Genetics, Breeding, and Nucellar Embryony



      This chapter summarizes evidence on heritable and nonheritable variation in citrus from studies carried out principally during the past seventy years.   Considerable description of the physiological effects of nucellar embryony is included.   Mutation, bud variation, and chimeral relationships, all prominent in many citrus varieties, are discussed in some detail.   The goals and results of some citrus breeding programs are described.
      In Chapter 10, Volume I, of the first edition of The Citrus Industry (Webber and Batchelor, 1943), A. D. Shamel presented extensive data on bud variation and bud selection in citrus.   Certain portions of his material are incorporated in this chapter.


      Variation in living organisms may be either genetic or nongenetic.   Differences such as those distinguishing the Valencia orange from the Washington navel orange are inherent in the genotypes of the trees; thus, trees budded from a Valencia regularly remain Valencia in type.   On the other hand, Valencia trees are seldom phenotypically identical, although most of the differences among them are not heritable.   Nonheritable variations are usually produced by differing conditions in the environment such as climate, nutrition, and disease.
      In vegetative propagation by such means as budding, grafting, or nucellar embryony, the genetic constitution of the progeny is usually identical to the parent because the somatic cells of a plant commonly have the same constitution.   Sexual reproduction, in contrast, recombines the genes and usually produces numerous genetic differences between parent and progeny.
      During the development of a plant or animal, differentiation of tissues and organs systematically occurs despite the genetic similarity of somatic cells.   Development is principally controlled by the actions of genes and gene products; it depends also on the interrelations among parts, such as position and arrangement of cells.   Recent studies indicate gene actions are highly selective in their timing and localization, so that critical processes are induced or repressed in orderly sequence.
      Heritable variations are the basis of genetics and breeding, but their analysis is greatly complicated by the problem of distinguishing them from nonheritable ones.   In citrus, climatic factors are very effective in producing modifications among like parts of a tree, such as individual fruits or leaves.   Light, temperature, and wind are of particular importance.   Both environment (fig. 5-1) and the long period of juvenility following seed reproduction can markedly modify characteristics important in citrus breeding, such as size, shape, yield, and flavor of the fruit.   Adequate evaluation of the genetic characters of a new variety, for example, requires that it be observed in various environments over a long period of time.
      Modifications due to virus infection have become of primary importance in citrus.   The number and variability of virus symptoms often make it very difficult to define the effects of other factors, genetic or nongenetic.   This is especially true with respect to the analysis of age changes, which are discussed below.


General Biological Considerations
      Age changes must be considered with respect both to individuals and clones.   Changes in individual organisms accompanying changes in age are widely recognized and have been the subject of many studies.   Because of the nature of the animal body, with its usually determinate growth and highly differentiated parts, much more is known concerning age changes in animals than in plants.
      Minot (1908) concluded that normal cell differentiation in animals tends to go too far, with injurious results.   Bidder (1932) considered mechanisms limiting final size in vertebrates at least indirectly responsible for senescence.   Later workers have demonstrated many conditions contributing to the aging process.   Comfort (1964), in a comprehensive review of animal senescence, concluded that no single explanation is widely applicable.   He defined senescence as "a change in the behavior of the organism with age, which leads to a decreased power of survival and adjustment."   Some important characteristics usually associated with aging (and at least partly verified by experiments) are cessation of growth, a high degree of differentiation, and sometimes the speeding up of metabolic rates.
      Age changes in plants have proved even more difficult to explain than those in animals.   As reviewed by Sax (1962), various species show great differences in their life spans, ranging from thousands of years in Sequoia and certain pines to a few weeks in some annuals.   In annual plants, death systematically follows flowering and seed maturation, although it can often be delayed by the prevention of these stages.   Thus, Leopold, Niedergang-Kamien, and Junick (1959) showed that soybean plants with their flowers removed lived nearly three months longer than plants which matured seed.   But Lockhart and Gottschall (1961) found that the Alaska pea, despite deflowering, finally differentiated an apical node that no longer contained a vegetative meristem.
      In perennials propagated asexually, there has long been uncertainty as to the meaning of changes occurring with age.   Knight, as early as 1795, held that clonal senescence did occur.   Other early workers concluded that there is no unavoidable aging process (Möbius, 1897; Winkler, 1920).   Yet Molisch (1938), after extensive consideration of the evidence, concluded that changes due to old age occur in clones, and that sexual reproduction produces a rejuvenation which does not result from artificial vegetative propagation.
      In considering the problem of age changes in citrus, we should differentiate between early effects of juvenility in seedling selections and possible old-age effects occurring after reproductive maturity has been reached.   It is clear that a juvenile period exists as a part of the life cycle of many plants, including citrus.   Robbins (1957a), Brink (1962), Sax (1962), and Doorenbos (1965) have reviewed the subject.   Differences between juvenile and adult leaf shape are marked in certain eucalyptus and juniper varieties.   In the beech, old trees shed their leaves in autumn, but young seedlings retain them in a dried condition.   The trunk region and root system remain juvenile after the upper parts of the tree have passed into the leaf-shedding phase.   In apples, leaves of one-year-old seedlings are thin and lack pubescence (Stoutemyer, 1937); these characteristics change as the seedlings grow older.   However, studies by Wellensiek (1952) showed that with certain mature woody plants, adventitious shoots produced on disbudded stems displayed some juvenile characters.
      The English ivy (Hedera hèlix) provides evidence for contrasting juvenile and adult stages, the latter being sometimes reversible (Robbins, 1957b; Stoutemyer and Britt, 1961).   The juvenile form, which is a trailing vine without flowers, can change abruptly to an upright form bearing inflorescences.   Seedlings always first display the juvenile condition.   Reversion from adult to juvenile form has been observed following pruning, grafting of adult scions onto juvenile stocks, and treatment with gibberellic acid.

