Seed Reproduction: Development of Gametes and Embryos

HOWARD B. FROST and ROBERT K. SOOST

      Seed reproduction of citrus is a continuous process that begins in the flower and ends with maturity of the fruit (fig. 4-1).   All or nearly all of the flowers of some sterile varieties, including certain triploids and also certain defective flowers, such as those with abortive pistils, fall from the tree with little if any growth after the petals separate.   In fertile varieties, flowers or ovaries may fall at any time from the unopened-bud stage up to the time of definite fruit "setting," some ten or twelve weeks after the flowers open.   By then, the normal removal of surplus ovaries is complete; except for rare accidents, the remaining fruits stay firmly attached to the tree until maturity.   The setting of fruits depends on complex physiological conditions that relate to the supply of water, food materials, and hormones, and which therefore are much affected by competition among flowers and young fruits.   In many varieties, the reproductive processes of pollination, fecundation, and embryo formation are of primary importance for fruit setting.

DEVELOPMENT OF EMBRYO SAC AND FEMALE GAMETE

Carpels and Placenta
      The pistil of Citrus is composed of about ten units, the carpels, each somewhat resembling a bean pod in structural plan, which are joined to each other and to a central axis to form a compound pistil.   In the mature flower, the pistil is composed of the ovary (the young fruit) at the base, the stigma (apical knob), and usually a slender style between.   At a very early stage (fig. 4-2, C), the pistil is not closed at the top, but consists of a circular wall or ring and a protuberance within the ring (Osawa, 1912).   This wall is composed of the joined carpels, which according to Ford (1942) arise as a whorl of crescent-shaped primordia.   The carpels grow upward and their margins project inward to meet the central protuberance (fig. 4-2, D).   At the same time, the central protuberance grows upward to produce the axis or core of the fruit, and unites with the carpel margins.   Each carpel thus encloses a locule of the ovary (fig. 4-2, E).
      At the inner angle of each locule is developed the placenta, the region of thickened tissue which bears the ovules.   Biermann (1896) and Ford (1942) stated that the ovules arise from the carpellary margins, but Osawa (1912) considered the placenta to be an outgrowth from the axis.
      A stylar canal, lined with epidermal cells (Tillson and Bamford, 1938; Ford, 1942; Banerji, 1954), extends from the placenta of each carpel to a narrow slit on the surface of the stigma.   Through these canals, the pollen tubes grow downward to the ovules.   However, Banerji (1954) stated that pollen-tube growth is mainly intercellular and not through the stylar canals.

Ovules
      While the flower is still very small, the ovule arises as an outgrowth of the placenta; this outgrowth involves several cell layers.   The ovule soon begins to grow upward, and finally takes the typical anatropous (inverted) form, with the micropyle facing the axis of the ovary.   The mature ovule consists of the funiculus (stalk), a central mass of cells (the nucellus), the embryo sac within the nucellus, and the two integuments surrounding the nucellus; at the apex of the nucellus is an opening through the integuments, the micropyle.   In the anatropous ovule, the end away from the placenta is the chalaza region.

Archesporium and Embryo-Sac Mother Cell
      Before the integuments develop, one cell in the second layer, near the apex of the nucellus, is distinguishable by its greater size and its larger nucleus (fig. 4-3, A).   This is the archesporial cell.   The archesporial cell soon divides once; the outer of the two daughter cells is the tapetal cell and the inner one is the embryo-sac mother cell, or megasporocyte (fig. 4-3, B).   The tapetal cell divides, and from it are developed eight layers of cells; by this time, therefore, the embryo-sac mother cell is deeply buried in tissue near the center of the nucellus.   Figure 4-3, C shows the development of the tapetal cells and also that of the integuments partly completed.   Occasionally more than one embryo-sac mother cell is formed in an ovule.   Bacchi (1943) and Banerji (1954) reported more than one normal embryo sac in a single ovule.

Meiosis and Megaspores
      The embryo-sac mother cell grows to several times its original size and becomes elongated (fig. 4-3, D).   The chromosomes of its nucleus meanwhile pair during the prophase stage of the first division.   This division occurs before the ovule is fully developed (Bacchi, 1943).   By metaphase of the first division, the nuclear membrane disappears.   Each chromosome pair separates during anaphase and telophase, forming two groups of chromosomes.   Following a short interphase, the second division splits each chromosome in two, resulting in four equal groups of chromosomes.   After the second division, cell walls are formed, producing four cells in a row that extends longitudinally in the nucellus (fig. 4-3, G).   These four cells are the megaspores.

Embryo Sac and Egg Cell
      Only the lowermost megaspore develops further; the remaining cells degenerate (fig. 4-4, A).   The lowest cell grows longitudinally and occupies the space vacated by the others.   This cell, the functional megaspore, develops further to form the embryo sac.   It enlarges greatly, but its cytoplasm does not increase correspondingly; large vacuoles appear, and when the embryo sac is mature they occupy the largest part of the volume within the cell wall (fig. 4-4).
      As the megaspore grows, its nucleus divides, and the two daughter nuclei go to the opposite ends of the embryo sac (fig. 4-4, B).   Each nucleus divides again, so that there are four nuclei in the embryo sac (fig. 4-4, C).   Each of these four nuclei divides once more, thus producing eight nuclei; the embryo sac now contains four nuclei near each end (fig. 4-4, D).   Three nuclei remain near the basal (chalazal) end, where they organize the three antipodal cells (fig. 4-4, E).   In the micropylar end, three complete cells are also arranged.   One of these is the egg cell, and the other two are the synergids, which perhaps facilitate the process of fecundation of the egg cell (fig. 4-4, E).   At this stage, the egg is mature and ready for fecundation.
      The two remaining nuclei, one near each end of the embryo sac, are the polar nuclei.   These move toward the middle of the sac (fig. 4-4, F) and supposedly fuse to form the endosperm nucleus.   (See below).
      The corresponding stages of development are attained later in the ovary than in the anthers of the same flower, so that the microspores have begun to develop into pollen grains before the megasporocyte has passed the prophase of the first division (Osawa, 1912).   At the time of the opening of the flower (fig. 4-1), the embryo sac is usually in the eight-nucleate stage, but may be in the one-, two-, or four-nucleate stage (Bacchi, 1943).

DEVELOPMENT OF POLLEN AND MALE GAMETES

Archesporium and Pollen Mother Cells
      At an early stage in development of the anther, the archesporial cells are recognizable by their larger nuclei and different staining; as in the megasporangium, these cells are derived from the hypodermal layer (Banerji, 1954).   According to Banerji (1954), the archesporial cells divide periclinally to produce an outer layer of parietal cells and an inner layer of sporogenous cells.   Additional divisions of the parietal cells produce four cell layers.   The innermost layer forms the tapetum, and the others with the epidermis form the anther wall.
      The tapetum is mainly one-layered but may be two- or three-layered (Banerji, 1954).   It surrounds the slender cylinder of pollen mother cells or microsporocytes (fig. 4-5), which are formed by successive divisions of the primary sporogenous cells.   Before the mother cells reach mid-prophase, the nucleus of each tapetal cell divides once; later more divisions occur, so that the tapetal cells become binucleate or irregularly multinucleate.   While the development of the pollen grains is in progress, the tapetal cells disintegrate, probably supplying food materials to the pollen.

Meiosis and Microspores in Diploids
      Virtually all cultivated forms of Citrus, Fortunella, and Poncirus are diploid, and the monoploid number of chromosomes in these genera is nine (fig. 4-6).   (See Longley, 1925; Frost, 1925a; Nakamura, 1934; Ruggieri, 1935; Miedzyrzecki, 1936; Lapin, 1937; Kandelaki, 1938; Bacchi, 1940; Yarnell, 1940; Nakamura, 1941, 1942; Krug, 1943; Nakamura, 1943; Chen, 1944; Singh and Shah, 1950; Rhandhawa and Choudhury, 1960; Ozsan, 1961; Naithani and Raghuvanshi, 1962; Raghuvanshi, 1962a.)   Diploidy (n = x = 9) occurs also in Severinia, Triphasia, Citropsis, Aeglopsis (Longley, 1925); Feronia, Murraya (Toxopeus, 1933; Pathak, et al., 1949); Afraegle, Atlantia [sic], Clausena, Microcitrus (Krug, 1943); Eremocitrus, Micromelum (Yarnell, 1940).
      Before their first division, the pollen mother cells are distinguishable from the tapetal cells which surround them by their large size and possession of only one nucleus, and by differences in staining (fig. 4-5, B).   During the late prophase of the first division, the pollen mother cells round off.   The original mother-cell wall persists about in its original form until it breaks down at the end of the sporad stage, whereas the cytoplasm develops a new rounded wall, usually well inside the original wall (fig. 4-6).
      In the pollen mother cells of Citrus, as in the embryo-sac mother cells, there has been little study of the prophase of the first division.   The processes must usually proceed normally, however, since in later stages the diploid varieties studied generally have nine bivalent chromosomes (fig. 4-6).
      At diakinesis the chromosomes are widely scattered in the nucleus (fig. 4-7).   The nuclear membrane disappears prior to metaphase, and at metaphase the nine-paired chromosomes become oriented at the equator of the spindle, which has meantime formed, and parallel with its axis.   Yarnell (1938) and Nakamura (1939) reported secondary associations of the chromosomes during the beginning of first metaphase.   Naithani and Raghuvanshi (1958) and Raghuvanshi (1962b) reported multivalents, inversions, and univalents in four varieties.   Pereau-Leroy (1950) also reported univalents in the lemon variety Sanguin Panache.   Iwamasa (1966) reported a translocation in Valencia, an inversion in Mexican lime, and asynapsis in Mukaku Yuzu.
      At anaphase I, the nine bivalents usually separate normally, resulting in two daughter nuclei.   Following a short interkinesis stage, the membranes of the daughter nuclei disappear and the second division usually proceeds normally.   Usually there are two sets of nine chromosomes each at the metaphase of second division (fig. 4-7, E), but irregular distributions have been found, indicating irregular separation in the first division.   Chen (1944) and Banerji (1954) also reported secondary association of chromosomes.   At the end of the second division, the four sets of nine chromosomes each organize four nuclei within the rounded wall of the mother cell, producing the four-nucleate stage.
      Walls form between the four nuclei, and the four new cells (microspores or young pollen grains) separate and round off within the original wall of the pollen mother cell.   This is the pollen-tetrad, microspore-tetrad, or quartet stage.   Since more than four microspores may be formed by one mother cell (occasionally in diploids, and abundantly in most polyploids), these terms are often literally inappropriate, and therefore it seems preferable to use the terms sporad and microsporad (Webber, 1933).
      The chromosomes are small in bulk.   At maximum condensation in the first division, the length is commonly near two microns (1/12,000 of an inch) or less (fig. 4-6), but just before separation in early anaphase I, they may be stretched to a considerably greater length.   Yarnell (1937), Kandelaki (1938), and Sharma and Bal (1957) related these differences to various species.

