triploids arising from 2 n ovules have the chance to germinate and multiply in natu- ral conditions.

Polyploid characteristics

Vegetative and fruit characteristics

An increase in cell size is the most common and universal effect of polyploidization. This change in cell volume may alter meta- bolic processes, especially those that involve membranes because of variation in the surface to volume ratio (Otto and Whitton, 2000). This could explain the slower developmental rate generally observed in some polyploids compared with diploids.

In citrus, the autotetraploid nucellar lines allow an accurate evaluation of the effect of polyploidization itself because of the homogeneity of the allelic constitution of these tetraploids and their parental diploids. Cameron and Frost (1968) have given a precise description of the autote- traploid plants cultivated at Riverside.

Tetraploid leaves are considerably broader relative to their length than diploids. They are also thicker and darker. Polyploids remain more thorny, thicker and darker than diploids. Growth is slower in tetraploids and the top is smaller and more compact. Tetraploids are slower than diploids to bloom and set fruit, and they generally produce less fruit. It appears that tetraploid grapefruits are much more vigor- ous and productive than most other autote- traploids. Tetraploid fruits are generally

smaller and less elongate than diploid fruit. In most cases, the fruit shape is irregular and the rinds are thicker and rougher with more prominent oil glands. The proportion of juice relative to whole fruit weight is much lower than for diploids, but the fl avour, acid and soluble contents are little affected. Seed number is generally lower as well as the number of embryos per seeds. An exception is a tetraploid ‘Lisbon’ lemon which produces more seeds and bigger fruit than the diploid parent. In conclusion, tree and fruit characteristics of autotetraploids are not valuable for production but they constitute an interesting germplasm for triploid breeding.

These observations are generally con- firmed by more recent studies. Barrett (1992) described an autotetraploid of Key lime having darker leaves and larger fruit with the same flavour as the parental diploid. Autotetraploid lines of P. trifoliata used as rootstock confer a reduced vigour to the tree (Jacquemond and Blondel, 1986). Autotetraploid C. volkameriana display a very depressed growth compared with diploids (Ollitrault and Jacquemond, 1995), while autotetraploid ‘Kinnow’ mandarin has shown a moderately reduced growth and interesting morphological characters (Khan et al., 1992). The authors suggested that the leaf thickness, size, breadth and the number and size of stomata should be good morphological markers to select sponta- neous tetraploids in nucellar seedlings.

The identifi cation of triploid plants by morphological traits appears much more diffi cult because they present a great vari- ability due to their zygotic origin (Cameron and Frost, 1968). Starrantino (1992) observed that triploid hybrids arising from a diploid female ´ tetraploid male cross generally present a good vigour and display the fruit level similarities to the male parent providing the diploid gametes. For exam- ple, it appears that pigmentation of blood oranges used as male parent is highly heri- table. According to Cameron and Frost (1968) and Starrantino (1992), most of the triploid citrus hybrids are nearly seedless.

Meiosis and fertility of polyploids

MEIOSIS OF TETRAPLOID CITRUS. Gametic viability is generally lower in autotetraploid geno- types having multivalent chromosome asso- ciation during meiosis than in allotetraploids forming bivalents, leading to equilibrated disomic segregation. In citrus, it has been shown that the degeneration of PMCs is more frequent in autotetraploids than in their diploid parental genotypes (Frost and Soost, 1968). These authors also observed great variability in chromosome conjugation (tetravalents, trivalents, biva- lents and univalents) during metaphase I and showed that one-third to one-half of sporads have more than the normal number of four microspores (generally six or seven). As a consequence, most autotetraploids generally produce few pollen grains with the normal chromosome complement and had a lower pollen viability than diploid parental lines. However, the remaining fer- tility is enough to allow controlled hand pollination.