Changes Associated with Age in Citrus
      In citrus, juvenile characters are very prominent and often persist for a long time.   They include thorniness, vigorous and upright habit of growth, slowness to fruit, alternate bearing in early years, and physical differences in fruit characters.   In early years, the most striking character is thorniness.   Swingle (1932) emphasized thorniness in discussing the marked rejuvenation which occurs in citrus after seed reproduction.
      Thorniness.—Most species and varieties of Citrus produce thorns, which are imperfectly developed branches (Uphof, 1935a).   Occasionally a vigorous thorn shows its true nature by producing leaves and even flowers, and rarely a thorn may be replaced by an ordinary leafy shoot which develops at the same time as the thorns of neighboring nodes (fig. 5-2).   This shows that thorns are rudimentary lateral branches which develop during the same growth cycle as the stem that bears them, whereas ordinary lateral branches develop in a later growth cycle.   Very vigorous trees and shoots are especially likely to be thorny.
      On ordinary trees of most old budded varieties, thorns are few and small, although in the Lisbon lemon, for example, they are often pronounced enough to cause annoyance.   Young nucellar seedlings, grown from such old varieties, show thorniness which is many times as great as on the parent tree.   The immediate budded progeny of these seedlings may also be very thorny.   The thorniness of young zygotic seedlings and their budded progeny is variable.   It is usually high, however, and on the average is similar to that of young nucellar individuals from the same parent varieties.   The following considerations apply in general to both kinds of plants.
      As thorny seedlings become older, much of the new growth, especially high in the top, is decidedly less thorny.   If bud sticks are cut from top shoots low in thorns, they are likely to produce trees which are not highly thorny.   In early studies at Riverside, this procedure often gave nearly thornless trees within ten to fifteen years from the germination of the seed, but the reduction differed greatly with different varieties.
      The lower portions of a seedling tree retain, perhaps indefinitely, the ability to produce very thorny shoots.   The physiological change that causes the decrease of seedling thorniness thus cannot depend simply on the age of the tree or clone from seed; it may be favored rather by repeated cell division and perhaps by the position of the shoot.   Thornlessness perhaps also can originate by somatic mutation.
      Detailed evidence on seedling thorniness was obtained by Frost (1938) in breeding studies beginning about 1915.   The correlation in relation to certain parental types was marked.   With the satsuma mandarin, which is low in thorns, young nucellar seedlings and their budded progeny usually had little thorniness, whereas in sweet oranges thorniness was high and especially persistent.   Certain hybrids of the Bouquet sour orange seemed to produce no thorns at all.   Many hybrid and nucellar seedlings from various species were grown in crowded plantings, at 2- by 5-foot spacing, and were repeatedly pruned back severely for about ten years.   Most of the seedlings were originally very thorny, and at the end of the period thorniness of the new growth seemed undiminished.   This was true also of budded trees from some of the seedlings, handled in the same way.
      An orchard planting at wider spacing was later made with other budded trees from these seedlings; this planting was not cut back appreciably.   After a few years, the newer growth, especially on high branches, was much reduced in thorniness.   When a second budded generation was propagated from some of these trees, after three to five years, budwood selected for thornlessness gave less thorniness in the nursery than had the earlier budding.   A few original hybrid seedlings, first held in close planting until sixteen years old, were also severely pruned and set in the orchard.   In the first years thereafter, these trees were much more thorny than second-budded-generation trees earlier derived from them.   Similar trends in thorniness have since been generally observed by citrus breeders.
      Figure 5-3 shows the contrasting degrees of thorniness in nursery budlings of Kara mandarin, resulting from selection of thornless and thorny budwood from the same tree.   Figure 5-4 illustrates low thorniness in nursery trees budded from Minneola tangelo, contrasted with high thorniness on budlings from a recent nucellar seedling of Minneola.
      Frost (1952) contrasted thorniness in old budlines and recent nucellar selections of fifteen citrus varieties.   Nursery budlings were propagated in 1930 from old orchard trees and from the first budded generation of selections about fifteen years of age from seed.   The seedling selections had usually arisen from the same trees which served as old budline sources.   Budwood from the seedling selections was taken from both high and low positions.   At three years of age, resulting trees from almost all of the seedling selections were much more thorny than those from the old budlines.   Trees from low budwood were more thorny than those from high budwood.   Some of the data are shown in table 5-1.
      Other workers, in studies on flowering and fruiting in young nucellar selections, have emphasized the thorniness of such material.   Furr, Cooper, and Reece (1947) observed heavy thorns on shoots of two-year-old Jaffa orange seedlings that had developed from buds inserted into branches on older Valencia trees; the Valencia branches themselves remained low in thorns.   Cooper et al. (1958) in Texas, reported that four-year-old budded trees from young nucellar selections of Webb Redblush grapefruit were more thorny than trees budded at the same time from the parent budline.   Bitters, Batchelor, and Foote (1956) described plantings of several Valencia selections involving old budlines and recent nucellar seedling sources.   Again, thorniness seemed inversely related to age of the selection from seed; recent selections such as the Campbell and Cutter Valencias showed rather high thorniness.
      Valencia and Washington navel oranges propagated by stem cuttings from old clonal sources by Halma (1931) gave trees which for several years were more thorny than budded trees, but less thorny than seedlings.   Frost (unpublished) propagated sprouts from roots of some of these stem-cutting trees; the resulting root-sprout trees were initially as thorny as seedlings, but thorniness declined rapidly.   Two of these Valencia trees (one stem cutting and one root sprout) were planted next to a budded tree of Frost nucellar Valencia in 1933.   By 1963, none of these thirty-year-old trees showed appreciable thorniness on peripheral branches, but recent, low, interior sprouts on the nucellar selection were thorny, while such sprouts on the other two trees were nearly thornless.   Budded repropagations made from these trees in 1958 again showed an initial higher thorniness in the nucellar budline.
      Growth Rate, Flowering, and Fruiting.—Important differences in growth rate and tendency to bloom, between old clonal and young nucellar selections, were early emphasized by various workers (Swingle, 1927, 1932; Tanaka, 1927a; Hodgson and Cameron, 1935, 1938; Frost, 1938).   Swingle suggested that vigorous growth, like thorniness, may be a persistent effect of special conditions occurring during the development of the embryo.   In the study by Frost (1952), the average cross-section area of trunk three years after budding was clearly greater in the nucellar budlings than in the propagations from the seed parents in nearly all of fifteen varieties.   After eighteen years, many of these same nucellar trees, then in an orchard planting, showed excesses of 33 to 108 per cent in trunk area compared to the parent clones (Cameron and Soost, 1952).   Tendency to early flowering and fruiting among the three-year-old trees was much lower in the nucellar selections (see table 5-1 for examples).   Among eighty-five budlings of this group, only nine bore fruit, while among eighty-two from the parent budlines, sixty-five carried fruit.
      Furr et al. (1947) attempted to induce early flowering in juvenile citrus varieties by various means, including budding, inarching, stunting, and girdling.   Most of these treatments were unsuccessful, but a few seedlings and shoots girdled in their second year from seed flowered the next year.   Many seedlings girdled in their sixth year flowered in the next season.
      Furr (1961) compared time of first flowering and fruiting in pairs of budlings propagated from basal and top shoots of seedlings which had already fruited.   Among thirty pairs of trees of Rangpur lime and Iran lemon, eleven trees from basal buds and nineteen from top buds flowered two years after budding; in the next year, nearly all trees from both bud positions bore fruit.   With grapefruit, however, which is slower to come into fruit, much more early fruiting occurred from top buds than from basal ones.   Similar behavior with respect to early fruiting of nucellar Lisbon lemon budlings was reported by Cameron, Soost, and Frost (1959).
      Other tendencies characteristic of young seedling selections are marked biennial alternation of flower and fruit production and irregular distribution of the fruit on the tree.   These differences seem to be mainly due to more extensive inhibition of flowering and fruit setting by preceding fruit production and perhaps by effects of wind and shade on the foliage.
      Fruit Characters and Yield.—Fruits of early propagations of seedling selections have a tendency to elongated shape, puffing, hollowness of axis, and low seed number (Frost, 1938; Hodgson and Cameron, 1938).   These characteristics are usually modified in older trees, especially after repeated repropagation.   Batchelor and Cameron (1949) and Cameron and Soost (1952) found that fruit size and shape, and percentages of juice, soluble solids, and acid were generally the same in eighteen- to twenty-year-old budded trees of nucellar selections as in the comparable seed-parent budlines.   Some fruit characters, however, still showed differences.   Among seven varieties which produce appreciable numbers of seeds, seediness still tended to be lower in the nucellar selections of five, including the Marsh grapefruit.   Later (Cameron, Soost, and Frost, 1959), this difference was still apparent in the Marsh.   Another character showing persistent difference is size of navel structure.   Swingle (unpublished) grew several hundred nucellar seedlings of the Washington navel; in none did the fruit entirely lack the navel structure, but there was a decided tendency for smaller navels than in the seed-parent trees.   Seedling selections of the Washington navel and Ruvel (a navel orange) made by Frost from 1915 to 1917 showed much smaller average navel structures than their seed parents even as late as 1951 (Cameron and Soost, 1952).   Smaller navel aperture has also been reported by Oberholzer and Hofmeyr (1955) in a nucellar Washington navel in South Africa.
      In the Washington navel and the Marsh, certain undesirable differences in physical characters of the fruit have tended to persist in nucellar selections (Soost and Cameron, 1961b.)   With the Washington, an irregular tendency to tapered stem ends and early looseness of core has existed, particularly in seasons conducive to large, slightly rough fruit.   Similar tendencies, as well as a frequently thicker peel, have been observed in the grapefruit.   With such characters, it is difficult to differentiate between juvenile effects and possible slight genetic changes.   Somatic mutations are frequent in citrus (see [below]).   Since most of them are unfavorable, they must constantly be guarded against.
      The most useful horticultural characteristics of most nucellar selections are high tree vigor and high fruit yield.   Higher yields were shown in nucellar selections, as contrasted to parental budlines, in nine varieties out of ten in a study of Cameron and Soost (1952).   In selections much younger from seed, however, yields are sometimes poorer than in trees from old budlines, which is due partly to the tendency to alternate bearing.
      Gardner and Reece (1960) reported on characteristics of twenty-eight navel orange selections, budded on sour orange rootstock in the same orchard.   These included nine Washington navels derived from nucellar seedlings in 1909 and 1910.   The budded trees were rated for yield at 11 through 14 years from planting.   Eight out of nine nucellar selections appear to have been among the better yielding group, whereas only nine out of nineteen of the other sources were high in yield.   Most of the nucellars also ranked within the top one-half of the group in tree size.
      Marloth, Basson, and Bredell (1964, and unpublished) reported preliminary results of a study of Valencia oranges in South Africa that included two nucellar selections (Olinda and Frost) from California.   In the first few years of bearing, these selections did not show a greater yield than two old budlines, but over an eight-year period they averaged a greater yield than the latter.   The Olinda and Frost showed marked alternation in bearing during part of the period.   They also showed low total soluble solids at first, but improved notably in this respect after the third year.
      Earlier workers sometimes attributed other character differences to seedling selections of normal nucellar origin.   Greater frost hardiness, larger fruit size, and added resistance to disease have been reported (Traub and Robinson, 1937), but these can often be indirect or temporary effects, dependent on greater seedling vigor.   Cooper, Olson, and Shull (1959) found that yellow-vein chlorosis, associated with a high water table, did not occur on virus-free nucellar selections, but was severe on old budlines affected by viruses, The difference appeared to be due to the weak root systems of the virus-affected trees.

The Question of Clonal Senescence
      It is uncertain whether any of the more persistent differences between nucellar citrus selections and their seed parents reflect a physiological aging in the parent clones beyond that associated with sexual maturity.   Various differences may be secondary effects of continued nucellar vigor of growth.   Virus infections not showing acute symptoms can depress growth (for example, see Calavan and Weathers, 1959); injurious effects associated with the stubborn disease complex have been particularly variable.   In both plants and animals, tissue-culture studies suggest that somatic cells can reproduce and thrive indefinitely if the environment is favorable enough.   As mentioned earlier, certain plant clones in nature appear to be of extreme age.   Cottam (1954) described a form of aspen which seems to have spread only by asexual means for perhaps 8,000 years.   Hall (1929) described an ancient clone of tulip, a sterile polyploid, which was known in Italy in 1606 and still exists.
      Plant forms which can maintain clonal life over very long periods usually do so by resprouting from crown, root, or underground stem tissue.   This is true of such plants as olive, banana, date palm, sugarcane, and chrysanthemum.   Trippi and Montaldi (1960) have postulated that the degeneration which often occurs in sugarcane clones as they grow older arises partly from the practice of propagating from the upper, and perhaps physiologically more specialized, portions of the stalks.   They suggest that basal cuttings may give more vigorous plants that are less susceptible to disease.
      In citrus, the existence of clear juvenile characteristics, and the evidence for associated, slower-changing characters, make it conceivable that continued physiological aging can take place.   Brink (1960, 1962) has discussed the increasing evidence in plants for a type of cellular process called paramutation by which the classical genes may sometimes interact with other chromosomal components during somatic development, so that somatically permanent changes occur both in the chromosomal component and in the cellular end products of the gene action.   In some cases, chromosomal components are so modified that the change persists in subsequent sexual generations.   Brink postulated that phase change, including sexual maturity, in plants may be dependent on a normal form of this mechanism.   If this is so, additional chances might also be brought about by such causes during the life of a clone.

Nucellar Embryony and Citrus Variety Improvement
      The information derived from long-term studies of the effects of nucellar embryony has had an important impact upon commercial citrus production in the United States and elsewhere.   The investigations described on the preceding pages, and the extensive information developed by plant pathologists on the behavior of citrus viruses, have led to widespread use of selected nucellar budlines for the establishment of vigorous, relatively virus-free citrus orchards.   Since most citrus viruses are seldom seed-borne, nucellar-seedling budlines have made it possible to establish desirable rootstock-scion combinations which would often fail if older, virus-infected bud sources were used.   The superior health and vigor of these combinations commonly result in longer-lived, higher-yielding trees, even though some problems associated with juvenile characteristics still remain in a few varieties.
      In California, the majority of all orange and lemon trees now (1968) propagated by nurserymen, and a substantial proportion of the grapefruits and mandarins, are derived from budlines of known nucellar origin.   In many other citrus-growing areas of the world, including South Africa, Central and South America, and various Mediterranean countries, the commercial use of such budlines is rapidly increasing, both by the introduction of California selections and through the development of local sources.
      The pioneering research of W. T. Swingle, H. B. Frost, H. S. Fawcett, R. W. Hodgson, and L. D. Batchelor, together with studies by later workers representing the combined disciplines of horticulture, plant breeding, and plant pathology, can be largely credited with making this development possible.


General Principles
      Genetic variations originate through several processes that are fundamentally distinct.   Most genetic variations depend on new combinations of genes, which are carried on the chromosomes.   New combinations in the offspring are either the direct result of hybridization or arise from segregation and recombination of genes within a hybrid parent.   Recombinations are usually produced abundantly in sexual reproduction, which is the main reason seedling fruit trees differ genetically from their parents.   Occasionally, change occurs in the nature of the gene; this has been called "point" mutation.   Changes resembling point mutations in their effects on the organism can be produced by chromosome aberrations, or by abnormal segregation of chromosomes, leading to loss, duplication, or rearrangement of genes.   Somatic segregation signifies gene segregation in somatic cells; this may occur in an occasional cell by unusual crossing-over, abnormal chromosome disjunction, or other exceptional processes.   Finally, tissues of two different genetic types may become combined in the same plant, either side by side or one within the other, forming a chimera.
      The processes listed above may be grouped into three categories: (1) the regrouping, duplication, or loss of unchanged genes within cells; (2) change in the nature of the gene; (3) the combination or rearrangement of cells of two or more genetic types in one individual (chimeras).
      If a genetic variation first occurs in the germ cells, it is a gametic or germinal variation; if it arises in the soma or body of a plant, it is a somatic or bud variation.
      A large body of evidence shows that heredity, in higher organisms at least, depends primarily upon the chromosomes.   Chromosomes have also been identified in certain fungi and bacteria, where they had formerly been undetected.   However, certain forms of extra-chromosomal inheritance are likewise well established.   Transmission through the cytoplasm of abnormal plastid characteristics, such as albinism, is well known.   Numerous cases of cytoplasmically controlled male sterility are also on record.   Evidence for phenomena such as paramutation is more complex and not yet fully understood.