Meiosis in Polyploids
      The presence of extra genomes in triploid and tetraploid forms of citrus causes certain characteristics from related diploids in tree and fruit (see chap. 5).
      In triploids, chromosome reduction cannot be regular, since there are three somatic chromosomes of each kind and the first division separates the chromosomes of each kind into two groups.
      Although Russo and Torrisi (1951b) reported allotetraploids as well as autotetraploids, most, if not all, tetraploid forms of Citrus are basically autotetraploids in regard to chromosome homology.   Therefore, meiosis in tetraploid Citrus is irregular, due to multivalent conjugation.
      Triploid forms of Citrus seem to originate from diploids only as gametic seedlings; tetraploids originate either as nucellar seedlings or in stem growing points (see chap. 5).

Meiosis in Tetraploid Forms
      Pollen mother cells and sporads from tetraploid forms of twelve varieties belonging to four species of Citrus, have been examined at the University of California Citrus Research Center, Riverside.   With some varieties at least, the pollen mother cells of tetraploids degenerate before the reduction division much more often than do those of corresponding diploids.   With some varieties, complete counts were difficult to obtain; however, all forms classed as tetraploid had a chromosome number of thirty-six, or at least near thirty-six.   Lapin (1937), Furusato (1952), and Russo and Torrisi (1951a, 1951b) have also reported plants with tetraploid chromosome counts in several varieties.
      At and near metaphase I, there is much variation in chromosome conjugation.   In especially open figures, at a stage when the chromosomes are not yet oriented at the equator of the spindle, occasional pollen mother cells clearly have nine compound chromosomes (fig. 4-7, F).   These quadrivalents often look like ordinary bivalents joined by somewhat slender connections, but they also take other forms.   Often these are several separate bivalents, and sometimes apparently unconnected bivalents predominate (fig. 4-7, H).   Trivalent and univalent chromosomes are also common (fig. 4-7, G).
      At metaphase II (Frost, 1925b), the numbers are very irregular, metaphase groups of eighteen univalents being in the minority.   In some forms at least, unquestionable bivalents are common (fig. 4-7, J).   Stray chromosomes, which lagged at anaphase I, are common in the second division (fig. 4-7, K).
      The sporad stage usually indicates much irregularity of chromosome reduction (fig. 4-6).   In most tetraploid varieties studied, from one third to one half of the sporads counted have shown more than the normal four microspores, often six or seven.   In tetraploid Owari satsuma, however, only about one sixth to one fourth of the sporads had extra microspores.
      Where extra microspores are common, possibly extra nuclei are also present in some microspores.   Presumably, also, many irregular chromosome assortments do not lead to the production of extra nuclei.   Therefore, tetraploids usually produce few pollen grains whose chromosome complement is completely normal.   However, tetraploids have been successfully used as pollen parents in the production of triploid hybrids (see chap. 5).

Meiosis in Triploid Hybrids
      Longley (1926) studied the chromosomes in the pollen mother cells of a triploid hybrid from a cross of limequat by kumquat [(Citrus aurantifolia (Christm.) Swing. X Fortunella japonica (Thunb.) Swing.) X F. hindsii (Champ.) Swing.], the last mentioned species having thirty-six chromosomes.   Apparently trivalent chromosomes were common; sometimes several univalents were present.
      Frost (1927, 1929, 1930, and unpublished) examined chromosomes from about twenty triploid or approximately triploid hybrids from diploid parents.   Sporad counts were also made for triploid trees.   As in Longley's (1926) material, first-metaphase and pre-metaphase figures were seldom completely countable in the sense of positive determination of the number of univalent elements present, but they were diagnostic of triploidy or approximate triploidy because of the frequent occurrence of conspicuous trivalents.   The three univalent elements were very often joined end-to-end.   Other common shapes observed were the T, the V, the Y, and the ring and rod (fig. 4-7).   Trivalents usually predominated, but bivalents and univalents were common, often being more numerous than shown in the figures.
      Luss (1935), Lapin (1937), Yarnell (1939), and Russo and Torrisi (1951b) reported twenty-seven chromosomes in the root tips of various seedling forms.
      In the second division, the number of chromosomes in each metaphase group is highly variable, but usually exceeds the normal nine (fig. 4-7, M).   Some of the chromosomes are often astray, scattered through the cell or in groups (fig. 4-7, N).   Occasionally, under favorable conditions permitting a complete count, the total number of univalent elements found was twenty-seven.
      In triploids, as in tetraploids, the determination of chromosome number at anaphase I and metaphase II may be complicated by the presence of persistent bivalents.
      From the number and size of the chromosomes at metaphase II, it appears that the separate univalents of metaphase I must often divide in anaphase I, presumably by precocious separation of the chromatids (fig. 4-7, O).   This is indicated also by the occurrence at anaphase II of single equatorial laggards equal in size to the anaphase chromosomes.
      The sporad stage (fig. 4-6) shows much irregularity of cell division, although in some triploids, as in Longley's (1926) material, a large majority of the sporads have no extra microspores.   As with tetraploids, different masses of sporads from the same bud often differ surprisingly in the number of extra microspores; such variations seem to be due to variation in the number of nuclei.

Aneuploids
      Luss (1935), Krug and Bacchi (1943), Bacchi (1944), Ohta and Furusato (1957b), and Sharma and Bal (1957) reported as many as three chromosomes more than the diploid number in somatic cells.   Meiotic studies on aneuploids have not been reported.

Pollen Grain and Sperm Cells (Male Gametes)
      Development follows the usual course for angiosperm pollen.   The microspore enlarges, and develops two heavy coats, exine and intine.   Before the anther dehisces, the nucleus divides, forming the vegetative or tube nucleus and the generative nucleus.   J. L. Brewbaker (unpublished) reported the pollen to be binucleate.   Banerji (1954), however, reported the pollen of C. paradisi Macf. was trinucleate.
      Normal anthers are bright yellow when mature, owing to the pollen which they contain.   In varieties with very defective pollen development, the anther color may be considerably lighter, whereas anthers containing no pollen are pale cream or white and usually do not dehisce.
      Herrero-Egaña and Acerete (1935) made a detailed study of the pollens of varieties belonging to five of the most important species of Citrus.   They found that lemon pollen was readily distinguished from that of other species by its morphological characteristics.   Pereau-Leroy (1950) reported differences in the average diameter of the pollen grains of five species and two hybrids.   Nakamura (1943) listed pollen diameters for seven lemon varieties.   All but the Villafranca are within the range (34 to 36 microns) given by Pereau-Leroy for C. limon (L.) Burm. f.   Nair and Mehra (1962) reported distinguishing differences of the colpi and the lumina of the surface reticulations among the five species, Citrus aurantifolia, C. sinensis (L.) Osbeck., C. grandis (L.) Osbeck., C. limon, and C. aurantium L.

POLLINATION AND FECUNDATION

Selfing and Crossing
      Pollination of a pistil by pollen from stamens on the same plant is self-pollination, and fecundation of an egg cell by a sperm cell of the same plant is self-fecundation; both processes are often called selfing.   When a plant is divided by asexual propagation to produce a clone or clonal variety, all the trees of this variety are genetically identical (unless changes are produced by bud variation); pollination or fecundation between such identical trees is self-pollination or self-fecundation of a clone, and is equivalent to selfing of an individual tree.
      Cross-pollination and cross-fecundation involve the transfer of pollen from one plant or clonal variety to another.   Obviously, there is a genetic distinction between cross-pollination and self-pollination only where the two cross-pollinating plants are in some degree unlike genetically.

Agencies of Pollination
      Natural self-pollination can be brought about in Citrus by contact of anthers with the stigma, by pollen falling or being blown against the stigma, or by transfer of pollen by insects.   The pollen is of the sticky, adherent type characteristic of entomophilous (insect-pollinated) plants, and wind is therefore usually a minor factor in pollination.   Four characteristics of the flowers make them attractive to insects: the conspicuous corolla, the strong perfume, the pollen, and the abundant nectar (Webber, 1930).   Thrips are present in the flowers in great abundance, some apparently feeding on the pollen.   Honey bees and other insects work the flowers for both nectar and pollen.   Mites are sometimes found in the flowers and may occasionally inadvertently carry pollen (Uphof, 1934).   Natural cross-pollination is undoubtedly accomplished mainly by bees.
      In many citrus varieties with functional pollen, fruits with seeds frequently set on branches or trees that have been caged to exclude insects.   Thus, in some varieties at least, self-pollination occurs without the agency of insects.   Toxopeus (1931) found certain species and varieties to be somewhat proterandous (stamens maturing before the stigma), whereas in other forms the stamens and stigma were found to mature at about the same time.   In the former, there is a greater tendency for abundant self-pollination in the absence of insects, because the anthers begin to shed their pollen while they are still pressed against the stigma in the unopened or opening flower.   However, other factors may still prevent self-fecundation (see below).   The relative length of pistil and stamens may also affect the opportunity for self-pollination (Uphof, 1934).

Relation of Pollination to Setting of Fruit
      The significance of pollination for fruit production differs greatly in different varieties.   The principal factors concerned are the following: the amount of functional pollen, the facilities for pollination, the relation of the pollen to the setting of seed, and the ability of some varieties to produce seedless (parthenocarpic) fruits either with or without pollination.   The main evidence on the factors affecting seed production is discussed under the section headings of "Polyembryony" and "Sterility."

Parthenocarpy (Seedless Fruits)
      The production of fruits without seeds is parthenocarpy.   The setting of seedless fruits without any external stimulation is autonomic parthenocarpy.   The ordinary varieties of navel orange, satsuma mandarin, Tahiti lime, and certain other citrus varieties regularly produce parthenocarpic fruits, and many varieties usually seedy are more or less capable of parthenocarpy.   In varieties normally more or less seedy, the tendency to parthenocarpy is highly variable.   Ikeda (1904-06) and Nagai and Tanikawa (1928) found that when pollination was experimentally prevented, some varieties set no fruits, and other normally seedy varieties produced seedless fruits, usually setting much less abundantly, however, than when pollinated.   Many lemon varieties and Marsh grapefruit readily produce seedless fruits.   Valencia orange sets some seedless fruits when pollen is excluded (Wong, 1939).   Wilking mandarin set no fruit when pollen was excluded in trials at the University of California Citrus Research Center, Riverside.
      In certain plants, the formation of seedless fruits may be induced by pollen which, in the particular combination at least, is incapable of producing fecundation or the setting of seed.   Seedless fruits sometimes develop following self-pollination of self-incompatible citrus varieties.   Vituskina (1953) reported the set of lemon fruit after pollination with lily pollen.
      In citrus, growth-regulating compounds have not generally been effective in setting parthenocarpic fruits.   Furusato and Ohta (1957) and Furusato and Suzuki (1955) reported a few seedless fruits in Citrus natsudaidai Hayata after treatment with 2,4,5,-trichlorophenoxyacetic acid (2,4,5-T).   Gibberellin set seedless fruits on Clementine mandarin (Soost, 1958) and Orlando tangelo (Krezdorn and Cohen, 1962).
      Sokol'skaja (1940) and Plaut (1947) reported that seedless fruits were induced by breaking off the pistil prior to anthesis.   The seedless Valencia fruits reported by Wong (1939) were also set from flowers in which the pistils had been removed.   The stimulation of the formation of plant-growth regulators or the removal of inhibitors by the removal of the pistil are interesting conjectures.