MEIOSIS OF TRIPLOID CITRUS. Triploids have gen- erally been considered to be an evolution- ary dead-end because they have very low fertility and tend to produce aneuploid gametes, due to problems of chromosome pairing during meiosis. However triploids can produce haploid, diploid to triploid gametes at low rates that can lead to diploid, triploid and tetraploid progeny (Otto and Whitton, 2000). In the case of citrus, early cytogenetic studies have described triploid meiosis. Trivalent chro- mosomes and some univalents have been observed by Longley (1926). Frost and Soost (1968) described a predominance of trivalents but also the presence of numer- ous bivalents and univalents, as well as a great variation in the number of extra microspores in some genotypes. Abortion of megaspores has been observed for ‘Oroblanco’ and the LCNR46 C. limon ´ C. sinenis triploid hybrid, from the fi rst divi- sion of the embryo sac to the stage of fertil- ized egg cell (Fatta Del Bosco et al., 1992). In contrast, a low percentage of pollen abor-

tion has been observed by the same authors. As a consequence of megasporogenesis abortion, most of the triploid hybrids obtained from diploid ´ autotetraploid crosses are seedless (Cameron and Frost, 1968; Starrantino, 1992). However, some triploid seeds have been found in triploid maternal progeny (Lapin, 1937; Russo and Torrisi, 1951b), while we found half triploids and half diploids among seeds of ‘Oroblanco’ (P. Ollitrault, unpublished data). In the same study, only diploids have been found in progeny of clementine fertil- ized by ‘Oroblanco’ pollen, suggesting an unfavourable competition of diploid pollen with the haploid pollen, or the absence of diploid pollen.

Ploidy manipulation for triploid breeding; practical and theorical aspects

Several strategies have been developed for triploid citrus breeding. Some of them exploit natural events such as 2 n gametes, while the more recent strategies combine haplomethods and somatic hybridization for a direct synthesis of triploid hybrids. For each method, we will discuss briefl y their biological limitations and practical aspects as well as their implications for gene segregation and recombination.

In the citrus literature, tetraploids aris- ing from chromosome doubling (sponta- neous in nucellar tissue, or induced by colchicine) are generally called autote- traploids while tetraploid somatic hybrids obtained by protoplast fusion are consid- ered as allotetraploids. We adopt this termi- nology herein, keeping in mind that it is quite different from the general defi nition of auto- and allotetraploidy (Otto and Whitton, 2000) and that it may not be related to the disomic or tetrasomic mode of chromosomal segregation.

Selection of spontaneous triploid in 2x ´ 2x sexual crosses

BIOLOGICAL LIMITATIONS AND PRACTICAL ASPECTS.

The selection of triploid hybrids arising

from 2 n megagametophytes was described in the 1970s (Esen and Soost, 1971, 1973; Geraci et al., 1975). Triploids were sought on small seeds of monoembryonic cultivars with a high rate of diploid megagameto- phytes such as ‘Temple’, ‘Wilking’ or ‘Fortune’ (Esen and Soost, 1977; Wakana et al., 1981; Soost, 1987). However, this approach was limited by a relatively low effi ciency and the diffi culty of screening large populations of seedlings by classical cytogenetic methods of chromosome count- ing. More recently the use of in vitro embryo rescue and ploidy evaluation by fl ow cytometry has provided much greater effi ciency (Ollitrault et al., 1996b). In this way, it is possible to exploit low rates of spontaneous diploid megagametophytes such as that of clementine (1%), and this strategy (Fig. 8.2) is actually routinely used by several teams in the Mediterranean Basin to select new easy peel cultivars (France, Spain and Morroco).

Concerning polyembryonic cultivars,

Wakana et al. (1981) show that triploid zygotic embryos should be found with diploid nucellar embryos in small seeds. However, the practical possibility of select- ing these triploid embryo is greatly limited by polyembryony (Wakana et al., 1981). To avoid this problem, Geraci et al. (1977) have proposed carrying out a very early rescue of zygotic embryos from immature fruit, but it seems that selection of sponta- neous triploids from polyembryonic seedlings has not found real applications in citrus breeding.