Genes and Chromosomes
      The concept of genes as macromolecules, or combinations of such molecules, arranged in linear order on the chromosomes, remains important to genetics.   In somatic cells of diploid higher plants, the chromosomes are present in duplicate; the two chromosomes of a pair, which are similar in their gene content, are homologous chromosomes.   When a somatic cell divides, each chromosome separates longitudinally, and each of the two daughter cells normally receives chromosomes and genes identical with those of the mother cell.   In two cell divisions closely preceding the formation of the gametes, in contrast, the two homologous chromosomes pair and then separate so that each gamete receives only one chromosome from each pair, and the total number of chromosomes is reduced to one half.   These two reduction divisions constitute meiosis.   When an egg and a sperm unite, the full somatic number of chromosomes is restored; the embryo thus usually receives half of its genies from its mother and half from its father.
      The process of chromosome duplication—including gene duplication—has become much better understood in recent years.   Precisely arranged linear sequences of chemical units are assembled within the cell to reproduce exactly the already-existing chromosome strands, apparently with the latter serving as templates.   Sequences of units, which correspond to functional locations of genes, can sometimes be delimited.   Some gene effects are controlled by complex loci that behave as though they were composed of several subunits.
      Genes situated at corresponding positions (loci) in homologous chromosomes are allelic.   When the chromosome pairs segregate at meiosis, such genes, of course, also segregate.   Genes at different loci on the same chromosome (linked genes) can normally be separated only through crossing-over, which is an exchange of parts occurring between chromosomes of the same pair during the reduction divisions.   Nonhomologous chromosomes, together with the genes they carry, recombine freely during the reduction division.   New genetic combinations can occur in progeny by such regrouping of genes in either or both parental gametes.
      If two allelic genes are identical, the individual is homozygous for this gene pair.   Each gamete then receives one of these two genes; all the gametes, and all offspring from selfing, are thus alike with respect to this pair of genes.   If a plant is homozygous for all its pairs of genes, it normally produces only one kind of gamete.   Upon selfing, its offspring will all be genetically identical, that is, it will "breed true."   If allelic genes are not identical, the individual is heterozygous for this gene pair; half of the gametes will receive one kind of gene and the other half will receive the other kind.   Such a parent will not breed true; its offspring from selfing will differ in any characters that depend on the differences between these two genes.
      The genes of an organism jointly direct its life processes and, within limits, determine its characteristics.   Genes can best be identified and studied when one or a few gene differences produce clear differences in the organism.

Hybridization, Hybrid Vigor, and Inbreeding
      The word hybrid as used here refers to forms which are heterozygous as a result of interbreeding of unlike parents, regardless of remote ancestry.   Probably all citrus forms are somewhat heterozygous, because of ancestral mutation and crossbreeding.   In many kinds of plants, different species within a genus will not intercross or if they do they may produce sterile hybrids.   In Citrus, however, interfertility of species is usual, and considerable fertility of the F1 hybrids is common.
      The genera Poncirus and Fortunella (the kumquat) can be crossed with Citrus and some of the hybrids are somewhat fertile.   But Swingle and Robinson (1923) found it difficult to cross Poncirus with Fortunella, and nearly all the hybrids were weak.   Citranges have been crossed with Fortunella, giving some vigorous hybrids.   Toxopeus (1936) obtained weak hybrids from the cross of Citrus aurantifolia (Christm.) Swing. with Murraya paniculata (L.) Jack.   Traub and Robinson (1937) described a hybrid between Eremocitrus and a citrange.   Efforts at Riverside to cross Severinia with Citrus have been unsuccessful.
      High variability of zygotic progenies seems characteristic of at least all citrus forms which produce nucellar embryos.   Variability occurs whether the parents belong to the same or different species or to distinct genera.   Some of the first citranges produced by controlled hybridization differed among themselves more widely than do various varieties within the parent species (Swingle, 1913).   Similar variability has been generally observed by citrus breeders.   Most characteristics are affected, including size, shape, and vigor of tree, susceptibility to disease, and many fruit characters.   Although a variably intermediate condition is the most common, certain characters may closely approximate those of one parent, or even be outside the parental range.
      There is evidence for hybrid vigor and inbreeding depression in citrus, although critical proof is lacking.   Webber and Swingle (1905) reported citranges to be more vigorous than nucellar progeny of the parent species; citrange crossed with grapefruit also gave high vigor, as did kumquat with citrange.   The authors have obtained small populations of C. grandis (L.) Osbeck X P. trifoliata (L.) Raf. (trifoliate orange) and of a zygotic hybrid (C. sinensis [L.] Osbeck X C. paradisi Macf.) X trifoliate orange, which are highly vigorous.   Evaluations of vigor in young seedlings must take into account the effects of competition from nucellar embryos when these are present.
      Within Citrus, it appears that wider crosses are the more likely to produce vigorous offspring, but exceptions are not uncommon.   Interspecific crosses involving C. grandis, like intergeneric ones, appear to be unusually fast-growing.   Frost (1943), in limited tests, reported weak hybrids from the narrow crosses of Ruby X Valencia orange and Eureka X Lisbon lemon, and Torres (1936) found that even the wider cross, sour orange X rough lemon, gave hybrids which were weaker than either parent.   Within the diverse mandarin group, various crosses have given considerable numbers of vigorous hybrids, yet at the University of California Citrus Research Center, Riverside, present studies suggest that backcrosses and "near" backcrosses among mandarin types are not as vigorous as second-generation hybrids involving two or three Citrus species.   New hybrids of Eureka lemon with Lisbon lemon (R. K. Soost, unpublished) show poor average vigor.
      Clear evidence on the effects of selfing is difficult to obtain, since many polyembryonic varieties give few zygotic seedlings.   Self-incompatibility, present in some citrus forms (see chap. 4) further interferes with selfing studies.   Weakness of variants from selfing or presumed selfing (fig. 5-5) was reported by Webber and Barrett (1931) Webber (1932) Toxopeus (1936) Torres (1936) and Frost (1943).   At Riverside, California, recent orchard plantings of sexual progeny from selfing of a hybrid of C. grandis (L.) Osbeck X C. reticulata Blanco are not as vigorous as F1 or advanced-cross trees involving comparable parents.   J. R. Furr (unpublished) has obtained populations from selfing of the zygotic Temple, and many of the plants seem initially vigorous.   Temple (tangor?) is apparently a hybrid, and perhaps its selfs (and those just cited from Riverside) are not as weak as selfs within a homogeneous and polyembryonic group such as the sweet oranges.

Nucellar Embryony and Heterozygosis in Citrus Evolution
      Nucellar embryony must have had an important relationship to natural and artificial selection in the evolution of Citrus.   Natural selection no doubt acted not only on trees, but on embryos during seed formation.   To be successful under natural selection, citrus forms had to possess adequate growth vigor and climatic adaptation and also satisfactory capability of seed reproduction.
      Abundant heterozygosis is present in most citrus forms; the agencies responsible for this are presumably: (1) frequent gene mutation; (2) cross-pollination within a wide range of interfertility; and (3) nucellar embryony.   Reproduction by nucellar embryony preserves any heterozygosis that may originate by hybridization or mutation.   It thus must have favored accumulation of mutant recessive genes, including those detrimental to vigor and fertility.   When such genes became numerous, they would have tended to eliminate successful sexual reproduction, especially by selfing.   Hybrids between forms not closely related would have had better prospects of survival, because of greater freedom from homozygosis.   Nucellar embryony, by providing for asexual reproduction, acts as an isolating agent which should favor evolutionary differentiation.
      Even without asexual reproduction, mutation and frequent cross-pollination should tend toward the accumulation of unfavorable recessive genes.   If nucellar embryony should then originate, forms with this character would probably be preserved by natural selection.   But excessive evolutionary increase in the number of embryos per seed should be lessened by the unfavorable effects of reduction of embryo size.
      Many forms of Citrus, and their progenitors, may have been reproduced almost entirely through nucellar embryos from remote antiquity.   Differentiation of varieties, in such a group as the sweet oranges, may have occurred mainly by somatic change without frequent sexual reproduction.   Occasionally an exceptional recombination of genes from heterozygous parents might have given rise to a different and horticulturally valuable new type.   Such a type might be so unlike the parents that its origin would remain uncertain.
      Tanaka (1927a) concluded that such abrupt origin of radically different types has occurred, but he considered that mutation, rather than recombination, probably was the main cause.   He suggested that the satsuma, the Mediterranean mandarin, and the grapefruit may have originated in this way, and he indicated forms which seem closely related to these.   For the grapefruit, the obvious relative is the pummelo, which lacks nucellar embryony and seems to be much less heterozygous than most citrus forms.   Origin by interspecific hybridization has also been suggested for the grapefruit; if this occurred, segregation and recombination evidently established a definite new type.   Traub and Robinson (1937) suggested that the grapefruit seems intermediate between sour pummelos and "salad" pummelos.
      The low variability among the true lemons, and among the sweet oranges, may represent mainly somatic variation.   Mandarins, however, have a broader genetic base, and suggest more differentiation by sexual reproduction.
      Comparatively wide crossing in Citrus has no doubt been more common under domestication than in earlier times, because forms have been brought together under cultivation.   Although Tanaka (1927a) considered that definite hybrids among the varieties of the Pacific region were rather few, he (1961) described several probable cases, including the Benikoji, the Funadoko, and the Iyo.   Many Indian citrus varieties described by Bonavia (1888-90) had characters suggestive of hybridity between lemon and citron.   Luss (1935) concluded that the Ponderosa and Melarosa lemons and the Asahikan "pomelo" are interspecific hybrids.   Other probable natural hybrids include the Rangpur lime, Meyer lemon, Japanese citron, and Temple orange (tangor?).