Pollination and Fruit Yields
      Although the tendency to parthenocarpy is remarkably prevalent among the citrus fruits, the horticulturally successful parthenocarpic varieties seem to be exceptional in their high ability to set fruits without seeds.   If a variety incapable of producing seeds is to be horticulturally successful, it must be highly parthenocarpic; probably many seedless varieties have arisen which were worthless because they were not sufficiently parthenocarpic.   In decidedly seedy varieties, on the contrary, the degree of capacity for parthenocarpy is unimportant and ordinarily unnoticed.
      There is some reason to believe that seedlessness is rather generally a handicap to the setting of fruit.   Even the Washington navel orange is probably limited in yield by its seedlessness.   High temperatures during or immediately after fruit set can cause severe shedding of fruit in Washington navel orange.   Seedy varieties are less severely affected.   Coit and Hodgson (1919) also found that abscission of the flowers occurs more readily in Washington navel than in several less sterile varieties.   It is probable that the low productivity of many triploids is due to both ovule and pollen sterility (see chap. 5).
      Cross-pollination cannot have much effect on yields unless it occurs on a wide scale.   Any effect of pollination is likely to be of little importance in large solid plantings of varieties that have no good pollen of their own, since only trees not far apart are ordinarily likely to interpollinate extensively.   Shamel's (1918) observations indicated that there was little pollination in the Washington navel orange plantings which he studied, since seeds were found in only about ten out of more than 25,000 fruits of performance-record trees examined, though artificial cross-pollination of this variety very often produces a few seeds.   Webber (1930) concluded that insect pollination is not likely to have any important effect on the crop of navel oranges in solid plantings.
      The evidence from mixed plantings of the major varieties grown in California (Washington navel, Valencia, Eureka, Lisbon, Marsh) indicates that pollination by bees is probably a negligible factor in the production of these varieties.   Wong (1939) reported no difference between cross- and self-pollination in Valencia oranges.   Plaut (1947) got no increase in fruit set in Valencia with cross-pollination, although he did get an increase in seediness.   Cameron, Cole, and Nauer (1960) got an apparent increase in fruit set in Valencia by cross-pollination.   El-Tomi (1954, 1957b) reported improved yield in some years from cross-pollination of Washington navel.
      In two varieties of satsuma, Nagai and Tanikawa (1928) obtained substantially the same set of fruit with no pollination, with self-pollination, and (one variety tested) with cross-pollination.
      Although the presence of bees in solid or mixed plantings of some varieties may have no significant effect on yield, other varieties may benefit from the presence of bees even in solid plantings.   In the absence of other pollen sources, Lacarelle and Miedzyrzecki (1937) and Van Horn and Todd (1954) reported increased fruit set when bees were placed in cages built over individual Clementine mandarin trees.   Fujita (1957) got increased set with bee pollination in satsuma.   Hassanein and Ibrahim (1959) got reduced set in Khalili orange when bees were excluded.   Self-pollination by bees of flowers prior to normal anthesis may be a factor in this increased fruit set.   Flower buds of some varieties often have the stigma exposed prior to anthesis.   Iwasaki, Ohata, and Nishiura (1954) got increased seed set in the satsuma when flowers were pollinated at least four days prior to opening.   Pereau-Leroy (1950) also got increased fruit set in Clementine by using bud pollination.   Seedy fruit can also set by self-pollination of buds in at least some varieties of C. grandis that normally set only seedless fruit when self-pollinated (Soost, 1964).
      With self-incompatible or pollen-sterile varieties possessing a low degree of parthenocarpy, the presence of other pollen sources and bees is essential to fruit production.   Uphof (1934), who made comparative tests of self-pollination and cross-pollination on small branches of many varieties, found that fruit setting was commonly more abundant with crossing.   He concluded that the abundant occurrence of defective pollen in some varieties seemed an adequate explanation of this result, and that some pollen produced too little growth of pollen tubes.   This latter condition, if characteristic of selfing as compared with crossing, would indicate true self-incompatibility.   Because of self-incompatibility (see below), cross-pollination is needed for adequate fruit production in Clementine mandarin, Orlando tangelo, Minneola tangelo, Robinson mandarin, Hyuganatsu, and several pummelo varieties.   It is probably also needed in Sukega grapefruit and Osceola mandarin.

Pollination and Fruit Size
      The number of seeds per fruit may affect fruit size, at least in some varieties.   Krezdorn (1959) and Cameron et al. (1960) found significant correlations between fruit size and seed number in the Orlando tangelo and Valencia orange, respectively.   Significant correlation or a marked trend toward correlation was also reported by Cameron et al. (1960) in a seedy sweet orange, Mediterranean Sweet orange, Trovita orange, and Frua mandarin.   Minneola tangelo (Cameron et al., 1960) and Clementine mandarin (Coste and Gagnard, 1956; Soost, 1956) also have higher seed numbers in the larger fruits.   Because the number of seeds can be increased in several of these varieties by cross-pollination (see below), adequate pollination can affect not only yield but fruit size.   Hearn and Reece (1967) reported that Page mandarin fruits produced from self-pollination were seedless and too small for commercial use.   Seedy fruits were produced by cross-pollinating with Lee mandarin pollen, producing larger fruits than either Orlando tangelo or Temple tangor pollen.

Pollen Tube
      The thick, sticky liquid secreted by the stigma serves not only to catch and hold the pollen grains, but also provides suitable conditions for their germination.   Climenko (1936), Singh and Dhuria (1960), Randhawa, Nath, and Choudhury (1961) reported that the stigma is receptive one to three days prior to anthesis and retains receptivity for six to eight days after anthesis.   Bud-pollination (Fujita, 1957) has demonstrated successful pollen-tube growth as much as four days prior to anthesis.   The pollen grain absorbs moisture from the stigmatic liquid and forms the pollen tube.   Pollen tubes grow through the style to locules of the ovary.   According to Webber (unpublished), the pollen tubes grow in the stylar canals.   However, Banerji (1954) stated that growth is intercellular and not in the canals.   In the ovary, a pollen tube grows through the micropyle of each ovule, disorganizing one or both synergids as it enters the embryo sac (Bacchi, 1943; Banerji, 1954).

Fecundation
      When the pollen tube has reached the embryo sac by way of the micropyle, a sperm nucleus passes into the embryo sac, where it fuses with the egg nucleus, thus accomplishing fecundation.   According to Banerji (1954), the second sperm nucleus fuses simultaneously with the two polar nuclei to produce a triploid endosperm nucleus.   Toxopeus (1933) found 27-chromosome (triploid) endosperm in two diploid forms of Citrus.   Figure 4-4, E and F, which show embryo sacs ready for fecundation, indicate that the polar nuclei themselves probably fuse (Osawa, 1912).   Figure 4-8 shows the egg cell after fecundation has occurred and the endosperm nucleus has divided.
      The period from pollination to fecundation has been reported to be from two days to four weeks.   Strasburger (1878), Toxopeus (1930), and Osawa (1912) reported periods from two to four weeks.   In Foster grapefruit, Bacchi (1943) found fecundation occurred in four days.   Furusato (1953) reported a period of two to eight days for satsuma.   Ikeda (1904-06) also observed that removal of the stigma two or three days after pollination did not prevent setting of seed.   The shorter period to fecundation probably is more representative of Citrus when growing under favorable conditions.   Pollen-tube growth could be altered greatly by environmental f actors or by pollen incompatibility.   Gerasimova-Navashina (1960) indicated that fusion of the gametes depends on temperature and the condition of the gametes.

DEVELOPMENT OF EMBRYO AND ENDOSPERM

The Gametic Embryo
      Strasburger (1878) stated that in greenhouses in Germany the egg, fecundated in the fall, usually overwintered before dividing, but that in southern countries development was uninterrupted, Schacht finding that on the Madeira Islands off the coast of Morocco a mandarin variety showed the first division about two months after flowering.   Osawa (1912) found that in Poncirus the first division of the egg took place between two and four weeks after fecundation (fig. 4-8).   Bacchi (1943) stated that the first division occurs about fifty days after fecundation, and Furusato (1953) indicated a period of about one month.   The first divisions are transverse (fig. 4-8); the cells nearest the micropylar end of the embryo sac form the slender suspensor, and the apical cell at the free end of the row divides to form a rounded mass of cells.   Later outgrowths initiate the various parts of the complete embryo: hypocotyl, radicle, cotyledons, and plumule.

Nucellar Embryos
      In many varieties of Citrus, and also in Fortunella and Poncirus, extra embryos, derived not from cells of the embryo sac but from somatic cells of the nucellus, are developed in the ovules.   These grow into the embryo sac and lie alongside the normal embryo.   After fecundation, and before or soon after the egg undergoes its first division (Strasburger, 1878; Biermann, 1896; Osawa, 1912; Toxopeus, 1930; Bacchi, 1943; Furusato, 1953; Vasiljcova, 1951), stained sections of ovules show a few large cells, with large nuclei and dense cytoplasm, in the nucellus.   These cells are mainly near the micropylar end of the embryo sac and within one or two cell layers of it (fig. 4-8, C).   Some of these cells begin to divide and produce small masses of cells which protrude into the sac alongside the embryo derived from the egg cell (fig. 4-8, D).   While very young, these embryos may be distinguishable by their irregular shape and lack of a suspensor.   However, Maheshwari and Ranga-Swamy (1958) reported the presence of suspensors.   These extra embryos are called nucellar embryos (Toxopeus, 1930).   The time relationship between the initiation of development of the zygotic embryo and the nucellar embryos may be quite variable.   Biermann (1896) concluded that at an early state the nucellar embryos are less advanced than the zygotic embryo.   Toxopeus (1930) and Vasiljcova (1951) indicated that most of the nucellar embryos were several-celled when the egg began to divide.

Endosperm
      Before the egg cell divides, the embryo sac begins to enlarge rapidly, crowding and destroying many cells of the nucellus.   The endosperm nucleus divides, and successive divisions form many free nuclei, which become scattered through a thin layer of cytoplasm that lines the embryo sac (fig. 4-8).   Bacchi (1943) found the free nuclear condition still present 67 days after pollination.   Cell walls then form, and these cells divide further to form the endosperm.   In Citrus and closely related genera, the endosperm transports nutrients for the developing embryos.