SEGREGATION AND RECOMBINATION. Second mei-

otic division restitution (SDR) has been pro- posed for diploid megagametophyte development in clementine (Luro et al., 2004) while Chen et al. (2007) concluded first meiotic division (FDR) for sweet orange. In both cases only a part of the maternal heterozygosity is transmitted to the triploid hybrid but the structures of triploid populations are very different. The

Fig. 8.2. Scheme for selection of triploid hybrids from dilpoid ´ diploid hybridization.

rate of maternal heterozygosity transmis- sion (RHT) varies among the loci and is directly linked to the crossing over rate between the centromere and the locus (RHT

= r for SDR and 1–1/2r for FRD, where r is the crossing over frequency between the locus and the centromere). For SDR and FDR respectively, RHT varies from the cen- tromere to the telomere between 0 to 1 and

1 to 0.5 in the case of systematic single crossing over. Therefore, the homozygosity level of 2n gametes is higher for SDR than for FDR.

Segregations and recombinations occur for both the male and female parents. In the case of SDR for one locus, the maximum allelic combinations of the triploid hybrid vary between two in the case of systematic single crossing over, four for a locus very close to the centromere and six for others. In the case of FDR it is two for locus very close to the centromere and six for others.

Analysis of the origin of 2n gametes for several seed parents must be done to con- fi rm that SDR observed in clementine is generalized in the mandarin group.

Sexual crosses between diploids and autote- traploids

BIOLOGICAL LIMITATION AND PRACTICAL ASPECTS.

Most of the autotetraploids (doubled diploids) used for triploid breeding are spontaneous nucellar tetraploids from poly- embryonic cultivars and are themselves polyembryonic. Obtaining such autote- traploid seedlings should be enhanced by selecting seeds from giga sectors of chimeric fruits (Grosser et al., 1998). Relatively few studies have concerned chromosome stock duplication by colchicine treatments. Tachikawa (1971) and Barrett (1974) mentioned some autote- traploids as well as periclinal ploidy chimeras obtained by such treatments. To avoid chimera development, several authors have combined in vitro techniques and colchicine treatments. Gmitter and Ling (1991) have obtained a non-chimeric tetraploid of ‘Valencia’ sweet orange and ‘Orlando’ tangelo via somatic embryogene-

sis from culture of undeveloped ovules treated with colchicine, and Gmitter et al. (1991) selected ‘Hamlin’ and ‘Ridhe Pineapple’ tetraploid sweet oranges from embryogenic cultures treated with colchicine. Wakana et al. (2005) obtained tetraploid forms of acid citrus cultivars by top grafting of shoots with sprouting axial buds treated with colchicine.

Considering that autotetraploid selec- tion from polyembryonic genotype seedlings is now very easy using flow cytometry (even for cultivars with low rates of chromosome duplication), colchicine treatment remains interesting mostly with the objective of monoembryonic tetraploid creation. It should be applied in budwoods (because in vitro techniques involving somatic embryogenesis are mostly useful for polyembryonic genotypes). Flow cytom- etry will help greatly to select non-chimeric tetraploid shoots. Tetraploid plants have also been regenerated by somaclonal varia- tion from C. limon embryogenic callus obtained from ovule culture on medium containing 2, 4-dichlorophenoxyacetic acid (2, 4-D) (Vardi and Spiegel-Roy, 1982).