Inherited Characters in Citrus
      Citrus normally has nine pairs of chromosomes, which carry perhaps thousands of genes.   For a number of reasons, little has been proven as to the effects of the individual genes.   Each tree occupies considerable space and usually requires several years from seed to fruiting.   The frequent presence of nucellar progeny, which may not be identifiable until fruiting, complicates genetic studies.   Zygotic seedlings from any one cross are highly variable, and those from close inbreeding are often weak or sterile, making gene analysis difficult.
      Occasionally a conspicuous character difference and a fairly simple progeny ratio suggest segregation for one pair of genes.   Thus, when the dwarf and short-petioled Bouquet sour orange was crossed with the Imperial grapefruit, about half the hybrids had very short petioles (fig. 5-6); nearly all of these trees also resembled the Bouquet in other foliage aspects (Frost, 1943).   But when the Maltese Oval orange was crossed with the Bouquet, the hybrids were not sharply separable by these characters.   Perhaps one chromosome of the Bouquet carries a gene with major effect on leaf form, while the allel on the homologous chromosome has less effect.   Such an effect was obviously modified when Maltese Oval was the other parent.
      Toxopeus (1936) reported segregations for dwarf plants from the cross of a pummelo (Citrus grandis) with a citron (C. medica L.).   When the pummelo was used as seed parent, only dwarf hybrids were obtained; the reciprocal cross produced 50 per cent dwarfs.   Such ratios can occur when one parent is heterozygous for a dwarfing gene and there is also elimination of the allelic normal gene from the gametes of one sex but not the other.   At the University of California Citrus Research Center, several progenies involving a particular pummelo, obtained in 1945 and 1946, have also shown segregations for extreme dwarfs.   Hybrids of the Karn Lau Yau pummelo (CRC 2341) as female, crossed by Clementine mandarin, were exclusively dwarfs, which remained only about 20 to 30 cm high at three years of age.   Crosses of this pummelo with Temple, Kara, and Valencia each produced some but not all dwarfs.   Normal siblings were many times larger than the dwarfs.   When plants which died in their first year, without making appreciable growth, were included as probable dwarfs, the proportion of dwarfs ranged from 42 to 62 per cent in these families.   In contrast, a progeny of 176 plants of the same pummelo crossed by the Pearl tangelo showed no dwarfs.   A different pummelo, Moanalua (CRC 448) when crossed by an orange, yielded a lower proportion of dwarfs.   A single major gene could be involved.   In the Riverside progenies, after several years, one dwarf and a budling, probably from a different dwarf, began to grow vigorously, indicating a radical change in physiology.   Fig. 5-7 shows two dwarfs and one normal plant.
      There is evidence that within the genus Citrus, nucellar embryony is inherited in a rather simple fashion.   Parlevliet and Cameron (1959) found that hybrids from sexual, monoembryonic parents were themselves always essentially monoembryonic.   Crosses of monoembryonic by polyembryonic parents of several Citrus varieties yielded hybrids of both types in ratios which often approached 1:1; certain parents conditioned higher proportions of polyembryonic offspring.   Monoembryonic offspring were occasionally obtained from polyembryonic parents.   Nucellar embryony may be principally controlled by one dominant gene; monoembryonic varieties would then be homozygous for the recessive allele.
      Recent crosses of the Clementine mandarin by Poncirus trifoliata (monoembryonic X polyembryonic) have given different results in F1.   Thirty-one out of thirty-two hybrids tested appear to be monoembryonic.
      Evidence that certain essentially monoembryonic varieties do not produce nucellar embryos at all was obtained by Ozsan and Cameron (1963).   (See Chapter 4.)
      Levels of acidity in citrus fruits are also inherited.   Both acid and essentially acidless forms are known in many varieties.   Soost and Cameron (1961a) obtained hybrids of relatively low acidity from crosses of a low-acid pummelo by medium-acid pollen parents of several varieties.   Other pummelos of rather high acidity, when crossed with the same pollen parents, produced progeny whose acid levels were mostly above those of either parent.   The gene basis is not known; it was noted that the differences in acidity were not closely correlated with differences in levels of soluble solids.
      The trifoliate leaf character of Poncirus is known to show dominance in crosses.   In the F1, it is nearly always fully dominant over the monofoliate leaf of Citrus; a few hybrids occur which show mixtures of bifoliate and monofoliate leaves.   Patterns of segregation in advanced generations do not indicate single gene inheritance.   Some unifoliate seedlings from open pollination of F1 plants are produced, but the proportions are low.   Yarnell (1942) grew open-pollinated progenies from citranges and citrumelos.   A total of fifty seedlings from six zygotic citrumelos included nine unifoliate and seven variably unifoliate plants.   Among 3,589 seedlings from twelve highly nucellar citranges, segregants of any kind were rare as would be expected.   Webber (1932) had earlier reported leaf segregation in open-pollinated progenies of only one (the Sanford) of six citrange varieties studied.
      The authors have studied leaf segregation in open-pollinated progenies from a series of highly zygotic F1 hybrids of the parentages Clementine X trifoliate orange, and Sukega grapefruit X trifoliate orange.   Second generation seed was obtained where location and flowering times almost insured that only selfing or crossing among the hybrids had occurred.   From a total of 1,716 seedlings from thirty-three hybrids with Clementine, 7.17 per cent were monofoliate.   From 326 seedlings from seven hybrids with Sukega, 5.8 per cent were monofoliate.   A small added percentage of each group were fluctuatingly bi- or trifoliate.   Classification as monofoliate was based upon seedlings which had reached at least the eight-leaf stage.
      J. R. Furr (unpublished) has grown seedlings from monoembryonic Temple tangor and Clementine mandarin pollinated by the Troyer citrange; these represent backcrosses to monofoliate leaf types.   Only 20 to 30 per cent of the seedlings were monofoliate.   These data appear to fit the pattern of duplicate factors by which either of two dominant genes conditions trifoliate leaves.
      Toxopeus (1962) reported that among 50 two-year-old seedlings from the cross of C. grandis (L.) Osbeck X C. hystrix DC. (both monofoliate) about one third were trifoliate-leaved.   He suggested that two complementary, dominant genes may be involved, with each species being heterozygous for one of them.   This gene system does not account for the breeding behavior of Poncirus, just described.   Toxopeus also found that in the cross of C. grandis (three varieties) X Meyer lemon about 50 per cent of 598 seedlings showed in young leaves the purple color characteristic of lemons; he postulated that the Meyer may be heterozygous for a dominant gene for this color.
      The trifoliate orange is notably resistant to the citrus nematode, Tylenchulus semipenetrans Cobb (DuCharme, 1948; Baines, Bitters, and Clarke, 1960), and it transmits this resistance to its F, hybrids with Citrus, as measured by root infestations of young seedlings in the greenhouse.   Cameron, Baines, and Clarke (1954) found that 95 per cent of 484 hybrid seedlings, involving five Citrus species, showed low infestation, as did 100 per cent of 846 Poncirus seedlings.   From 63 to 100 per cent of the nucellar seedlings of the Citrus parents became heavily infested.   Older trees of Poncirus hybrids in the field do not show such consistent resistance.
      In Florida, search for resistance to the burrowing nematode, Radopholus similis (Cobb) Throne has lately been the subject of intensive study.   Great differences in susceptibility have been reported among Citrus varieties, selections, and relatives (e.g., Ford and Feder, 1961).   Among more than 1,200 forms tested by 1962, nearly all have been found susceptible, although a few specific source trees within several taxonomic groups are tolerant or resistant.   Two rough lemon selections, six sweet oranges, and the Carrizo citrange are classed as tolerant.   Ridge pineapple sweet orange and Milam, a lemon hybrid, are resistant.   Among twelve nonsusceptible selections, seven are sweet orange types, but these fall among several subgroups.   Such isolated instances of resistance suggest unexpectedly specific genetic differences, especially since some of the selections are in most characters highly similar to susceptible members of their groups.
      A striking range of segregation for tree and fruit characters was described by Reece and Childs (1962) among about seventy seedlings from seed collected from pulp of the nearly seedless Persian lime.   It was concluded that the Persian lime can segregate for many interspecific characteristics, presumably after selfing.   The Persian lime studied by Bacchi (1940) was found to be triploid, as was the Bearss lime (Krug and Bacchi, 1943); Nakamura (1934), however, reported diploidy in the Persian lime.   A wide range of characters is not unlikely in the progeny of a triploid, especially if it is a hybrid.   However, some of the seedlings in question probably arose by outcrossing or other contamination.   Two seedlings were found to be essentially identical to the Persian lime (Childs and Long, 1960) and thus seemingly of nucellar origin.   If they are not nucellar, they would represent unusually rare products of sexual reproduction.
      Data on inheritance of salt tolerance, Phytophthora tolerance, and cold hardiness are discussed [below].   Segregation for male sterility is described [below].