Embryo Culture
      Embryo culture has been used as an aid in breeding and in investigating various aspects of nucellar embryony.   Sobrinho and Gurgel (1953) reported growth of 2 mm embryos in "Sachs" solution at pH 5.   Stevenson (1956) reported shoot and root growth of embryos extracted from fruits within five to six months after pollination and placed on nutrient agar.   Hurosvili (1957) got satisfactory growth on Tukey's medium at pH 7 to 7.5 with vitamins B₁ and B₂ added.   Ohta and Furusato (1957a) unsuccessfully attempted to culture small embryos (less than 1 mm) from mature seeds.   Embryos larger than 3 mm grew best on nutrient agar without sucrose at pH 6.   Addition of sucrose caused growth reduction.   They further reported that addition of 0.01 ppm of vitamin C or vitamin B produced increased growth, but higher concentrations retarded growth.   The addition of 2,4,5-T may have increased growth at 1 ppm or less; however, no stimulation of 2,4-dichlorophenoxyacetic acid (2,4-D) or naphthaleneacetic acid (NAA) was noted and concentrations of NAA greater than 0.1 ppm retarded growth.   Ozsan and Cameron (1963) found 2 per cent sucrose better than 1 per cent for growth of 1- to 3-mm embryos, but 1 per cent was better than 2 per cent for 4 and 5-mm embryos.   Absence of sucrose reduced growth of all embryos.   With 2- and 4-mm embryos, root growth was much greater when only sucrose was added to the basic nutrient agar.   For 1-mm embryos, growth was best when immature coconut milk and a combination of growth factors were added in addition to sucrose.   The inorganic media of Tukey and Nitsch were better than those of Randolph and Cox or White (modified).   A 0.7 per cent agar was used with all media.   Ranga-Swamy (1959) reports enlargement and differentiation of "proembryos" on 5 per cent sucrose, but not on 2 or 3 per cent when they were included with extracted ovules.   Proembryos extracted from ovules did not grow on White's medium.   Sabharwal (1962) was unable to induce the development of nucellar embryos in vitro unless they had been initiated prior to the extraction of the nucelli.
      Dr. Toshio Murashige and co-workers of the Citrus Research Center and Agricultural Experiment Station, Riverside, California, are currently developing culture techniques that may provide a method of obtaining nucellar embryos from the nucellus of varieties that normally do not produce nucellar embryos.   Nucellar tissue extracted after the initiation of the zygote has differentiated on a defined nutrient medium.   Partially developed zygotes extracted from varieties that normally produce very few sexual seedlings have also differentiated on the same medium.

POLYEMBRYONY

Multiple Zygotic Embryos
      In several cases among hybrids produced at Riverside, California, two separate seedlings grew from one seed.   In each case, the two hybrids seem identical in genetic type, although such identity has not been observed elsewhere among citrus hybrids.   It is most likely that each of these pairs of "twins" were produced by the development of two embryos from one fecundated egg.   Twin hybrids were also found in Swingle's seedling populations (Traub and Robinson, 1937).
      Traub and Robinson (1937) also reported one triple hybrid and one quadruple hybrid in Swingle's populations.   Ozsan and Cameron (1963) proved in four cases that three seedlings from the same seed were hybrid.   These multiple embryos could be produced by multiple fission of the fecundated egg or by the functioning of more than one embryo sac per ovule (see also chap. 5).

Nucellar Embryony
      Since nucellar embryos develop asexually by ordinary mitotic division of cells of the nucellus, no male cells contribute to their formation, and no reduction division occurs in the seed-parent cells which give rise to them.   Nucellar seedlings therefore not only inherit from the seed parent alone, but are actually identical with it in genetic constitution, except for possible differences due to somatic (bud) variation.   This asexual reproduction has very important consequences for evolution, breeding, and culture of the citrus fruits, all of which is discussed in Chapter 5.
      Terminology of Nucellar Embryony.—The term nucellar embryony (Ernst, 1918; Sharp, 1934) is especially suitable because it precisely indicates what occurs in Citrus and its near relatives, and includes nothing else.   Embryos which develop from the nucellus may appropriately be called nucellar embryos, and trees which grow from these embryos are nucellar seedlings; these embryos and seedlings are the nucellar progeny or offspring of their parent.
      The more inclusive terms apomixis and apomictic are obviously applicable to nucellar embryony.   They are preferable to apogamy and apogamic, which have sometimes been employed for Citrus, but which are commonly restricted to formation of embryos from gametophytic cells.
      Other terms for the nucellar embryos are: false, extra, adventive, adventitious, asexual, and vegetative.   Terms applied to embryos developed from the egg are: true, normal, sexual, generative, zygotic, and gametic; of these, zygotic seems to be the most satisfactory, since it precisely indicates the origin, except in possible instances of parthenogenesis.
      Relation of the Pollen to Nucellar Embryony.—The available evidence on the initiation of the development of nucellar embryos indicates that pollination is usually, but perhaps not invariably necessary.   The usual seedlessness of the ordinary varieties of navel oranges, satsuma mandarin and Tahiti lime in solid plantings indicates that in these varieties seeds do not often form, if ever, without pollination by varieties having good pollen.   Such varieties show that auxin may be adequate for fruit setting without inducing nucellar embryony.   Furusato and Suzuki (1955) produced only seedless fruits when using 2,4,5-T.   Only seedless fruits were obtained by Soost (1958) by application of gibberellin to unpollinated Clementine fruit.
      Except for two reports (Webber, 1930; Wright, 1937), seed formation has not been observed when pollen has been excluded.   Ikeda (1904-06) and Coit (1915) found that prevention of pollination resulted in seedlessness when any fruits were produced.   Nagai and Tanikawa (1928) emasculated and bagged, without pollination, flowers of eighteen citrus varieties which produced, when self-pollinated, averages ranging from one to forty-three seeds per fruit.   Of these eighteen varieties, thirteen produced some fruits without pollination, but all these fruits were seedless.   Wong (1939) found that four varieties, which commonly have seeds if pollinated, produced only seedless fruits when pollen was excluded.   Toxopeus (1930) obtained no fruits when pollination was prevented in several hundred flowers of three seedy varieties.
      With certain interspecific hybrids, which rarely, if ever, produce zygotic seedlings, pollination seems to be necessary for the formation of nucellar embryos (Swingle, 1927).   The fact that the seed production of some such hybrids is much greater when suitably cross-pollinated than when they are self-pollinated also indicates dependence of nucellar embryony on pollination with functional pollen; the Thomasville citrangequat and the Morton citrange are examples.
      Although there may be exceptions, pollination is unquestionably usually necessary for the formation of nucellar embryos.
      Whether or not fecundation is a prerequisite for the formation of nucellar embryos is not clear.   Several varieties which seem to be self-sterile (see below) occasionally produce fruits having "imperfect seeds with poorly developed embryos" after self-pollination or evidently incompatible cross-pollination (Nagai and Tanikawa, 1928).   The embryos in these seeds may be nucellar embryos which result from pollen-tube stimulus, without fecundation and therefore without normal endosperm development.   Toxopeus (1930) obtained no fruits when pollen of Feronia, a near relative of Citrus, was used on Citrus.   This lack of fruit set could be due to lack of pollen-tube growth rather than lack of fecundation.   The initiation of nucellar embryos after fecundation has indicated the need for fecundation in their development.   However, it is not clear that some development does not occur without fecundation.
      Johri and Ahuja (1956) described the development of endosperm and nucellar embryos in embryo sacs of Aegle marmelos (L.) Corr., a Citrus relative, in which the egg and antipodal cells degenerate.   Presumably, union of polar nuclei and a sperm cell occurred.   If such a phenomenon occurs in Citrus, nucellar embryos could develop without the initiation of a sexual embryo.

Interrelations of the Two Classes of Embryos
      Except in the forms which rarely or never produce nucellar embryos, the zygotic embryo must often or usually compete for space and nutrients with one or more nucellar embryos.   Strasburger (1878), Bacchi (1943), and Furusato (1953) observed that embryo sacs may contain one, two, or sometimes more embryos developing normally, together with others that are more or less suppressed.   In table 4-1 data are given illustrating the elimination of embryos at germination.   The result of this competition must depend on the number of nucellar embryos, their time of starting, their location, and the relative inherent or genetic vigor of the two kinds of embryos.   Reports by Sokol'skaja (1938) and Majsuradze (1961) indicated that the lower the number of embryos per seed, the larger the average size of the embryos and the greater the probability of hybrids surviving.   In many varieties, a seed produces only one or more nucellar seedlings, without a gametic seedling, indicating that the zygotic embryo (if ever formed) was crowded out or was simply too weak to survive.   When the zygotic embryos fail to survive in all or most of the seeds, the seedlings produced will be all or nearly all from nucellar embryos, and the seed parent will "breed true" with respect to all or nearly all of its offspring (Frost, 1926; Toxopeus, 1936).
      Toxopeus (1936) mentioned three factors as being relatively unfavorable to the zygotic embryos: (1) in the material which he examined, the fecundated egg cell began to divide when most of the nucellar embryos already consisted of several cells each; (2) the young zygotic embryo, situated near the apex of the embryo sac, seems less favorably placed than the nucellar embryos, with respect both to transport of nutrients from the vascular system and to the opportunity to occupy the space within the sac; and (3) zygotic seedlings for the most part are weaker in genetic constitution than the corresponding nucellar seedlings, and the zygotic embryos which survive to produce seedlings are probably more vigorous, on the average, than those which are crowded out.   Evidence indicates that the pollen parent may influence the relative proportions of zygotic and nucellar progeny (Toxopeus, 1931, 1936; Torres, 1936; Frost, 1926; Mamporija, 1957; Lomija, 1961a, 1961b; Soost and Cameron, unpublished).   Toxopeus (1936) concluded that this is most likely a result of unlike influences of different kinds of pollen on the development of nucellar embryos.   However, differences in the vigor of the hybrids produced may be the main cause.   Environmental and internal factors undoubtedly also affect the proportions of zygotic and nucellar seedlings.   In breeding work at Riverside, California, with the Orlando tangelo, using a single variety of pollen, a few fruits produced over 10 per cent zygotic progeny while all other fruits produced only nucellar progeny.
      The proportions of zygotic and nucellar progeny resulting from cross-pollination of various varieties (table 4-2) have been reported by Webber (1900), Frost (1926), Toxopeus (1930), Torres (1932, 1936, 1938), Brieger and Moreira (1945), and Russo and Torrisi (1953).   The proportions from self or open pollination have been discussed by Frost (1926), Toxopeus (1930), Webber (1932), Torres (1936), Moreira and Gurgel (1941), Naik (1949), and Russo and Torrisi (1953).   Additional unpublished information on the proportions of zygotic and nucellar progeny has been obtained in breeding programs of the Citrus Research Center at Riverside, California, and of the U.S. Department of Agriculture at Orlando, Florida, and Indio, California.