Crosses of 2 x female ´ 4 x male and 4 x

´ 2 x have been used by breeders to obtain triploid hybrids (Cameron and Soost, 1968; Esen et al., 1978; Oiyama et al., 1981; Starrantino and Recupero, 1982). It appears that 4 x ´ 2 x crosses have been more suc- cessful in producing triploid embryos (Cameron and Frost, 1968; Soost and Cameron, 1975). It was proposed that the 3/5 ratio between embryo and endosperm ploidy was more favourable than the 3/4 ratio obtained in the reciprocal crosses. However, most of the available tetraploid generations are polyembryonic. Thus, monoembryonic diploid female ´ tetraploid male hybridizations have been exploited much more than the reciprocal ones. Precocious embryo rescue, 3–4 months after anthesis has been proposed by Starrantino and Recupero (1982) to enhance the production of triploid hybrids from 2 x ´ 4 x crosses. ‘Oroblanco’ and ‘Melogold’ triploid pummelo ´ grapefruit hybrids (Soost and Cameron, 1980, 1985)

resulted from such crosses, as well as inter- esting tangor and mandarin triploid hybrids were selected in Italy (Starrantino, 1992).

Tetraploid hybrids are found in addi- tion to triploids in 2 x ´ 4 x hybridization (Cameron and Soost, 1968; Oiyama et al., 1981), while only triploids are obtained in reciprocal crosses. These tetraploids are found in normal seeds, which is in agree- ment with the spontaneous production of diploid megagametophytes and the good embryo/endosperm ploidy ratio of EBN when a diploid pollen fertilized an unre- duced megagametophyte. These tetraploid hybrids should be incorporated in the tetraploid gene pool for further triploid breeding. The occurrence of tetraploid hybrids in 2 x ´ 4 x crosses implies that the ploidy of seedling hybrids should be checked before any other evaluations.

SEGREGATION AND RECOMBINATION. Segregation

and recombination of tetraploids is com- plex (Wu et al., 2001a, b). It involves pref- erential pairing between homeologous chromosomes that defi nes the proportion of bivalents and multivalents formed (Wu et al., 2001a) and the double reduction fre- quency in the case of tetravalent formation (Mather, 1936). The latter parameter is con- stant for a specifi c locus depending on the distance to the centromere. A lot of eco- nomic Citrus species such as sweet orange and grapefruit are of interspecifi c origin. Moreover, differences in nuclear genome size exist among the three ancestral taxa of cultivated Citrus (Ollitrault et al., 1995) cer- tainly leading to structural heterozygosity in interspecifi c hybrids. It should explain the variability of chromosome association during meiosis observed in autotetraploids. Preferential pairing of duplicated chromo- somes can be relatively frequent for dou- bled interspecifi c hybrids.

The restitution of the heterozygosity (ab) of the diploid line that generates the tetraploid (aabb) will be a function of the preferential pairing, of the rate of tetrava- lent formation and of the distance of the locus to the centromere. It will vary:

● from 100% in the case of total preferen- tial pairing between duplicated chromo- somes leading to systematic bivalent formation and disomic segregation. It will be 66% in case of random bivalent for- mation and tetrasonic segregation and less in case of tetravalent associations

● to 40% for loci independent of the cen- tromere in the case of systematic tetrava- lent formation (Wu et al., 2001a).

For ‘autotetraploid’ ´ diploid hybridization, segregations and recombina- tions occur for both the male and the female parents. For one locus, the maximum potential allelic combinations of triploid hybrids varies between:

● two (one heterozygous diploid gamete ´ two allelic haploid constitutions) in the case of total preferential pairing between duplicated chromosome (disomic segre- gation)

● and six ((two homozygous + one het- erozygous diploid gamete constitution) ´ two allelic haploid constitutions).

Sexual hybridization between diploid females and tetraploid somatic hybrids

BIOLOGICAL LIMITATIONS AND PRACTICAL ASPECTS.