      In a broad sense, mutation may be defined as change in genetic constitution which arises suddenly and is not due to segregation and recombination of genes in sexual reproduction.   Mutation may occur either in germ cells or somatic cells.   More narrowly, gene mutation signifies change of a gene to a somewhat different gene.   As noted [above], loss or rearrangement of chromosomes or their parts or inherited changes in cytoplasmic particles may simulate gene mutation in their phenotypic effects.   Heritable bud variations and chimeral conditions usually are the results of mutation in one form or another.
      Even in the best-analyzed organisms, it is often impossible to distinguish between gene mutation and the loss of a minute part of a chromosome which included that gene.   However, there is evidence in lower organisms for reversion, or back mutation, of genes.   In such cases, the gene apparently could not have been lost.   In other cases, the action of a gene can be shown to have been suppressed by another gene.
      Among sexual seedlings of citrus, it is rarely possible to distinguish new gene mutations from variations due to normal recombinations of genes from the parents.   Among variant nucellar seedlings, or variant branches or fruits, gene mutation is likewise seldom separable from grosser chromosome changes.   The possibility of chimeral conditions introduces uncertainty, since tissue of one genetic type may remain within that of another for an indefinite time, and then emerge to produce a new effect.
      Unstable Genetic Factors in Citrus.—In some sectorial fruit-rind chimeras, the visible changes appear to be caused by unequal nuclear division rather than by gene mutation.   A fruit may have two adjoining variant sectors of rind, often about equal in width, which are changed from the normal in opposite directions (for an early example, see Toxopeus, 1933).   Thus, one sector may be abnormally thick and a contiguous sector abnormally thin.   Since one member of such a pair of variations may often be inconspicuous, or entirely lost (perhaps due to growth competition), rind chimeras showing single sectors may often also be the result of unequal nuclear division.   Various bud-variation types often differ from the parent type in several characters, which leaves it doubtful whether the original change was limited to one gene.   In maize, Jones (1937) concluded that certain endosperm changes were produced very much oftener by somatic segregation than by gene mutation.
      When mutations occur in citrus, their expression in somatic tissues should be favored by the apparently extensive heterozygosis of most forms.   Such heterozygosis, in turn, implies that mutation has occurred frequently in the past.   Certain unstable types, such as are discussed below, could result from gene mutation.
      In an early study at Riverside, California, six of the nucellar offspring of one Valencia orange tree bore fruits that were partly without juice (Frost, 1943).   The fruits varied from abnormal firmness and toughness of juicy pulp to complete absence of juice (fig. 5-8).   The fruits on each tree differed greatly among themselves, but in addition the affected trees clearly differed in average amount of fruit dryness.   This behavior can still be observed in trees more than thirty years of age in the orchard, which are the budded progeny of the original seedlings.
      A pollenless navel orange of unknown origin, known as Ruvel, produced in five separate nucellar seedlings from two small branches a pollen-bearing and mainly non-navel form called Rufert (Frost, 1943).   Trees budded from different seedlings have differed in average grade of the new characters; some are intermediate, and some vary within the individual tree much like the Valencia progeny described above.   Their characteristics, although variable, have been maintained by budding.   The Trovita orange (Frost, 1935) is somewhat similarly variable.   Three seedlings from a single fruit gave rise to the Trovita type; trees propagated from one of these has produced frequent navel-marked fruits and fewer seeds than the others.
      The red flesh color and rind color of the Ruby orange may be genetically unstable.   Variant rind sectors with increased or decreased redness occur frequently, and segments and half-segments showing a discontinuous increase or decrease in pulp redness have been observed (fig. 5-9).   In addition, nucellar progeny trees from seeds of a single parent tree at Riverside have included more than one genetic color type.   Trees budded from different seedlings have differed greatly and persistently in average pulp color.   In a first-budded generation trial between 1929 and 1935, nucellar selection 41 had a much higher color rating than selection 35 (table 5-2).   In the second-budded generation, the selections maintained this relationship and both appeared to differ from budded trees of their seed parent.   The red color is very susceptible to environmental modification, but the relative intensities in these selections have been consistent for many years.
      Wood pocket disease or ligno-cortosis of semidense Lisbon lemons (Calavan, 1957a, 1957b) appears to result from an unstable genetic factor and may involve chimeral interrelations.   Its symptoms include leaf variegation, fruit rind sectoring, and wood cankers.   A similar or identical disease, sometimes called lime blotch, is known in Tahiti limes (Gates and Soule, 1950).   Wood pocket disease recurs in lemon trees grown from buds or grafts from diseased sources, and it is transmitted through a considerable proportion of both zygotic and nucellar seedlings.   In contrast to citrus virus diseases, however, it is not transmitted to healthy trees by tissue grafts, and its known occurrence is limited to certain lemons and limes.   Degree of symptom expression is variable, but there is a strong tendency for the disease to increase in severity in successive budded or seedling generations.
      Bud Variation.—Mutations resulting in bud variation are very prominent in citrus.   The mechanisms which can be involved were noted [above].   In a plant which is a synthetic chimera, bud variation can occur without having been preceded by mutation.
      A somatic mutation presumably may originate in any cell.   If the new cell type can reproduce itself, a sector descended from it may develop, and the shoot or fruit becomes a chimera.   Sooner or later, unmixed shoots of the new type may be produced.
      In studies of bud variation, one must distinguish between environmental variations and occasional heritable changes.   Only when a bud variation is markedly different in tree or fruit is it likely to be detected among the environmental fluctuations which are commonly present.   A new form differing only slightly in some quantitative character such as yield or fruit size will often be overlooked, and would require extensive trial to demonstrate its genetic difference from the parent type.
      A. D. Shamel and associates, of the U.S. Department of Agriculture, carried out a long series of studies on bud variation in citrus, which extended over the period from about 1909 to 1936 (Shamel, 1943).   Shamel sought out and described several score of variations occurring among important commercial varieties, principally in growers' orchards in California and Arizona.   The variations occurred both as limb sports and whole trees; the permanence of many of these changes was demonstrated by bud propagation and performance records of the normal and variant progeny types.   In addition, many single-fruit chimeras were observed that could not be propagated.
      Shamel classified the variations that he studied as morphological or physiological, and as physical or chemical.   Variations were found which affected tree or leaf size, tree habit, chlorophyll production, fruit size or shape, rind thickness, and acidity.   These classifications, of course, represent the end results of various genetic and developmental interactions.
      In commercial orchards of Washington navel and Valencia oranges and Eureka lemons surveyed by Shamel beginning about 1910, an average of some 25 per cent of the trees were judged to be entire-tree variations.   The percentages ranged from less than 10 per cent to more than 75 per cent, depending upon the orchard; limb variants of the same types were also often found.   It seemed clear that the whole-tree variants had arisen from budwood taken unintentionally from mutant twigs or branches on budparent trees, and that the same mutations were sometimes recurring on normal progeny trees.   Single-fruit variations also occurred on many trees.
      From surveys of the Washington navel variety, Shamel and his associates designated at least twenty-four variant types.   These included unproductive variants, some of which were classed as "Australian" types, variants with rolled or slender leaves, and many variants bearing different sorts of abnormal fruits.   Among other citrus varieties, some fifteen variant types were described in the Valencia variety, about eight major types in the Eureka lemon, and six in the Lisbon lemon.   Several variants were also recognized in the Marsh grapefruit and in other varieties less extensively studied.   The classes of changes observed were rather similar among all varieties.   Table 5-3 is a summary of data from Shamel (1943) showing the characteristics and progeny performance of some of the unfavorable bud variations and their normal counterparts.
      Several bud variants showing potentially useful characters were described by Shamel.   One of these, a limb variant of Washington navel designated Everbearing, flowered over a considerable period and bore dense clusters of fruit.   Thorns on this variety sometimes bore flowers and fruit.   Another productive variant of the Washington, called Vivial, bore fruits without navel structures.   The Robertson navel (Shamel and Pomeroy, 1935) is a variant usually ripening slightly earlier than normal Washington from which it arose.
      Since most bud mutations are undesirable, Shamel rightly emphasized the importance of careful selection of budwood from source trees typical of the variety and preferably with proven performance records.   Much of his work consisted in helping to establish dependable sources of normal or possibly improved selections of existing citrus varieties.
      Numerous bud-mutation types, in addition to those discussed above, are known or believed to have arisen from the Washington navel orange, which itself is considered to be a sport from the Selecta orange.   Variants that have received some recognition include the Thomson, Carter, and recently, in California, the Gillette and Newhall.   Most of these variants appear to be slightly earlier or sweeter than the Washington.   Another mutant, the Marrs, has attained prominence in Texas (Waibel, 1953; Olson, 1963), where it was discovered as a limb sport in 1927.   In Texas, the Marrs shows several differences from the Washington, including greater productivity, earlier marketability, smaller tree size, fertile pollen, and a lack of bitterness in the juice.   The fruit seldom has a navel.
      The Salustiana orange, reported to have arisen in Spain as a bud sport of the Comun, ripens early in North Africa (Chapot and Huet, 1963).   It is almost seedless, juicy, and highly productive.
      A recent apparent bud mutation, resulting in an inferior type, is a high-acid form of navel orange recognized in California in 1960 (Soost et al., 1961).   The strain was through an accident widely propagated, so that many young plantings of supposedly normal Washington navels contained varying numbers of trees with the variant characters.   Growth habit of the trees was unusually upright, vigorous, and thorny.   Fruits averaged smaller than comparable normal Washington navels, with a more pebbled rind texture and a smaller navel opening.   The flesh was softer and more juicy than in normal fruits, and acidity averaged two- to four-tenths percentage points higher during the main part of the harvest season.   Because of high acidity, the fruit has often failed to meet legal maturity standards early in the season.   This variant is believed to have originated from a budstick from a normal Washington navel tree (Frost selection) of nucellar origin.   The budded variants were not immediately detected, since vigorous growth is typical of propagations from nucellar selections.   A widespread top-working program was carried out to eliminate the mutant.
      Bud Variation in the Satsuma.—Satsuma is a citrus group in which bud variation has been very important.   Extensive studies in Japan by Tanaka (1932) over a period of many years showed that numerous varieties have arisen by limb variation.   At least three distinct varieties, the Owari, Ikeda, and Zairai, seem to have been established in various areas of Japan by the time the studies began in about 1909.   A fourth type, called Wase, was also being grown, and this was traced back to a group of four source trees, from which it became known as Kuwano Wase.
      In some orchards of Kuwano Wase, further bud variations were found; these took the form of reversions to the Owari fruit type.   At about the same time, it was shown that a bud variation branch on a known Owari tree had fruit characters essentially identical to those of Kuwano Wase.   This was the Higuchi Wase.   Such evidence led to the conclusion that Kuwano Wase is a bud sport from Owari.   Many other bud variations classed as Wase types were identified on Owari, Ikeda, and Zairai trees.   These varied among themselves, although early maturity is apparently a common feature.   At least part of their differences could be ascribed to differences in the parent varieties.
      New variations have continued to be reported in Japan.   Thus, Yamaguchi and Yakushiji (1960) described two early satsuma sports, the Matsuyama and the Tichima, both from Owari.   Iwasaki et al. (1962) described a strain test of seven early satsuma types, carried out from 1939 to 1961.   Five of the selections previously described by Tanaka were included, and two newer ones, Iseki and Mikkabi-10.   The Miyagawa was judged best; vegetative reversion was observed in a Suzuki selection during the study.
      The mechanisms involved in satsuma bud variation are uncertain.   It is not easy to ascribe all the correlated changes seen in a mutant type to single-gene mutation, although this is not impossible.   The tendency of the Kuwano Wase to frequent reversion is suggestive of a shift in germ-layer composition of a chimera, although Tanaka did not favor this explanation.   All sports classed as Wase types may not be due to the same genetic change.   All seem to have been selected for earliness, but Iwasaki et al. (1962) report that fruit shape, coloring time, susceptibility to disease, and other characters are not the same among the variants.
      Nakamura (1929), in cytological studies of some of the Wase type variations, found no differences in the chromosome number (2n = 18) and reported that meiotic behavior seemed normal.   Small changes in chromosome structure, however, are difficult to detect in citrus.
      Frost, Cameron, and Soost (1957) reported a long-term study of nucellar-seedling selections of satsuma, which were judged to show effects both of nucellar embryony and of genetic variation (table 5-4).   Three selections were identified from one seed-parent tree.   Selections A and B were obtained repeatedly from seeds of fruits from different branches on the parent tree, and a third selection C, arose from a seed from the same branch which produced a seedling of A.   Budded trees of A and B both showed larger tree and leaf size, greater yields, and more alternate bearing than trees from the seed-parent budline; these differences seemed to be effects of nucellar embryony.   Fruit coloring was earlier in these two nucellar selections, and soluble solids were higher throughout the season.   Trees of A were more open and had flatter fruit, which usually colored earlier than those of B.   Selection C was more erect, with later-maturing, thicker-rinded fruit.
      It is almost certain that all of these selections were of nucellar origin; the seeds came from fruits from controlled pollination by parents very unlike satsuma.   Indexing tests indicated that the tristeza virus was not involved.   Selections A and B probably existed as shoots on the seed-parent tree as a result of bud variation.   Selection C may have occurred in the same manner, or it may have arisen in nucellar tissue; C, at least, is a genetically distinct type (a strain in the sense of Shamel, Scott, and Pomeroy, 1918).   The seed parent appears to be Owari, which gave rise to many of the variants in Japan.   Horticulturally, selection A has been superior to B and C, and to the parent.