Numbers of Embryos and Seedlings
      The number of embryos in the mature seeds has been reported by Frost (1926), Toxopeus (1930), Torres (1936), Navarro de Andrade (1933), Brieger and Moreira (1945), Nasharty (1945), Moreira, Gurgel and de Arruda (1947), Gurgel and Sobrinho (1951), Py (1951), Minessy (1953), Furusato (1953), Russo and Torrisi (1953), Chapot and Praloran (1955), Furusato, Ohta and Ishibashi (1955), Furusato (1957), Minnessy and Higazy (1957), Parlevliet and Cameron (1959), and Ozsan and Cameron (1963).   Additional information on the number of seedlings per seed has been obtained in breeding programs at the Citrus Research Center, Riverside, California, and from the U.S. Department of Agriculture at Orlando, Florida, and Indio, California.   Tanaka (1954) indicates whether or not many of the forms he discusses are polyembryonic.
      The number of embryos per seed varies greatly even on one tree, the average number differs greatly according to variety, and there is no general consistency within many of the species in which polyembryony is present.   Nasharty (1945), Chapot and Praloran (1955), and Minessy and Higazy (1957) presented data showing rather large differences observed between localities in the same year.   Moreira et al. (1947) and Minessy and Higazy (1957) demonstrated differences between years.   Moreira et al. (1947) also analyzed the variability between trees, fruits, and seeds.   Nasharty (1945) and Minessy (1953) also indicated an effect of rootstock with some scions.   Undoubtedly, many environmental and other factors affect the number of embryos per seed.   Traub (1936) reported a decrease in the average number of embryos by decreasing the food supply.   Furusato (1953) reported a reduction in the number of embryos per seed after injection of young fruit with either water or maleic hydrazide-30.   Furusato et al. (1955) reported significantly higher numbers of embryos per seed on the north side of trees and in old (thirty-year-old) trees.
      The embryos in a polyembryonic seed often differ widely in size and shape (fig. 4-9).   The average number of seedlings produced per seed is commonly much smaller than the total number of embryos, and the seedlings are doubtless derived mainly from those embryos that are comparatively large (table 4-3).   Although some varieties have more than two embryos in the great majority of their seeds, in most varieties few seeds produce more than two or three seedlings each.   Table 4-2 presents the percentages of nucellar seedlings for representative varieties.   Because of considerable confusion in the nomenclature of citrus varieties and species, the degree of polyembryony of any accession should be checked unless its identity is well known.
      In C. limon, the Eureka and Lisbon lemons have only one embryo in most of the seeds and give low percentages of nucellar seedlings (Frost, 1926).   Russo and Torrisi (1953) reported low percentages of nucellar seedlings for Monachello, but rather high percentages for Feminello.
      Rough lemon, classified as a separate species by Tanaka (1954), has a relatively large number of embryos, and usually has a high percentage of nucellar seedlings.   When P. trifoliata pollen was used, 40 per cent gametic seedlings were obtained (Soost, unpublished).   Meyer lemon is monoembryonic, and no nucellar seedlings have been reported.   Ponderosa lemon and sweet lemon (Moreira et al., 1947) have a very low degree of polyembryony.
      In C. reticulata, many varieties, including Dancy, Ponkan, Willowleaf, and satsuma, have numerous embryos and high proportions of nucellar seedlings.   The King has few extra embryos and usually gives low proportions of nucellar seedlings from cross-pollination, but the Japanese variety Kunenbo, very similar to King, is usually polyembryonic (Tanaka, 1954).   Kinnow and Kara, which have King as one parent, have many embryos per seed and produce few, if any, gametic seedlings.   However, Wilking and Kincy, also hybrids having King as one parent, are monoembryonic and have not been known to produce any nucellar seedlings.   Temple and Clementine, which are hybrids of unknown parentage, are also monoembryonic and no nucellar seedlings have been reported.   The Clement tangelo, Lee hybrid mandarin, and a considerable number of unnamed hybrids are also monoembryonic (see chap. 5).   Torres (1936) reported the variety Kishiu as monoembryonic.
      In C. sinensis the number of embryos may be either moderate or high, and the proportions of nucellar seedlings are rather high in most varieties.   No monoembryonic varieties have been reported.
      Various grapefruit varieties (C. paradisi) produce many nucellar embryos.   Several ordinary varieties have given high proportions of nucellar seedlings, whereas the Imperial, which is decidedly unlike the usual commercial varieties, has given a medium proportion.   Torres (1932) found a low proportion in the variety Panuban.   Sukega, a hybrid grapefruit, is monoembryonic (Parlevliet and Cameron, 1959), and no nucellar seedlings have been identified.   The Wheeny grapefruit, which appears to be a hybrid, is also monoembryonic.   Moreira et al. (1947) report some questionable grapefruit as monoembryonic.
      The pummelos (C. grandis) have been reported to be monoembryonic by most investigators (Toxopeus, 1930, 1933; Torres, 1936; Brieger and Moreira, 1945; Nasharty, 1945; Py, 1951).   No nucellar seedlings have been produced at the University of California Citrus Research Center, Riverside, using C. grandis as a seed parent in the production of several hundred hybrids.   Frost found a high percentage of nucellar embryos in the Frizzelle pummelo, and other C. grandis accessions at Riverside have also been polyembryonic.   Because C. grandis hybridizes readily with other citrus and many of the hybrids are pummelo-like in appearance, these polyembryonic varieties could well be hybrids.
      Although C. aurantium may give rather low numbers of embryos, the proportion of nucellar seedlings is high.
      The citrons (C. medica L.) appear to be mainly monoembryonic (Moreira et al., 1947; Py, 1951; Tanaka, 1954; Chapot and Praloran, 1955; and Carpenter, unpublished).
      The many diverse types classified as limes (C. aurantifolia Christm.) Swing., including the so-called mandarin-limes (C. limonia Osbeck) generally have fairly low numbers of embryos per seed but usually produce low proportions of hybrid seedlings.
      Although Tanaka (1954) indicated a number of his reported species were monoembryonic, it is not always clear whether they are strictly monoembryonic and produce no nucellar embryos.   His C. junos Sieb. ex Tan. is indicated to be monoembryonic, yet correctly identified accessions of this species at Riverside produce nucellar seedlings.
      Torres (1930, 1936), Nasharty (1945), Moreira et al. (1947); Chapot and Praloran (1955), and Parlevliet and Cameron (1959) found occasional extra embryos or seedlings in varieties usually producing only one embryo per seed and not known to have produced nucellar seedlings.   Ozsan and Cameron (1963) obtained seed in three of these monoembryonic varieties (Clementine, Wilking, and Siamese pummelo) by guarded cross-pollination with P. trifoliata, which imparts a dominant trifoliate leaf to its hybrids.   All nucellar seedlings would therefore be monofoliate.   Of a total of 131 embryos, excised from 50 seeds with more than one embryo each, 69 were determined to be zygotic.   The remaining 62 embryos did not grow large enough to be judged.   No seedling was shown to be nucellar.   In 14 cases, two embryos from the same seed were shown to be zygotic; in four cases, three from the same seed were proved so.   The data strongly indicates that these three varieties do not produce nucellar embryos, and suggest that other known cases of extra embryos in varieties normally producing only zygotic progeny are also not of nucellar origin.

THE SEED

Embryos
      The cotyledons, in which most of the food supply for the germinating seedlings is stored, are fleshy and constitute by far the largest part of the mature embryo (fig. 4-9).   They are attached to a very short hypocotyl, at the other end of which is the rudimentary radicle.   Lying between the cotyledons is the plumule, which until germination begins is merely a minute cone that has not yet developed the first true leaves (Casella, 1931).   Martin (1946) classified the embryo as investing.
      The one or more embryos form a more or less solid, rounded-to-corrugated mass, in which the radicles normally point toward the micropylar end of the seed.   A seed with one embryo usually has two cotyledons about equal in size and shape, but when there are two or more embryos the size and shape of their cotyledons often vary greatly (fig. 4-9).   Occasionally an embryo has three or more cotyledons.   When a seed contains several embryos, usually part of these are poorly developed, often with excessively small cotyledons (fig. 4-9).   (See also Chapot and Praloran, 1955.)   In polyembryonic varieties, the embryos are commonly crowded together at the micropylar end, but sometimes are located away from it (Nasharty, 1945; Parlevliet and Cameron, 1959).   Sometimes a small embryo lies in a hollow in the outer side of a large cotyledon, or between two large cotyledons.
      The color of the cotyledons may be creamy white, cream, or various shades of green.   Intense green color is characteristic of the tangerines and mandarins, but some varieties such as the King only show occasional green.   Chapot and Praloran (1955) described the satsuma as varying from deep pistachio green to white.   They reported that oranges (C. sinensis), sour orange (C. aurantium), grapefruit (C. paradisi), pummelos (C. grandis), limes (C. aurantifolia), and citrons (C. medica) have white cotyledons.   Varieties that are normally white may develop a greenish color in overripe fruit.   Chapot and Praloran (1955) particularly noted this behavior in C. limon.

Seed Coats
      The tegmen, or inner seed coat, is a thin, rather delicate membrane which closely invests the embryo or embryos.   The inner integument of the ovule probably forms the main part of the tegmen.   The tegmen must include whatever remains of the nucellus and the endosperm.   In most varieties, the tegmen is characteristically colored.   The area covering the chalazal end is usually purple, brown (various shades), pink, yellow, or cream and the rest of the tegmen is usually lighter shades of these colors or grayish white (Chapot and Praloran, 1955).   Chapot and Praloran suggested the use of the tegmen color, particularly at the chalazal end, as an identifying characteristic for the major groups of cultivated varieties and for relating uncertain varieties to these groups.   Chapot and Praloran (1955) reported the association of the color of the chalazal end of the tegmen with the coloring of other plant parts and the acidity of the fruits.   C. limon and acid varieties of C. medica are reported to have purple color in the flowers, new shoot growth, and chalazal end of the tegmen.   In contrast, the non-acid citrons (C. medica) have white flowers and no purple color in the new shoot growth or in the tegman.
      The outer seed coat or testa is usually grayish white, cream, or yellow in color and tough and woody in texture (figs. 4-10 and 4-11).   In Citrus, the testa is often considerably wrinkled or ridged and is usually extended beyond the rest of the seed at one or both ends (especially the micropylar end) to form a conical beak or a flat plate which may extend along the whole length of the seed.   In some polyembryonic seeds, the surface may be corrugated because of the unequal development of the embryos.   In Poncirus trifoliata, Moreira et al. (1947) correlated the degree of corrugation with the number of embryos.   The surface of the testa is mucilagenous, so that the fresh seeds are very slippery.   In P. trifoliata, the testa is coriacious and much less mucilagenous than in Citrus (Chapot and Praloran, 1955).