The creation of tetraploid somatic hybrids for triploid breeding is described in Chapter 10; a review can also be found in Grosser et al. (2000). Additional combinations have been obtained recently (Calixto et al., 2004; Grosser and Chandler, 2004; Wu et al., 2005). Good levels of pollen fertility have been found in several allotetraploid somatic hybrids (Deng et al., 1995) and some of them present an earlier fl owering than autotetraploids. Due to an unbalanced ploidy ratio, embryo rescue is systemati- cally used to recover triploid hybrids. The fi rst interploid hybridizations with somatic hybrids were reported in 1991 (Oiyama, 1991) between diploid clementine and tetraploid Trovita sweet orange + Poncirus trifoliata. Embryos of underdeveloped seeds at fruit maturity were rescued in vitro. They produced mostly triploid

hybrids and a few tetraploids (from normal seeds) resulting from the fertilization of unreduced ovules. Deng et al. (1996) observed a 74.5% pollen fertility in hybrid Hamlin Sweet orange + Rough lemon, and used it to fertilize diploid lines. Mostly abortive seeds were harvested three months after pollination for in vitro embryo rescue. Diploid, triploid and tetraploid plants have been obtained. This implies that the tetraploid somatic hybrid is able to produce haploid and diploid pollen. Similar obser- vations were made by Tusa et al. (1996) in sexual crosses between diploid Feminello lemon and three allotetraploid somatic hybrids (Valencia sweet orange + Feminello lemon, Milam lemon + Feminello lemon and Key lime + Valencia sweet orange). An equivalent number of diploid, triploid and tetraploid hybrids was obtained. The use of diploid pollen from allotetraploid hybrids has been systematized in Florida, and sev- eral hundred triploids from interploid crosses with allotetraploids are already planted in the fi eld (Grosser et al., 2000).

SEGREGATION AND RECOMBINATION. As for dou-

bled diploids, the gametic structures formed by tetraploid somatic hybrids depend on the mode of chromosome associ- ation at meiosis and on the position of the locus relative to the centromere. The main difference is that tetraploid somatic hybrids should have four different alleles (abcd) for the same locus. Therefore, a diploid gamete from an allotetraploid can display a very high heterozygosity. If the chromosomal differentiation between the two parents of the somatic hybrid is high at the interspe- cifi c level, it will lead to total preferential pairing of chromosomes from the same parent and disomic segregation. In this case, the diploid gametes will transmit interspecifi c heterozygosity. In other cases, diploid gametes should transmit either intraparental or interparental heterozygos- ity.

Recombination and segregation occur for both the female and the male parents, and this strategy leads to the greatest diver-

sity of the triploid progeny. For each locus, the maximum number of allelic structure of the triploid varies between:

● eight (four allelic structures of the diploid gamete ´ two allelic structures of the haploid gamete) in the case of an allotetraploid with disomic segregation

● and 20 (ten allelic structures of the diploid gamete ´ two allelic structures of the haploid gamete) if there is the possi- bility of formation of tetravalents (Wu et al., 2001a).

Endosperm culture of 2x ´ 2x sexual crosses

BIOLOGICAL LIMITATIONS AND PRACTICAL ASPECTS.

This technique could be a tool to overcome the barriers to sexual hybridization that result from nucellar embryony and can the- oretically be applied to all germplasm with female fertility. Successful regeneration of triploid plantlets has been reported by Wang and Chang (1978) and Gmitter et al. (1990). However, the step of shoot or embryo regeneration from endosperm calli appears to be critical (Jaskani et al., 1996), which limits the practical application of this technique for breeding purposes. Indeed it appears diffi cult to obtain large recombining populations in order to apply effi cient fi eld selection.

SEGREGATION AND RECOMBINATION. The triploid

structure of the endosperm results from the fertilization of two haploid polar nuclei of the embryo sac by a haploid vegetative nucleus of pollen. There is no restitution of maternal heterozygosity because the two polar embryos arise from the same haploid cell by mitotic division. This leads to a high level of homozygosity (in terms of a dupli- cated allele at each locus) in the triploid hybrids. Recombination and segregation occur for both the male and the female parent, and a maximum of four allelic con- stitutions should be found at each locus (two homozygous diploid ovule combina- tions ´ two allelic pollen constitutions).