      Composition, Behavior, and Terminology.—
A plant chimera is a combination of tissues of two or more genetic constitutions in the same individual plant or part of a plant (fig. 5-10).   The term chimera is not applied to ordinary budded and grafted plants, but is restricted to forms in which the combined genetic types (components) grow together, side by side, in the same part of the plant.   These components, like the stock and scion of a budded tree, maintain separate genetic constitutions in their respective cell lineages, but their close association results in mutual physiological influence and often compels compromise between conflicting growth tendencies.   As a rule, the components of a chimera no doubt differ in chromosome or gene constitution; occasionally, they may differ in plastid constitution only, as in some green-and-white variegated forms.
      The term graft hybrid correctly refers to chimeras resulting from graftage.   There is little evidence in higher plants for the existence of hybrids produced by actual fusion of somatic cells of different genetic types.
      Chimeras are classified according to the relative positions of the components.   If one component forms an outer covering surrounding an inner core of another component, the combination is a periclinal chimera (Jones, 1934).   If one component does not surround the other, but both extend to the surface or the center of an organ, the chimera is sectorial.   If the arrangement is externally sectorial, but one component does not extend to the center of the shoot, the chimera is often called mericlinal (Jorgensen and Crane, 1927).   If the two genetic types are irregularly mixed throughout the plant, as in some forms of variegation, they form a mosaic or mixed chimera.
      Chimeras may be classified also on the basis of their origin (Frost, 1926a).   If a chimera is produced by growth from tissues of two varieties or species, at a bud or graft union, it is synthetic; if it is produced by genetic change in a cell or cells originally like the rest of the plant, it is autogenous.
      All these kinds of chimeras seem to occur in citrus trees; in view of their importance in relation to bud variation and variety improvement, their characteristics are discussed in detail below.
      Developmental Relationships in Chimeras.—Chimeras can be analyzed on the basis of the arrangement of the component genetic types in leaf, stem, and flower tissues, together with the ways in which the various parts are developed from the meristems, or growing points, of the plant.   One growing-point cell layer, or primary histogen (Dermen, 1945) may produce one or more layers of the mature tissue; in the fruit, tissue of superficial origin may be deeply buried.
      In older studies, evidence for chimeras was usually based on the morphological and histological appearance of the tissues.   Some of the best-known cases in citrus can be recognized in this way.   Differences in chromosome number, when they occur, provide a much more accurate means of identifying the cell lineages.   If the epidermis, for example, carries a tetraploid complement of chromosomes, while the hypodermal tissue carries the diploid number, tissue derivation can be determined in the developing parts by chromosome counts, and often by cell size.   During the past twenty-five years, this method has been used in several plant studies.   In Datura, extensive investigation has shown that three primary histogenic layers are regularly present in the apical meristems, and that these layers give rise to certain tissues of the stem, leaves, and fruits (Satina, Blakeslee, and Avery, 1940; Satina, 1945).
      Dermen (1947, 1953a, 1953b) employed the same method to trace the origin of tissues in cranberry, apple, and peach, where three histogenic layers are also the usual basis for development.   In many of these fruits, and in Datura, the polyploid condition was induced by colchicine.
      The number of histogenic layers and the regularity of their behavior are incompletely known in most families of plants.   A review by Gifford (1954) showed that in a majority of the dicotyledons studied there were two primary outer layers and presumably one core layer, while in monocotyledons only one outer layer was present in many cases.   Citrus appears to have three histogenic layers.
      In dicotyledonous plants, the epidermis is regularly developed from the outermost growing-point layer (the dermatogen, or Layer I): Layer II produces a marginal zone of leaf tissue, or even all the leaf parenchyma inside the epidermis, as well as the gametes and at least part of the associated tapetal and nucellar cells.   In certain green-and-white chimeras a green-producing Layer II usually produces a larger part of the leaf tissue than does a white-producing Layer II (Rischkow, 1936).   In some exceptional cases Layer I, instead of Layer II, gives rise to the gametes (Jones, 1934, p. 42).   Differences in growth-rate tendencies between the components of a chimera may affect the regularity of tissue differentiation and the stability of the chimeral pattern.
      Sectorial chimeras are generally unstable; in the formation of lateral buds, one component is often eliminated or a periclinal arrangement may be formed.   The conditions producing mosaic chimeras include: (1) irregular assortment of green and colorless plastids in cell division; (2) frequent genetic change within cell nuclei; and (3) unbalanced growth relations in a periclinal chimera, such that one component frequently takes the place of another.   In citrus forms where there seems to be frequent genetic variation within cells, the variable phenotype may depend on several grades of chromosome constitution, apart from changes in the relationships of tissue components.
      Synthetic Chimeras.—In early experiments, Winkler (1908) grafted tomato onto its relative, the nightshade, and nightshade on tomato, after which he cut off the grafted plants at the union.   Sprouts developed from adventitious buds, and occasionally one arose at the line of the union so that one side of the shoot was tomato and the other nightshade, constituting a sectorial chimera (fig. 5-11).   Intermediate forms were also obtained, which when analyzed by chromosome counts and seedling characters proved to be periclinal chimeras (Winkler, 1910; Baur, 1910).   These chimeras had one or more outer layers of tomato tissue and a core of nightshade, or vice versa (fig. 5-12).   Where synthetic chimeras are so formed, by the combination of different species, identification of the chimeral condition is fairly easy.   Forms capable of somatic separation into two types having leaves or fruit clearly of two species have probably been produced by graft synthesis.
      The citrus bizzarria.—Several freakish citrus forms, known in Italy as bizzarria, are evidently synthetic chimeras.   The bizzarria of Florence was described by Nati (1674, 1929); it had branches, leaves, and fruits partly like citron, partly like sour orange, and mainly intermediate or mixed (fig. 5-10).   Nati concluded, mostly from testimony of the gardener under whose care the bizzarria had originated, that it probably arose at the union where a citron bud had been inserted on a sour orange seedling—the bud failing to grow at first, but a mixed shoot coming out later.   Strasburger (1907) concluded that this bizzarria did not result from actual fusion of somatic cells, since it had the same chromosome number as the parent types, not twice as many.   Tanaka (1927b) studied Nati's bizzarria, and concluded that it is primarily a periclinal chimera, with a core of citron and outer layers of sour orange, the sectorial structures resulting from irregularities in histogenic-layer behavior; most of the rind is superficially sour orange, with yellowish streaks where the citron core has emerged, and occasional wider sectors of thick citron rind.   The instability of this chimera may be due to the more vigorous growth of the inner component.
      Strasburger (1907) summarized accounts of other citrus bizzarria, including one of citron and sweet orange that was in existence before 1644, when the bizzarria of Nati is said to have originated.   Savastano and Parrozzani (1611) described as "natural hybrids" three bizzarria which presumably were synthetic periclinal chimeras.   Brown (1918) described a tree that appeared to be a lemon-over-orange chimera, and Uphof (1935b) described a tree that bore intermingled fruits of sweet and sour orange.   Intermediate fruits were not observed.
      In Japan, two citrus forms considered to be synthetic periclinal chimeras have been in existence for some fifty years (Samura and Nakahara, 1928; Takahashi, 1962).   The Kobayashi Mikan is recorded as having arisen at the junction where a satsuma scion was grafted onto Natsudaidai.   This scion was accidently [sic] broken off at its base, after which two adventitious buds emerged, one of which grew into the original chimeral tree.   This chimera produces fruit with rind like Natsudaidai, but with flesh like satsuma.   It is seedy, and seedlings from one fruit were reported to be Natsudaidai (Dr. Yuishiro Tanaka, unpublished).   In this chimera, histogenic Layer II appears to be genetically Natsudaidai, while Layer I should be satsuma.
      Kinkoji Unshu arose in a similar manner, in about 1912, after satsuma scions had been top-worked onto Kinkoji (C. obovoidea Takahashi).   A branch was later observed which produced fruits having outer parts like Kinkoji and inner parts like satsuma, with the fruits bearing some seeds.   Both of these chimeras have been maintained in Japan.
      Casella (1935) described and showed photographs of a tree of sweet lime budded on sour orange, which produced a vigorous chimeral branch at the bud union.   This branch produced some fruits similar to lime, some similar to orange, and some chimeral fruits with rind sectors of both types.   Casella reported that seeds and juice vesicles in these fruits and fruit sectors corresponded in character to the respective rind types.   He stated that six-year-old seedlings from seeds from orange-like fruit sectors had characters of sour orange, but that those from lime-like sectors also had sour orange characters.   Casella concluded that the variant branch was a synthetic chimera.
      Autogenous Chimeras.—Baur (1909) discovered that certain variegated forms of Pelargonium (garden "geranium") are white-over-green or green-over-white periclinal chimeras.   In later studies, such as those of Bateson (1916) and Chittenden (1927), other herbaceous plants were found to be periclinals.   Since the varieties concerned are propagated by cuttings, not by graftage, these chimeras no doubt originated by some change in cells of the parental clones.
      The components of such chimeras can often be identified by one of the following methods: (1) in green-and-white combinations, the components are visible by inspection in parts normally green; (2) in citrus, a component present as the subepidermal layer which produces the nucellus should appear in all histogenic layers of the derived nucellar seedlings; and (3) components with differing chromosome numbers can be identified during cell division, in certain tissues.   A component present as an inner layer, from which adventitious roots arise, can sometimes be detected by obtaining whole plants from such roots.   In some bud variation forms, root cuttings of this kind give the parental type that produced the variation; thus, certain thornless varieties of blackberry produce thorny plants from root cuttings.
      Many of the mutant types that arise in growing-point cells of Citrus must be inferior to the parental type in vigor and are soon crowded out.   Occasionally, however, a variant type is noticeable because it has come to occupy an extensive region in one or more cell layers.   If a bud-variation shoot contains only an outer layer of a new type, it may lose this type through emergence of the inner type, or it may lose the inner type by the reverse process.   Since chimeras are subject to such changes, a bud variation is not necessarily transmitted to all trees budded from it.