Seed Color
      The color of the intact seed depends chiefly on the color of the tegmen and the embryos and the extent to which their color is visible through the testa; in a variety which has green cotyledons, the green color may be distinctly visible through the seed coats.   Either drying or soaking in water changes the color of seeds by changing the opacity of the testa.   Fresh seeds, in general, range from grayish white, cream, or yellow to brownish or greenish, with characteristic differences manifest between varieties (Casella, 1931; Chapot and Praloran, 1955).

Size and Shape of Seeds
      Seeds of a given variety may be highly variable in size (Casella, 1931; Chapot and Praloran, 1955).   The size and shape of the seeds is also affected by the number of seeds per fruit.   Chapot and Praloran (1955) described seed shape as lenticular, cuniform, ovid, fusiform, almond, spherical, and modifications of these shapes (fig. 4-10).   They indicate typical shapes for each of the species containing cultivated varieties.   Some varieties also have characteristic shapes which aid in their identification.   Chapot and Praloran (1955) suggest the use of seed shape to determine parentage of unknown hybrids.

Germination
      The cotyledons are normally hypogean, remaining in the soil during germination.   They absorb moisture and swell, and the tip of the radicle elongates and emerges through a break in the micropylar end of the testa (Casella, 1931).   The radicle elongates rapidly to form a thick, fleshy taproot, which grows directly downward (fig. 4-11).
      Meanwhile the plumule, so rudimentary in the resting seed, has been rapidly lengthening its first internode (the epicotyl) and developing the first two true leaves.   The cotyledons lengthen somewhat, and after the taproot has made considerable growth the plumule emerges from the burst testa (fig. 4-11) and grows upward.   Its tip is bent while pushing through the soil but as soon as it emerges it straightens up (fig. 4-11).
      The first true leaves rapidly enlarge and soon separate.   Usually there are two of these leaves; they are situated opposite on the stem and are somewhat unlike the later leaves in shape, so that they may be mistaken for cotyledons.   Even in forms with normally trifoliate leaves, the first leaves are often unifoliate.   Except in albinistic plants (see the chapter on propagation in vol. III [text version, Revised Ed.]), which remain entirely or partly whitish, the first leaves are yellow in the soil and become green very rapidly after they emerge.   They soon reach their full size, and at that time germination may be considered complete.
      Secondary roots do not appear until the taproot is some three or four inches (8 or 10 cm) in length and the first leaves are considerably developed (Casella, 1931).
      Both root and stem normally emerge promptly at the micropylar end of the seed, but very often the tough testa obstructs the tip of the root so that it becomes bent or twisted before emerging.   The position of the embryo in relation to gravity may also produce bent roots (Nasharty, 1945).
      Many environmental factors, diseases, and pests affect the seeds during storage and after planting.   For a discussion of the relations of these factors to seed handling, see Chapter 1 of this volume and the chapter on propagation in Vol. III [text version, Revised Ed.].

Identification of Seedling Type (Nucellar versus Zygotic)
      For breeding, rootstock use, and experimental work, it is important to be able to accurately distinguish zygotic and nucellar seedlings at an early stage.   When the pollen parent is known and has morphological characteristics that clearly distinguish it from the seed parent, zygotic seedlings can be identified with some degree of accuracy.   However, when the pollen parent is either unknown (open-pollination) or when its morphological characteristics are identical or similar to the seed parent, as in self-pollination or in "close" crosses, it becomes extremely difficult to distinguish between zygotic and nucellar seedlings.   Even when the parents differ considerably in their characteristics, hybrids could be obtained that closely resemble the seed parent and might be identified as nucellar seedlings.
      Furr, Reece, and Hrnciar (1946) and Nishiura, Matsushima, and Okudai (1957) used modifications of the rootstock color-reaction test in attempts to distinguish nucellar and zygotic seedlings in progenies obtained from known cross-pollinations.   When both parents were very closely related or had similar color reactions, it was not possible to distinguish zygotic seedlings.   When the parents had dissimilar color reactions, it appeared that the nucellar seedlings were often intermediate in color.   However, Furr and Reece (unpublished) indicated that some seedlings identified as zygotic subsequently proved to be nucellar.   Thus, this particular test does not appear to offer a basis for accurately distinguishing zygotic seedlings in progenies where it is also difficult to distinguish them on a morphological basis.   Pieringer and Edwards (1965) used infrared spectroscopy to distinguish between nucellar and zygotic seedlings.   Although it is a very sensitive technique, it may prove to have the same limitations as the calorimetric method.   Pieringer, Edwards, and Wolford (1964) indicated that gas chromatography may be a usable technique for distinguishing between nucellar and zygotic seedlings.

STERILITY

Degree of Seediness in Cultivated Varieties
      The average number of seeds per fruit, considered in relation to the number of ovules per ovary, gives a rough measure of fertility and sterility, although nucellar embryony complicates the interpretation of the evidence.   There are characteristic varietal differences in seediness although much variation occurs (Moreira and Gurgel, 1941; Chapot and Praloran, 1955).
      Complete seedlessness under all conditions is rare among citrus varieties.   It may be characteristic of the Mukaku-Kishiu mandarin which is said to be a bud-variation strain of the seedy Kishui mandarin (Nagai and Tanikawa, 1928).   Chapot and Praloran (1955) reported the Otaheite to be seedless.   The Tahiti lime has been described as seedless (Hume, 1926; Uphof, 1931; Torres, 1932), but it is clear that occasional seeds can be produced (Traub and Robinson, 1937; Chapot and Praloran, 1955; Reece and Childs, 1962).
      The Washington navel and various other varieties of navel orange rarely produce any seeds in ordinary culture because they entirely lack pollen.   Since they produce some good embryo sacs, seeds can be produced by cross-pollination (Moreira and Gurgel, 1941; Majsuradze, 1951; El-Tomi, 1957a).   Therefore, although under ordinary conditions of large-scale culture the common varieties of navel orange usually are entirely seedless, it is possible to use them as seed parents in crossing.   Because of the absence of good pollen, however, they cannot be used as pollen parents.
      The satsumas produce some pollen, but the fruits are usually seedless in the absence of cross-pollination.   Pacher (1938) and Furusato (1951) indicate that it is possible to obtain adequate functional pollen for breeding under some environmental conditions.   Using satsuma pollen, Reece (unpublished) has produced hybrid seedlings.   With cross-pollination the satsumas ordinarily produce seeds considerably more readily than does the Washington navel (Miki, 1922).
      Next in the scale of sterility after the satsuma come varieties which presumably all produce some good pollen, and which usually have seeds in many or all of their fruits, but in which the number of seeds is small or very small, the average usually being less than five, in some varieties commonly a little higher.   Among such varieties are the Cadenera, Valencia, Boone, Shamouti, Enterprise, Hamlin, Verna, and Doblefina oranges; the Eureka, Lisbon, Berna, and Villafranca lemons; and the Marsh and other "seedless" grapefruits.   In these varieties, only a very small number of the ovules usually develop into seeds, indicating a high degree of sterility.
      Next to the few-seeded varieties may be grouped varieties which commonly have an average of from eight to ten or twenty seeds, including many of the mandarin varieties, the Paperrind, Parson Brown, Pineapple, and Ruby oranges and sour orange varieties.   Even in these the numbers of seeds are very much smaller than the number of ovules that have been observed; for example, in the Shamouti orange Oppenheim and Frankel (1929) found about seventy ovules per ovary.
      In other varieties, the number of seeds are still higher; in some varieties of grapefruit and shaddock (C. grandis), for example, the number may be as high as one hundred seeds per fruit.
      It is therefore probable that most of the commonly cultivated varieties, if not all, have some degree of ovule or pollen sterility.
      The seediness of some varieties can be greatly increased by cross-pollination.   This effect can be produced in some varieties such as Shamouti orange (Oppenheim, 1929a, 1929b) and Kara mandarin (Soost, unpublished) that have a low percentage of viable pollen.   In varieties that are self-incompatible (see below), fruits set by self-pollination may be seedless, but fruit set by cross-pollination with compatible pollen will be seedy.   Cross-incompatible varieties (see below) could set seedless fruit when pollinated by incompatible pollen, but would produce seedy fruit when compatible pollen was used.