Diploid + haploid protoplast fusion

BIOLOGICAL LIMITATIONS AND PRACTICAL ASPECTS.

The fi rst step is to establish haploid Citrus lines. These should be obtained by andro- genesis (Germanà, 1992) or by induced gynogenesis (Ollitrault et al., 1996a). Haplomethods in citrus are described in Chapter 7.

The technique of haploid + diploid somatic hybridization, developed simulta- neously by Ollitrault et al. (1997) and Kobayashi et al. (1997), is described in Chapter 10. It should be applied to polyem- bryonic or monoembryonic diploid culti- vars if haploid embryogenic callus lines are available. For combination with leaf proto- plasts of haploid plantlets it is necessary to use diploid protoplasts from embryogenic callus lines arising mainly from polyembry- onic cultivars. In the case of haploid callus lines, this method leads to triploid hybrids as well as tetraploid and pentaploid hybrids being obtained, due to the ploidy instability of haploid calli (Ollitrault et al., 1998b).

The great limitation of this strategy is the scarcity of haploid lines available in Citrus. This could be overcome by the development of gametosomatic hybridiza- tion.

SEGREGATION AND RECOMBINATION. This strategy

is the only one allowing a complete restitu- tion of the diploid cultivar nuclear genome and its heterozygosity, without segregation and recombination. Therefore, it will be more interesting in terms of effi ciency of multilocus selection at the diploid cultivar level to confer desirable traits to the triploid hybrid.

Segregation should occur at the level of cytoplasmic genomes when fusions are car- ried out with embryogenic callus protoplast for both diploid and haploid lines. This should lead to four different kinds of nucle- ocytoplasmic interaction.

For each nuclear locus, the maximum number of allelic combinations will be one in the case of combination with a haploid line, while it will be two with gametoso-

matic hybridization (segregation and recombination at the level of haploid cells).

Tools and protocols for ploidy manipulation in Citrus

A scheme for selection of triploids from diploids ´ diploids hybridization from using small seeds is presented in Fig. 8.2. In vitro techniques and fl ow cytometry have greatly modifi ed the potential of ploidy manipula- tion for citrus breeding. Somatic hybridiza- tion methods are described in Chapter 10, so we will focus here on two technical aspects: embryo rescue and fl ow cytometry.

Embryo rescue

The diffi culty of recovering triploid hybrids from interploid hybridization or diploid ´ diploid hybridization (due to an unbal- anced EBN or ploidy ratio between embryo and endosperm as well as polyembryony) should be overcome by in vitro embryo rescue. In vitro culture of excised fully developed embryos was described 40 years ago to obtain zygotic plantlets from poly- embryonic cultivars (Maheshwari and Ranga Swamy, 1958). Embryos should be rescued at different developmental stages, from the globular stage in undeveloped seeds of mature fruit (Starrantino and Russo, 1980) to heart-shaped embryos (Rangan et al., 1968). Starrantino and Recupero (1981) obtained triploid hybrids from 2 x ´ 4 x crosses by rescuing globular and heart-shaped embryos 3–4 months after anthesis.

Adenine sulphate (25 mg/l) and malt extract (500 mg/l) appear favourable for citrus embryo rescue. The addition of hor- mones should help embryo germination and shoot elongation. Many authors add 1 mg/l GA3 (gibberellic acid), while a balance

of 0.5 mg/l BA (6-benzylaminopurine), 0.5

mg/l kinetin and 0.11 mg/l NAA (1-naph- thaleneacetic acid) are favourable when regeneration is diffi cult (Deng et al., 1996).

As a standard culture medium, we can propose:

● Murashige and Tucker or Murashige and Skoog basic medium

● 30 g/l sucrose

● 25 mg/l adenine sulphate

● 500 mg/l malt extract

● 1 mg/l GA3

● pH 5.7

● 8 g/l agar (agar is better than other gelling agents to avoid problems of hyper- hydricity)

Seeds may be sterilized: 5 min in 70% alcohol + 10 min in sodium hypochorite solution (0.2% active chloride) and washed three times in sterilized water.