      Variant sectors of rind occur commonly but very irregularly on fruits of many varieties of Citrus.   Certain characters may be conspicuously changed, such as thickness, smoothness, and color of rind, size of oil glands, and color of pulp; often, several rind characters are changed together.   When such changes occur only occasionally within a tree or clone, they presumably represent separate new occurrences of mutant types in cells of the young fruits.
      Forms showing leaf variation give valuable evidence on the behavior of chimeras, because of the ease of recognition of the cell types concerned.   In normally green seed plants, the epidermis is without green color, except that usually there are chloroplasts in the stomatal guard cells.   If a mutant subepidermal layer of the growing point also produces non-green tissue, but the layer within it produces green tissue, the leaves commonly have a white marginal zone and also have at least one colorless layer of cells under the epidermis in the green central portion of the leaf (fig. 5-13, E-H).   In contrast, if the subepidermal histogen is green-producing, but the third layer is not, the marginal zone of the leaf has normal green color, while the central portion is pale because it has a white core under one or more layers of green cells (fig. 5-13, A-D).   Irregularities may produce a shoot of the reverse form or one which is entirely green or entirely white.   For examples in Poinsettia and in the privet, see Dermen (1950).   The patterns of variegated leaves in different species indicate that tissue zones are formed from the growing point in a much more regular way in some kinds of plants than in others, citrus chimeras being especially irregular.
      White-over-green chimeras.—Shamel et al. (1920a; 1920b) described variegated bud-variation strains of Eureka and Lisbon lemon.   Another form somewhat similar in type of leaf variegation, designated as Variegated Pink lemon, was discovered as a whole tree (Shamel, 1932).   It has leaves with an irregular white marginal zone, the white regions often being much dwarfed (fig. 5-14).   The upper sides of the leaves show as many as four different grades of green, and the lower sides two or three, the upper and lower patterns being mainly independent.   The immature fruits have narrow longitudinal stripes of thin whitish and normal green rind, the green becoming more abundant toward the apical end.   Mature fruits are yellow, slightly ridged, and have pink flesh.   The bark of the young shoots is usually pale, with longitudinal stripes of deeper green commonly arising at the leaf axils.   In each growth cycle, the amount of green in bark and leaves usually increases from base to apex of the shoot.   Occasionally a shoot is completely white.   The characteristics of this chimeral variety are still regularly expressed in a budded tree now more than thirty years of age at Riverside, California.   The variety may not be a simple white-over-green form; mutations in plastids, followed by plastid segregation at somatic cell divisions, might account for its chlorophyll patterns.
      In some light-margined citrus forms, the non-green areas may be creamy white; in others they are light yellow.   In the Imperial orange, grown at Riverside since 1914, the leaves are variegated similarly to those of the Variegated Pink lemon.   Fruits are orange-yellow, with narrow, thicker stripes of reddish-orange.   These characters appear irregularly throughout the tree, year after year.   Small populations of seedlings of this variety, from open pollination, have been entirely white-leaved and soon die.   Seedlings from a variegated, white-over-green Eureka lemon described by Shamel (1932) also produce white leaves.   This behavior supports the assumption that germ Layer II is producing both the nucellus and the white leaf tissues in the seed-parent tree.
      In such chimeral forms, both Layer II and Layer III seem to be participating in the formation of the leaves.   Layer III apparently has normal chlorophyll-producing capacity; the formation of white leaf tissue from Layer II seems to undergo rhythmic changes with the growth cycles.   Completely white shoots suggest that Layer III is occasionally lost from a growing point, but the white-producing Layer II is very seldom lost.
      Green-over white chimeras.—A budded tree of a variegated Lisbon lemon (CRC 2519) grown at Riverside before 1930, had leaves whose upper sides showed central areas of light green, surrounded by marginal zones of normal, darker green.   A few leaves seemed entirely normal.   The lighter areas were variable in size and shape, and sometimes showed at least two tints lighter than normal, yet the variation was less than in the white-over-green lemons described above.   The lower sides of the leaves only occasionally showed light areas.   No completely white areas were found, and the bark of young shoots seldom had light streaks.   This lemon was evidently a green-over-white chimera, but the white core component did not emerge to produce subepidermal tissue.
      An instance of complex instability is evidenced by a clone of variegated sour orange (CRC 622) budded at Riverside in 1914, from a seedling which showed leaf variegation when about 18 inches high.   The original budded tree is still living.   Most of the foliage has been normally green but some leaves are variegated (fig. 5-15) similarly to those of Lisbon lemon just described.   Occasional leaf areas appear white, but these are often bordered by a green zone along the leaf margin.   Some twigs with much leaf variegation have bark with whitish stripes.   Mature fruits (fig. 5-16) usually show a normal, deep orange rind color, but they may have a narrow stripe or wider sectors of thin, smooth, lemon-yellow rind; rarely, a fruit is mainly yellow and of small size.   This sour orange chimera shows some characteristics of a green-over-white type, as if a white layer were occasionally emerging to form subepidermal tissue; however, seedlings, including both nucellar and zygotic ones, continue to show the chimeral condition.   In 1963, thirty-one young seedlings were grown from open-pollinated seed from variegated fruit by D. A. Cole, Jr.   Seventeen probably nucellar and three seemingly zygotic plants all showed some leaf variegation of the same type as in the seed parent.   Eleven entirely white-leaved seedlings died very young.   This chimera seems to be due to an unstable genetic factor; the white seedlings may originate from already mutated cells, while in predominantly green seedlings mutations occur during growth.   Plastid segregation could be involved.
      Other autogenous citrus chimeras.—Other bud-variation strains give indication of chimeral constitution, without leaf variegation.   The Golden Buckeye navel orange (Shamel, Pomeroy, and Caryl, 1925) is one of these; it appears to have an inner layer similar to Washington navel, surrounded by a subepidermal layer producing a smoother, more yellowish rind.   The rind shows narrow longitudinal stripes and ridges of rougher, reddish-orange tissue which seems identical to that of the Washington.   Fruit shape tends to be more elongated than with the Washington; the pulp is firmer and sweeter, with a different aroma, and a navel opening is less commonly present.   An occasional branch produces fruits which seem to be entirely Washington.
      The difference in the pulp of the Golden Buckeye may be explained by the fact that in citrus both the juice sacs and the outer tissues of the rind seem to be formed by outer histogens.   If only Layer III is of the Washington type, Washington characters should not appear in the juice sacs.   If this histogen frequently emerges to form stripes on the rind, after the development of the fruit is well underway, the stripes would not necessarily be correlated with underlying pulp characters.
      Shamel, Pomeroy, and Caryl, (1929) described other citrus types which seem to be periclinal chimeras.   The fruit rind of the Seamed form of the Washington navel shows narrow, longitudinal grooves, which may be caused by the emergence of a mutant Layer III.   The Dual selection of Washington navel has patches of smooth and rough rind, but not regularly in longitudinal sectors.   The smooth areas sometimes form an equatorial zone, between polar caps of rough rind.   These patterns suggest a chimera in which the inner component emerges frequently during the earliest stages of ovary formation, but seldom thereafter.
      Frost and Krug (1942) described two cytochimeral forms obtained from a bud variation branch on a hybrid mandarin tree.   Chromosome counts of root tips, pollen mother cells, and leaf and stem primordia indicated that three histogenic layers were present and that the ploidy of Layers I, II, and III in the two forms was 2-4-4 and 2-4-2.   Root tips from stem cuttings were 4n from the 2-4-4 plant and 2n from the 2-4-2 form, as would be expected if such roots are produced from Layer III.   Growth habit was lower and broader in the 2-4-4 plant than in the 2-4-2 form.   The origin of inner leaf tissues was uncertain; cell divisions in layers other than epidermis showed the 4n condition in both plants, suggesting that only Layer II is involved, but this is contrary to evidence from green-and-white leaf chimeras already discussed.
      Evidence that the pink-fleshed Thompson and Foster grapefruits are periclinal chimeras has been obtained by Cameron, Soost, and Olson, (1964).   These varieties arose, respectively, from the white Marsh and the white Walters grapefruits, by somatic limb variation.   In the Foster, the rind shows color, but in the Thompson it does not.   By further somatic variation, the Thompson has repeatedly produced red-fleshed forms such as Henninger's Ruby and Webb's Redblush (Waibel, 1953), which appear to be identical.   In all of these colored varieties, lycopene and carotenes are the coloring compounds (Khan and MacKinney, 1953).   Nucellar seedlings of Thompson, tested over several seasons in California, show no pink or red color in their fruits and appear to be identical to the Marsh.   Nucellar seedlings of Foster, produced both in Texas and in California, have fruit with red flesh color like the Redblush and continue to show color in the rind.   On the other hand, nucellar seedlings of Ruby, Redblush, and others of the red-fleshed group reproduce the fruit color of their seed parents without change.   It appears that a mutant color factor is present in the Thompson in histogenic Layer I; this layer provides at least part of the cells of the juice vesicles (see chap. 1) but should not form subepidermal rind tissues.   The factor is not present in Layer II, from which nucellar embryos apparently arise.   In the Foster, a factor must be present in Layer II, but perhaps not in Layer I.   Nucellar seedlings of Foster, and those of the Redblush group, should carry the factor in all layers.   Occasional sports to Redblush, produced on Thompson trees, would then represent the substitution of a cell lineage from Layer I into Layer II, and possibly also into Layer III.   The fact that the juice vesicles are pink in Thompson and Foster and red in Redblush may be related to interaction among the tissues involved (Purcell, 1959).
      An unusual pink sport of the Shamouti orange, called Sarah, also contains lycopene and carotenes, rather than the anthocyanins commonly found in pigmented oranges (Monselise and Halevy, 1961).   As in the red grapefruits, these carotenoids occur in the rind, septa, and juice vesicles, but apparently not in the filtered juice.
      Dr. K. Mendel (personal communication) has stated that the Sarah shows typical characters of a chimera.   Some branches bear fruit which is essentially Shamouti in type, while others carry fruit with pink color in the albedo and segment membranes.   One budded tree shows further color variation in the juice vesicles.
      Sinclair and Lindgren (1943) found that cyanide fumigation at an early stage of flower-bud development often produced abundant and varied sectorial rind variations.   Although the variant characteristics indicated or simulated genetic change, perceptible effects were limited to the immediate crop season; shoot buds appeared to be unaffected.