Kinds of Generative Sterility
      With respect to the stage at which failure occurs, the cases of generative sterility may be divided into two main classes: gametic sterility and embryo (zygotic) abortion.   Gametic sterility may be further divided into relative gametic sterility and absolute gametic sterility.
      Relative Gametic Sterility (Self-Sterility and Cross-Sterility).—Self-sterility properly means self-incompatibility, or inability to form embryos by self-pollination although both the sperm cells and the egg cells are functional in suitable cross-pollination.   Since the plants of one clone may all be considered as separate parts of one plant, the term self-sterility includes incompatibility in pollination between different plants of the same clone as between different trees of Clementine mandarin, for example.
      The corresponding form of cross-sterility is cross-incompatibility, or inability to produce embryos with cross-pollination in certain kinds of matings, although the egg cells and sperm cells of the same plants are functional in other matings.
      Self-incompatibility has been discovered in several varieties of Citrus.   Nagai and Tanikawa (1928) found four apparently self-incompatible varieties among twenty-eight varieties tested.   The Clementine mandarin has clearly been shown to be self-incompatible (Oppenheimer, 1948; Coste and Gagnard, 1956; Soost, 1956).   The Orlando tangelo (Krezdorn and Robinson, 1958), the Minneola tangelo (Mustard, Lynch, and Nelson, 1956), the Robinson, Osceola, Page, Lee, and Nova mandarins (Reece and Register, 1961; Hearn and Reece, 1967) all appear to be self-incompatible.   The Sukega grapefruit and hybrids of Sukega with Clementine, Orlando, and Minneola have all failed to set seedy fruits when self-pollinated although their pollen functions on other varieties and germinates in vitro (Soost, unpublished).   Miwa (1951) reported the Hyuganatsu to be self-sterile but cross-fertile.   Early reports indicated great differences in seediness of some pummelo varieties grown in different localities (Groff, 1927, 1930; Torres, 1932; Pope, 1934).   Earlier, Boyle (1914) suggested that cross-pollination was the cause for the increased seediness in pummelos.   More recent work definitely demonstrates the presence of self-incompatibility in at least several pummelo varieties (Nauriyal, 1952; Aala, 1953; Soost, 1964).   Anjaneyula (1955) reported self-incompatibility for the lemon varieties Italian, Nepali Oblong, and Lucknow.   Self-incompatibility is also indicated in C. limettioides Tan. (Singh and Dhuria, 1960) and in Malta lemon (Randhawa et al., 1961).
      Ikeda (1904-06) concluded from small-scale pollination tests that some varieties are cross-incompatible.   Of the four self-incompatible varieties reported by Nagai and Tanikawa (1928), three also appeared to be cross-incompatible.   Minneola and Orlando tangelos also appear to be intercross-incompatible (Mustard et al., 1956; Krezdorn, 1960; Soost, unpublished).   Data presented by Reece and Register (1961) indicated that Robinson mandarin may be cross-incompatible with Page and Osceola mandarins, although environmental conditions could have been responsible for the low set of seeds and fruit.
      Orlando and Minneola tangelos are hybrids of Duncan grapefruit with Dancy tangerine (Reece and Childs,1962).   Robinson and Osceola are, in turn, hybrids of Clementine and Orlando, and Page is a hybrid of Clementine with Minneola (Reece and Register, 1961).   The pattern of inheritance in these hybrids and in the hybrids of Sukega listed above indicates the presence in Citrus of a set of incompatibility alleles.   On this basis, it would be predicted that all progeny resulting from intercrossing self-incompatible varieties, such as Clementine and Orlando, would also be self-incompatible.   The presence of such an incompatible system in an insect-pollinated genus is not surprising and may be more widespread in Citrus and related genera than has generally been realized.
      Whether cross-sterility due to general unlikeness of parents occurs at all among ordinary forms of Citrus is uncertain (see chap. 5).   There is some evidence of cross-sterility between forms with different chromosome number (see chap. 5).   Toxopeus (1931) discovered a peculiar form of partial cross-sterility related to the length of the style; when certain varieties which have very short styles are crossed reciprocally with ordinary long-styled varieties, the pollen of the short-styled varieties is comparatively ineffective in producing seeds, presumably because the pollen tubes grow too slowly or stop too soon.   Torres (1938) concluded that "some varieties of different species" seem to be intersterile, but that this is due to genetic differences, not to likeness in incompatibility genes, "since citrus [forms] are generally self-fertile."   However, the accumulating evidence suggests that self-incompatibility may be involved (see above).
      Absolute Gametic Sterility.—Male or pollen sterility consists of inability to produce functional pollen, that is, pollen with sperm cells capable of fecundating egg cells.   Female or embryo-sac sterility consists of inability to produce embryo sacs with egg cells capable of development into embryos.   Absolute gametic sterility may be roughly classed as genetic or nongenetic in origin.   In the former, the hereditary constitution is such that the plant or variety is sterile (partially or completely) under conditions which are favorable to fertility in other genetic types; absence or abnormality of the reproductive organs or cells is characteristic of the variety affected.   The Washington navel orange is an example of genetic sterility.   Nongenetic sterility is due to environmental causes which are comparatively unfavorable to the reproductive processes.   Iwamasa and Iwasaki (1963) associated low temperatures with the reduction of pollen fertility in C. aurantifolia.   Brieger and Gurgel (1941) reported that certain rootstock varieties decreased pollen fertility.
      Obviously, sterility may be both characteristic of a variety and variable in that variety according to the environment.   Nakamura (1943) found temperature influenced pollen fertility in varieties that had some degree of sterility under normal conditions.
      In some varieties, part of the flowers have abortive pistils whose development has been prevented, or interrupted in various stages, perhaps by competition of other flowers or buds for food material, or by production of inhibitors.   The flowers of a fertile citrus variety may be either (1) both perfect or hermaphroditic and staminate or male (polygamomonoecious), or (2) regularly perfect (Toxopeus, 1931; Uphof, 1932; Oppenheimer, 1935; Minessy, 1954; Singh and Dhuria, 1960; Randhawa et al. 1961).   In the former class, which includes Citrus limon, C. aurantifolia, and C. medica, and in forms that appear to be closely related to these species, the pistil is regularly underdeveloped or absent in a large part of the flowers.   In the latter class, which includes C. sinensis, C. grandis, C. paradisi, and C. reticulata, abortion of the pistil is much less common.   However, it may occur frequently among the later-opening and weaker flowers.   The proportion of staminate flowers is highly variable, according to the variety and the growth conditions.   Scaramella-Petri and Strigoli (1958) reported that low temperatures (0°C to 10°C) during flower formation increased the number of aborted pistils in Poncirus trifoliata.   In varieties with comparatively few perfect flowers, these flowers are likely to set fruit in much higher proportions than in varieties in which the flowers are regularly perfect.
      The stamens, unlike the pistil, show very little tendency to general failure of development, although in one variety of navel orange which produces some pollen (Shamel, 1918) part of the stamens are often abnormal, being more or less petaloid.   Hybrids of satsuma mandarin obtained by Iwamasa (1966), using several pollen parents, did not develop normal anthers.   Anjaneyula (1953) reported the transformation of other floral parts, including stamens, into petals in Sathgudi orange, kumquat (Fortunella sp.), and lemon.   Imperfect development of part or all of the pollen mother cells or the pollen is common, however.   The ability of pollen to produce fecundation is probably better indicated by its germination in suitable culture media than by its appearance under the microscope.   However, germination in vitro is not an absolute measure of the ability of the sperm to accomplish fecundation.   Techniques for in vitro germination have been used on many Citrus varieties by many investigators (Nagai and Tanikawa, 1928; Herrero-Egaña and Labiada, 1935); Moreira and Gurgel, 1941; Oppenheimer, 1948; Plaut, 1947; Pereau-Leroy, 1950; Zacharia, 1951; Nauriyal, 1952; Yatomi, Koga, and Uchida, 1952; Mustard et al., 1956; Krezdorn and Robinson, 1958; Damigella and Squillaci, 1959; Singh and Dhuria, 1960; Ozsan, 1961; Randhawa et al., 1961; Singh and Randhawa, 1961).   Most procedures have used 20 per cent sucrose solutions at 20°C to 25°C.   Resnick [sic] (1958) indicated varietal differences in response to the concentration of sucrose.   With most varieties that he tested, germination was best in 30 to 50 per cent sucrose, but Shamouti germinated best in 15 per cent.   With lemon and citron varieties, Randhawa et al. (1961) obtained the best germination in 15 per cent sucrose.   Germination on agar rather than in liquid media has also been satisfactory.   Various materials have also been added in attempts to increase germination.   Resnik (1956) reported increased germination from additions of 2,4-D, thiamine, indolebutyric acid, or boric acid.   Singh and Randhawa (1961) obtained increased growth of pollen tubes with additions of gibberellic acid, but the percentage of germination was decreased.   Large differences in the germination percentage for a single variety have been reported by various investigators.   In addition to possible errors in varietal identification, a lack of uniformity in germinating conditions and methods of handling the pollen is indicated.
      In some kinds of plants, certain genes or combinations of genes are lethal for the development of the particular pollen grains or embryo sacs, or both, which receive them, so that these pollen grains or embryo sacs do not produce viable gametes.   Also, possession of a particular gene constitution, sometimes of one particular kind of gene, may prevent the development of stamens or pistils, or of their reproductive tissues, throughout the plant; the sterility of the Washington navel orange is of this sort, since its pollen mother cells degenerate before reduction division.
      Irregular chromosome reduction, leading to presence of extra chromosomes or lack of a part of the complete normal set, may lead to gametic sterility or embryo abortion.   This cause of sterility is mainly responsible for the nearly complete sterility in citrus triploids, and must greatly reduce the fertility of tetraploids.   Irregular reduction of chromosomes often occur in F₁ plant hybrids from wide crosses; much irregularity was found by Longley (1925) in the Eustis limequat.   Some diploid varieties of Citrus show evidence of irregular reduction (see above).
      Even with regular reduction, F₁ plant hybrids from wide crosses often produce unbalanced sets of chromosomes, sets derived partly from each parent often being seriously incomplete or otherwise deleterious.
      Defective Pollen Development (Male Sterility).—In the Washington navel orange, casual inspection suffices to show that no pollen is shed, and that the mature anthers, at the time of the opening of the flower, are cream in color, strikingly unlike the bright yellow which is produced by the pollen in fertile anthers.   At the time when the normal anthers have just shed their pollen, the Washington navel anthers, usually unopened, are shriveling and changing to a dingy brownish color.
      Microscopic examination at earlier stages shows that the sporogenous tissue degenerates before the time of the first meiotic division (Webber, 1894, 1930, Osawa, 1912).   The anthers usually develop well-differentiated pollen mother cells and tapetum before development stops (fig. 4-12), although sometimes degeneration starts at an earlier stage, in the primary sporogenous cells.   If pollen mother cells have been produced, they degenerate without cell division, some appearing to form two or three nuclei within the original cell wall, and somewhat later the tapetum also degenerates (fig. 4-12).   Remnants of pollen mother cells and tapetal cells may sometimes be found in the mature anthers (fig. 4-12) but no mature pollen is produced (Nakamura, 1934, 1943; Moreira and Gurgel, 1941).   The Washington navel therefore is completely male-sterile, and is incapable of sexual reproduction except as a female parent in crosses.
      Occasional reports of well-developed pollen in this variety probably relate to distinct genetic types rather than to normal pollen development occurring under exceptional conditions in the Washington navel itself.   The pollen-bearing Rufert orange is a selection found among nucellar seedlings of a pollenless navel variety (Ruvel) at Riverside, California (see chap. 5).
      Nakamura (1934) found that in several varieties the pollen mother cells degenerate about as in the Washington navel.   These are the Thomson, Navelencia, Surprise, and Pyriform oranges (of which the first two, at least, were derived by bud variation from the Washington navel), and the Tahiti lime.   Uphof (1931) also observed this behavior in the Tahiti lime.   Iwamasa (1966) reported the same degeneration in hybrids of satsuma mandarin with Poncirus trifoliata.
      Satsumas are often completely male-sterile, producing very low percentages of well-developed pollen.   Osawa (1912) observed that most of the pollen grains degenerated at various stages of development (fig. 4-12, F), only a few pollen grains appearing normal at anthesis.   Longley (1925) and Ozsan (1961), however, found more than 50 per cent of the pollen grains appearing normal.   Several investigators (Osawa, 1912; Nagai and Tanikawa, 1928; Zorin, 1936; Pereau-Leroy 1950; Ozsan, 1961) have also been able to obtain germination in vitro and Reece (unpublished) has produced hybrids using satsuma pollen.
      The average number of pollen mother cells in a satsuma loculus is comparatively small; the number is highly variable, and some microsporangia produce none at all.   Osawa (1912) found that the pollen mother cells sometimes degenerate, but that in most cells the reduction divisions appear normal.   Nakamura (1929), however, found irregularities of meiosis to be common, and this suggests the presence of a chromosome constitution capable of causing pollen sterility.
      In the very few-seeded Shamouti orange, Oppenheim and Frankel (1929) found that the reduction divisions in the pollen mother cells, and the formation of the microspores, usually appeared normal.   Zacharia (1951) reports about 38 per cent of the pollen is shriveled and pale, about 23 per cent is normal appearing, and about 39 per cent is normal and large.   Germination results (Oppenheim and Frankel, 1929; Oppenheimer, 1935; Zacharia, 1951; Ozsan, 1961) have been variable but usually less than 10 per cent of the pollen has germinated; Ozsan (1961) obtained a maximum of 27.9 per cent.   Zacharia (1951) found that pollen-tube growth in the style was good but suggests that the ovules may not be reached, resulting in fruit set by induced parthenocarpy.
      High percentages (40 to 60 per cent of empty pollen have been reported for Valencia orange (Longley, 1925; Uphof, 1934; Moreira and Gurgel, 1941).   Cameron et al. (1960) obtained increased seed set from cross-pollination, indicating that low seed production is partly due to lack of functional pollen.
      Marsh grapefruit has been found to have from 5 to 15 per cent well-developed pollen (Longley, 1925; Friend et al., 1939; Moreira and Gurgel, 1941; Ozsan, 1961).   Ozsan obtained a maximum germination of 3.2 per cent.   Other seedless grapefruit also have low percentages of well-developed pollen; Ozsan found about 8 per cent in Thompson.
      Nakamura (1943) found rather large proportions of defective pollen in several varieties of C. limon.   Moreira and Gurgel (1941), Pereau-Leroy (1950), and Ozsan (1961), also found much defective pollen in the Eureka variety and others.   Nakamura (1943) found many abnormal tetrads and rather high frequency of univalents in Eureka lemon.   The frequency of univalents and abnormal tetrads increased as the temperature was reduced from 19°C to 10°C.   Chen (1944) also reported abnormal tetrads in Eureka.   Naithani and Raghuvanshi (1958) found inversions in Italian long and univalents in Italian oblong.   They also note "fusion of pollen mother cells to form giant binucleate cells."   However, Randhawa and Choudhury (1960) reported high pollen germination and very few meiotic abnormalities in the seedless varieties Malta, Nepali Oblong, and Seedless lemon, or in the seedy variety Genoa.   It seems clear that much pollen sterility, perhaps caused by meiotic abnormalities, occurs throughout C. limon.
      Moreira and Gurgel (1941) found several varieties of oranges that have high proportions of defective pollen.   In addition to the varieties already discussed, they include Lima, Pera, Cadenera, and several others.   All of these varieties are low in number of seeds.
      Nakamura (1943) found many other varieties and species, particularly those that appear to be taxonomically related to C. limon, that show a high degree of pollen sterility.   Among these are Otaheiti, rough lemon, Bouquet sour orange, and C. junos 'Kizu.'
      Pollen sterility of varying degree has therefore been found throughout the genus Citrus, culminating in the conditions found in varieties where no functional pollen is produced.
      Defective Embryo-Sac Development (Female Sterility).—Osawa (1912) found that both in the satsuma and in the Washington navel orange the megaspore mother cell or the embryo sac very often degenerates.   Sections of the ovule of the Washington show that its undivided embryo-sac mother cells often become shrunken with degenerating cytoplasm, which stains deeply.   More often, however, with both varieties, such changes are not noticeable until after the row of four megaspores has been formed (fig. 4-13, A) or even until one of the spores has divided to form the embryo sac (fig. 4-13, B).   A few embryo sacs reach full development and contain egg cells capable of fertilization, as is proved by the production of seeds and of seedlings, including occasional hybrids, by cross-pollination.
      Since the other tissues of the ovule develop normally, the degeneration is not a matter of generally poor development of the ovule, but must depend on some more specific tendency to poor development of the reproductive tissue.   Neither is the abnormality a matter of irregular chromosome reduction, since degeneration is often noticeable in the prophase of the first meiotic division.   It is presumably due to the same general genetic cause as is the degeneration of the pollen mother cells in the same varieties.   In this connection, it seems significant that in both anther and ovule the defective development is more extreme in the Washington navel than in satsumas.
      Apparently very few functional ovules are formed in the Tahiti lime.   Although Uphof (1931) reported that embryo sacs are not formed, occasional viable seeds are produced (Reece and Childs, 1962).   Bacchi (1940) reported the Tahiti lime to be a triploid, but Reece and Childs (1962) indicate the chromosome number to be diploid.
      Although the low percentage of functional pollen is a major factor in the low seed number of the Shamouti orange, the embryo sac is absent or abortive in a large part of the ovules (Oppenheim and Frankel, 1929; Zacharia, 1951).   Considerable female sterility is also indicated in the Valencia orange by the low seed set obtained from cross-pollination (Cameron et al., 1960).   Undoubtedly, other orange varieties that have few seeds, such as the Cadenera, also have much female sterility as well as pollen sterility.
      The low seed set obtained from cross-pollination indicates a high degree of female sterility in the Marsh grapefruit.   Other seedless varieties of grapefruit also must possess much female sterility.