The culture may be incubated at 26/28°C, and a 16 h photoperiod is favourable.

After ploidy control, shoots can be grafted on greenhouse rootstocks. This will allow a better growth than classical acclimatization and prevents fungus prob- lems, particularly for genotypes susceptible to Phytophthora spp. After grafting, it is necessary to maintain the plants for 2–3 weeks in saturated hygroscopy to prevent drying of the shoot.

Flow cytometry

Flow cytometry which was developed for medical research is now widely used for plant DNA content analysis (Arumu- ganathan and Earle, 1991a, b). This tech- nique allows an estimation of the volume and fl uorescence of isolated cells or nuclei. If the fl uorescent probe used is DNA spe- cifi c, nuclear DNA content can be quanti- fied. A large number (104–106/min) of nuclei can be analysed rapidly, and the results are presented as a histogram of inte- gral fl uorescence. The position of the peak in the axis is proportional to the DNA con- tent of the nuclei. Generally, if young leaves are analysed, two peaks are obtained. According to the cell cycle, the fi rst peak is defi ned by nuclei of non-cycling cells (G0)

or in the pre-DNA synthesis stage (G1),

while the second peak represents nuclei in

the post-DNA synthesis stage of mitosis. It is recommended to use an internal control

(nuclei with already known DNA content or nuclei of genotypes of the same species with identifi ed ploidy) for determination of absolute genome sizes or ploidy level of the sample.

Interest in fl ow cytometry for citrus ploidy analysis and then for polyploid breeding was fi rst emphasized by Ollitrault and Michaux-Ferriere (1992). Since then, this technique has opened up new avenues for citrus ploidy manipulation. Indeed it allows the selection of rare events such as 2 n gamete formation that are diffi cult to exploit for breeding by conventional cyto- genetic techniques (Ollitrault et al., 1996b; Luis Navarro et al., 2004). Flow cytometry is now used by several teams around the world as a routine tool for citrus breeding for both sexual progeny (Tusa et al., 1996; Ollitrault et al., 1998a) and somatic hybrids (Ollitrault et al., 1996c; Grosser et al., 2000). It has also allowed genome size dif- ferentiation between cultivated diploid species (Ollitrault et al., 1995). Larger genomes are observed for C. medica (0.81 pg/2C), while C. reticulata present the smallest ones (0.74 pg/2C). This differentia- tion does not allow prediction of the exact size of polyploid hybrids arising from sexual crosses because of chromosomic seg- regation.

The technique is very simple and allows the analysis of 150–200 genotypes per day. The procedure used is modifi ed from Arumuganathan and Earle (1991a). Small pieces (50 µg) of leaf of the sample and the internal control are chopped with a razor blade in 300 µl of extraction buffer (phosphate-buffered saline (PBS) contain- ing 1 mg/ml dithiothreitol and 0.3% Triton X-100). For exact nuclear DNA content evaluation, RNase (10–3 U/ml) must be added to the extraction buffer, while this is not necessary for routine ploidy analysis. The nucleus solution is fi ltered through a 30 µm nylon mesh. Then 150 µl of the fi l- tered solution is mixed with 150 µl of extraction buffer with the addition of 200

µg/ml propidium iodide (DNA stain). After

10 min incubation, the stained nucleus solution is analysed in the fl ow cytometer.

Staining with 4’, 6-diamidino-2- phenylindole (DAPI) also allows analysis of the ploidy level.

The constitution of PBS is as follows (for 1 l): 8 g of NaCl; 0.2 g of KCl; 1.44 g of Na2HPO4; 0.24 g of KH2PO4. This is made

up to 1 l with H2O after the pH has been

adjusted to 7.4 with HCl. The solution

should be aliquoted and sterilized for long- term conservation.