Embryo Abortion
      Functional gametes may be produced, fecundation may occur, and yet the fecundated egg may be unable to develop into an embryo capable of germination.   If no embryos are able to develop, complete zygotic sterility results.   This sterility, like gametic sterility, may be either genetic or nongenetic in origin.   The fate of the fecundated eggs, however, does not tell the whole story with respect to embryo production in Citrus and its nearest relatives, since many varieties are capable of producing embryos by an asexual process.   The relation of nucellar embryony to sterility must therefore be considered.
      The occurrence of empty seed coats and nearly empty seed coats suggests embryo abortion.   Seeds which produce only nucellar seedlings indicate failure or suppression or possible absence of zygotic embryos.

Relation of Nucellar Embryony to Sterility
      Evidence already discussed indicates that absence of pollen usually, if not always, prevents the development of nucellar embryos.   Complete male sterility, without cross-pollination, therefore results in seedlessness.   Also, since the nucellar embryos develop in the embryo sac, presumably such embryos cannot be produced unless the embryo sac is well developed; and it is probable that fecundation of the egg or union of the polar nuclei and sperm is usually necessary.   On the other hand, since certain varieties produce nucellar seedlings mainly or exclusively, and in other varieties many seeds produce nucellar seedlings only, it is probable that abortion of the zygotic embryo does not necessarily prevent the development of nucellar embryos.   In brief, then, it is probable that gametic sterility usually necessitates nucellar sterility, but that abortion of the gametic embryo does not.

Sterility of Diploid First-generation Hybrids
      The complications introduced by nucellar embryony and by the prevalence of sterility in ordinary varieties make it difficult to obtain adequate evidence on the relation of known hybridization to sterility.   In any adequate attempt to evaluate the fertility or sterility of particular selections, it is necessary to consider four features, namely, fruitfulness, seediness, number of embryos per seed, and the relative proportions of zygotic and nucellar seedlings.   The possibility that nucellar embryos may develop in spite of complete gametic sterility must also be kept in mind.
      First, it is certain that many fertile hybrids are produced from interspecific and even intergeneric hybrids.   From open pollination of fifty-six recent F₁ hybrids between four varieties of Citrus and Poncirus trifoliata (Cameron and Baines, unpublished) mainly zygotic seedlings were obtained from forty-three, and mainly apparent nucellar seedlings were obtained from thirteen others.   Altogether, this indicates a high degree of female fertility in these intergeneric hybrids.   Pollen fertility has not been checked.   In many hybrid populations between various Citrus species at Riverside, California, most individuals have a high degree of fertility as indicated by fruit set, seed number, and pollen viability.   Often the proportion of evident nucellar embryos is very high, making it impossible to determine generative female fertility.   However, many individuals produce some evident zygotic seedlings.   It is also clear that the amount of fertility is inherently variable among F₁ hybrids from the same cross.   Some hybrids involving parents that exhibit much sterility are highly sterile, exhibiting degenerated pollen, low seed production, and poor set.   Other individuals are decidedly seedy, although most of their seedlings are of nucellar origin.   Finally, it is obvious that the presence of nucellar embryony is involved in producing the appearance of female sterility.   In populations segregating for polyembryony (see chap. 5), monoembryonic individuals exhibit a high degree of generative fertility, but zygotic seedlings are low in number or lacking in progeny from polyembryonic individuals.   Such a close correlation between generative sterility and polyembryony is not likely.   Many individuals that produce nucellar seedlings almost exclusively also are highly pollen fertile.
      Zygotic progeny produced from self-pollination or cross-pollination of many closely related varieties are mainly very weak, some never producing flowers.   However, some individuals equal or surpass either parent in vigor of growth and seem fully fertile.   Hybrid progeny from crosses within the species C. reticulata are generally more vigorous than progeny from crosses within the other species to which most of the cultivated varieties are assigned.

The Horticultural Importance of Sterility
      The one general important horticultural advantage of sterility lies in the resulting absence or scarcity of seeds.   Because of the location of the seeds and the usual methods of using the fruits, seedlessness is very desirable from the viewpoint of the consumer.   In California, all of the main commercial varieties are few-seeded or seedless, and this characteristic unquestionably determined the adoption of the Marsh grapefruit.   Fahey (1940) reports that seeded varieties of grapefruit in Trinidad are practically not salable since the introduction of seedless varieties.   Low seediness has also been an important factor in the adoption of the Washington navel, Shamouti, and Cadenera oranges, the Tahiti lime, and satsuma mandarin.   The undesirable increase in seeds following cross-pollination of some varieties has led to suggestions that these varieties should be isolated from other varieties.   However, the lack of cross-pollination results in severe reductions in yields in at least some of these varieties (Clementine, Orlando, Minneola, and pummelo varieties).
      As previously discussed (p. 298), failure to set seeds often tends to prevent or reduce the setting of fruit.   Sterility therefore tends to be a disadvantage with respect to yield.   It may in some varieties, however, be advantageous by preventing excessive setting of fruit and consequent small size of fruit, overloading of trees, and alternate bearing.   Washington navel orange and satsuma mandarin exhibit much less alternation of yield than do mandarin varieties which have many seeds.
      Embryo abortion and perhaps gametic sterility may affect the degree to which a stock variety "breeds true" by nucellar embryony.   For rootstock use, it is, of course, desirable to have uniformity in the seedlings.   If sterility limits the number of seeds produced, it can limit the usefulness of a variety that would otherwise be a valuable rootstock.
      Low generative fertility is obviously a serious hindrance to the use of a variety in hybridization.   Complete or even a high proportion of pollen sterility is especially troublesome when, as in the navel oranges and the satsuma mandarins, the proportion of nucellar embryos is very high, such varieties are difficult or impossible to use as pollen parents, and when they are used as seed parents their seedlings include only small proportions of hybrids.   For the search for usable variations among nucellar offspring, "nucellar fertility" is necessary, and perhaps, therefore, some degree of gametic fertility also.   The unfavorable effect of seedlessness on the setting or retention of fruit is also a hindrance in the production of usable new seedless varieties.   However, the desirability of having seedless varieties has led to the breeding of triploid varieties (see chap. 5).

CYTOLOGICAL METHODS FOR CITRUS

LITERATURE CITED