      Diploidy is the general rule in Citrus and its related genera, the somatic chromosome number being 18 (see chap. 4).   However, forms with increased chromosome numbers have been identified or produced by several workers.   Such forms are useful in genetics and breeding.   Longley (1925) was the first to report that a form established in nature, the Hongkong wild kumquat (Fortunella hindsii Swing.), was tetraploid.   Frost (1925) described "thick-leaved" forms among the nucellar progeny of seed parents used in hybridization and determined cytologically that such plants of Paperrind orange and Lisbon lemon were tetraploids.   Frost (1926a, and cited in Krug, 1943) reported tetraploids in five species of Citrus and in Poncirus.   Lapin (1937) in Transcaucasia, U.S.S.R., identified tetraploid seedlings in a series of Citrus species and relatives.   Nakamura (1942) reported two tetraploids, the Shikinarimikan and an apparent tetraploid Sampson tangelo.   Triploids were reported by Longley (1926), Lapin (1937), and Bacchi (1940).   A hexaploid and a pentaploid have also been reported (Lapin, 1937; Krug, 1943) as well as occasional aneuploids (Krug and Bacchi, 1943).   Some of these forms are discussed below and in the section on breeding.
      Tetraploids.—Among many citrus varieties used for breeding at Riverside, California, from 1914 through 1916, nearly all seed parents (except several which gave few nucellar seedlings) produced one or more tetraploid seedlings (Frost 1926a, and unpublished).   About 2.5 per cent of some 3,600 nucellar progeny were tetraploid.   Open-pollinated seedlings from various hybrids, including the hybrid mandarins Kara and Kinnow, and from highly apogamic oranges, including the Washington navel, also gave seedlings that from their resemblance to cytologically determined forms are almost certainly tetraploid.   In Russia, tetraploids identified by Lapin (1937) occurred in eight Citrus species or varieties in percentages ranging from less than 1 to 5.66.   Lapin also found thirteen tetraploids among 353 known nucellar seedlings of Poncirus trifoliata.
      In Italy, Russo and Torrisi (1951a, 1951b) obtained spontaneous tetraploids in Citrus limon (L.) Burm. f. and C. aurantium L.   Furusato (1953a) found a low percentage of tetraploids among seedlings of four Citrus species.   Although the leaves were thicker and darker green than in sister diploids, he stated that the tetraploids within a variety were uniform, having the features of the seed parent, and were presumably nucellar autotetraploids.
      Origin of autotetraploids.—Spontaneous tetraploids appear to occur almost entirely as nucellar seedlings.   The older Riverside tetraploids that have been cytologically verified are of this type, as shown by the absence of pollen-parent characters after cross-pollination and by uniformity within a variety.   The tetraploids studied by Lapin included some of known nucellar origin and none of proven sexual origin.   Nearly one fourth of the early Riverside tetraploids came from seeds which also gave diploid seedlings, and nearly all these seeds came from fruits which also gave other diploid seedlings.   It is thus evident that doubling of the chromosome number occurs repeatedly in the parent trees, often in the tissues of the ovule or ovary.   Doubling does not often normally occur (or at least the polyploid cells seldom persist) in the young embryo, since tetraploid hybrids have seldom been reported from crosses of diploid parents.   Triploids do occur spontaneously from such crosses, indicating that one gamete was already double at fertilization.
      Tachikawa, Tanaka, and Hara, (1961) reported the occurrence of thirteen tetraploids from a total of eighteen seedlings from crosses of the reportedly monoembryonic, diploid Citrus tamurana Hort. ex Tan. and C. iyo, Hort. ex Tan. pollinated by tetraploid Natsudaidai.   The mechanisms responsible for these unexpected tetraploids are undetermined.   (See also [below].)
      Tree and fruit characters of autotetraploids.—Effects of tetraploidy in citrus can be rather accurately evaluated in nucellar selections, since meiosis and gene segregation are not involved.   Except for possible bud variation, such tetraploids differ genetically from sister nucellar diploids only in having the doubled number of chromosomes.   The following descriptions apply to tetraploids studied at Riverside, California.   Tetraploid leaves (fig. 5-17) are considerably broader in proportion to their length than diploid ones, and considerably thicker; leaf color tends to be darker.   The wings of the petioles are usually broader, and in some varieties they are often fused with the leaf blade.
      In young seedling selections and their immediate budded progeny, thorniness is at least as pronounced in tetraploids as in diploids, and the thorns are stouter.   As the selections become older from seed, new growth in tetraploids remains more thorny than in diploids.   Growth is slower in tetraploids; vigorous shoots are less common, and the top is smaller, less erect, and more compact.
      Tetraploids are slower than diploids in beginning to bloom and set fruit, and are generally much less fruitful.   Among the Riverside selections, only the tetraploids of the Lisbon lemon and of some grapefruits have been highly productive, often giving heavy yields, while a tetraploid Willowleaf mandarin has been the least productive of all, usually bearing no fruit.   In the latter tetraploid, excessive dying of the outer twigs occurs, possibly from sunburn.   A tetraploid Dancy tangerine has been very susceptible to dying and shelling of large areas of bark, with associated death of large parts of the tree.   Dying of branches also has been severe in the Ruby orange and Lisbon lemon.
      Many Riverside autotetraploids have now been maintained as budded trees or original seedlings for over thirty years.   Table 5-5 lists thirteen of the selections and indicates their present condition.   The grapefruit tetraploids have been strikingly more vigorous and healthy than any of the other varieties.   While some propagations of King, Dancy, and Lisbon have died, the budded trees of Imperial and Royal grapefruit are about equal in size and vigor to adjacent diploids.   Where tetraploids have been established on both rough lemon and sour orange rootstock, long-term vigor has been much the better on the sour orange.   Tetraploids on their own roots have often been smaller in size than when budded on a rootstock.   Furusato (1953a) obtained tetraploid seedlings of four Citrus species, and found that the main root was thicker and the number of lateral roots fewer than with diploids.   Furusato (1953b) also described orchard trees of satsuma accidentally propagated on tetraploid Poncirus trifoliata rootstocks of evidently spontaneous nucellar origin.   These trees were smaller and yielded less fruit than trees on diploid stock.   At Riverside, a nucellar tetraploid Poncirus seedling has grown vigorously for many years.   However, in an experiment with seedlings of this tetraploid as a rootstock, several Citrus scion varieties have shown marked variation in size and vigor, within varieties, over a fifteen-year period (Mukherjee and Cameron, 1958).
      In most varieties at Riverside (excepting Lisbon lemon and King mandarin), tetraploid fruits have been less elongate than diploid ones, and have shown more basining at the apex (Frost, 1932).   Tetraploid fruits are commonly smaller, especially on weak, outer branches; yet with the Lisbon the tetraploid fruits have been larger than those of the diploid and rougher and more irregular in shape.   Irregular fruit shape seems characteristic of tetraploids, and the fruits usually have thicker rinds (fig. 5-18) and larger and more prominent oil glands.   The juice vesicles are tougher and the yield of juice, on the basis of whole fruit weight, is much lower.   Tetraploid fruits tend to color later, but their flavor at maturity seems little affected, and the content of acid and soluble solids is similar to that in diploids.   In the high-yielding tetraploid Lisbon, acid content has usually been only slightly lower than in the diploid.
      Seed production and overall value.—Seed number is low in some tetraploids studied at Riverside, California, but in the Ruby and Paperrind oranges it has been about equal to that in the respective diploids, and in the Lisbon lemon it has been greater than in the diploid.   When seeds are few, many of them are often poorly developed; when they are well developed, they tend to be larger, especially longer, than in the corresponding diploid.   The number of embryos per seed tends to be fewer in tetraploids, and thus the average size of embryo in well-developed seeds is larger.   There is little evidence as to the proportion of sexual embryos in seeds from selfing.   Degeneration of pollen mother cells and irregular chromosome reduction (chap. 4) should result in fewer viable gametes than in corresponding diploids.   Embryo abortion may also be higher.   The variability in level of seed production among tetraploids suggests that more than one causal mechanism is involved.   If comparatively regular chromosome reduction occurs in the Lisbon, this would favor the observed fertility.   The large fruit size in this form may be a result of its higher seed number or of a general growth tendency which, in turn, favors seediness.   Cameron, Cole, and Nauer, (1960) found a strong correlation between seed number and fruit size in diploid Valencia orange and some other diploid varieties.
      Where normal seed production occurs despite very irregular meiosis, some other factor favorable to fertility is indicated.   In general, when genes are heterozygous, autotetraploids produce smaller proportions of pure recessive gametes and fewer progeny homozygous for unfavorable recessive genes than do corresponding diploids.
      Tree and fruit characters of Citrus autotetraploids show that these forms are not likely to have value for fruit production.   They produce poorer fruit and usually lower yields than corresponding diploids.   They are valuable in breeding, however, since by crossing with diploids some of them can be used to produce nearly seedless triploid varieties.
      Triploids: Origin, Characters, and Promise.—Triploids, like tetraploids, occasionally occur spontaneously among citrus seedlings; they may also be produced almost at will by certain crosses.   Triploids occur as sexual seedlings, however, while tetraploids characteristically arise as apogamic ones.   Frost, in early breeding studies, identified about twenty triploids and observed other probable ones, for a total of more than 5 per cent of about 1,200 hybrids from diploid parents grown to fruiting age at Riverside, California.   Indications of triploidy were also found among zygotic progeny from open pollination.   Lapin (1937) reported eighteen triploids and one modified triploid (4.27 per cent triploidy) among 445 hybrids from pollination of Citrus limon by eight other species and varieties.   He also found one triploid hybrid from C. medica, five triploids among 707 seedlings from open pollination of C. limon and C. paradisi, and five among forty-three seedlings of Shivamikan.   He found one hexaploid from open pollination of Ponderosa lemon.   Poncirus trifoliata as seed parent gave no triploids among 123 hybrids, although 353 nucellar seedlings and 147 other seedlings from open pollination included eighteen tetraploids.
      Triploidy indicates that an embryo has been produced by union of a monoploid (lx) gamete with one having the diploid (2x) chromosome number.   In the formation of spontaneous triploids, the doubled number is no doubt usually provided by the egg, since competition among pollen grains should favor the normal 1x gametes.   From the characters of spontaneous triploids, both Lapin and Frost considered that the seed parent had usually contributed the 2x gamete.
      The frequent occurrence of tetraploids among nucellar seedlings suggests that the doubling which produces triploids occurs prior to the reduction divisions, but it may also result from failure of reduction.   If it occurs before meiosis, modified triploids, with more or less than the exact 3x chromosome number, should sometimes occur, unless such unbalance seriously reduces viability.
      The extra set of chromosomes present in triploids may tend to produce the same general effects, in lesser degree, as do two extra sets in tetraploids, except for the greater sterility expected in triploids.   Triploids, however, result from fecundation and show the high genetic variability which characterizes zygotic progeny in citrus.   Positive identification of spontaneous triploidy by tree and fruit characters is thus difficult.   On the other hand, in controlled crosses where one parent is tetraploid, the occurrence of non-nucellar progeny having thick, rounded leaves and nearly seedless fruit is presumptive of triploidy or near triploidy.
      Triploids can be systematically produced by crosses of tetraploids by diploids, although the low incidence of viable seeds is a major obstacle.   Longley (1926) determined cytologically that two hybrids of a diploid limequat pollinated by tetraploid Fortunella hindsii were triploid.   (See also Traub and Robinson, 1937.)   Frost (1943) reported that a tetraploid grapefruit and tetraploid Lisbon lemon as seed parents produced triploids in crosses with diploid pollen parents.   When the grapefruit was used as pollen parent, very little seed was obtained.
      Russo and Torrisi (1953), in two seasons, obtained 307 seeds from 240 fruits of diploid lemons crossed by two selections of a tetraploid lemon (the Doppio).   Among sixty-three of these seeds, twenty from the cross of Feminello by tetraploid Doppio produced seven triploids, and forty-three seeds from Monachello by Doppio produced twelve triploids.   Tachikawa et al. (1961), after producing tetraploids of Citrus natsudaidai Hayata and Miyagawa-Wase satsuma by treatment with colchicine, made crosses between these tetraploids and various diploids.   In two seasons, diploid Natsudaidai as seed parent crossed by tetraploid Natsudaidai produced thirteen well-developed seeds in twenty-five fruits; the reciprocal cross gave twenty-one seeds in two fruits.   Three diploid, monoembryonic seed parents crossed by tetraploid Natsudaidai yielded forty-two seeds from sixty-four fruits.   From these crosses and one other, twelve triploids have so far been reported.
      At the University of California Citrus Research Center, Riverside, continued crossing to produce triploids has been carried out for several years.   Tetraploids have been used as seed parents and as pollen parents; eight tetraploid varieties and some twenty-four diploid ones have been included.   Table 5-6 summarizes the types of crosses and the seeds and seed germination obtained.   Numbers of developed, but empty seed coats have usually been high when the pollen was tetraploid, regardless of whether the seed parent produced only zygotic or in part nucellar seedlings.   When the seed parent was tetraploid (and partly nucellar) and the pollen parent was diploid, greater proportions of normal-appearing seeds resulted.   Germination of normal-appearing seeds has been variable in all groups.
      Recent cytological studies by J. W. Cameron and R. K. Soost have shown that among the crosses just described, some hybrids which arose from diploid, zygotic seed parents crossed by tetraploid pollen parents are themselves tetraploid.   This is in agreement with the report of Tachikawa et al. (1961).
      Since the chromosomes of triploids are present in triplicate before reduction, the gametes should usually receive unbalanced sets, and much sterility is to be expected.   In the known citrus triploids studied, the fruits are nearly seedless (fig. 5-19).   Some triploids set fruit well, whereas others are very unproductive.   Thus, among seven spontaneous triploid hybrids obtained by Frost, long-term yields have been good in four and very poor in three.   Among twenty-eight putative triploids from controlled pollination of tetraploid grapefruit by diploid Dancy tangerine and diploid Kinnow mandarin, about one fourth have yielded moderately well.   These twenty-eight and a larger number from tetraploid lemon crossed by various diploid pollen parents have all had fruit with very few seeds, averaging from less than one seed up to three or four.   Seed numbers in the parent types range from about eight to more than twenty-five.
      The production of triploids, by crossing tetraploids with diploids, provides an important method of producing essentially seedless varieties of Citrus at will.   Seedlessness is of great horticultural importance and is seldom found among diploids except in a few selected varieties, so that it is desirable to test extensively the possibilities of such crosses.   It is of interest, also, to determine the characters of hybrids in which one parent has provided two thirds of the chromosomes.   The tendency to low fruit yields is a serious practical problem as is the usual low incidence of superior genetic types from crossing.


      Breeding is the artificial control of reproduction and therefore of inheritance.   It involves the selection of parents and usually selection among the resulting progeny.   Breeding and genetics are intimately interrelated; each can further the development of the other.   Although simple selection for desirable forms, without knowledge or control of the pollen parent, may be considered a form of breeding, controlled cross- or self-pollination is usually basic to a breeding program.   Within the last twenty years, "mutation breeding" by the use of irradiation and of chemical mutagens has become an approach of interest with many plants, but it has so far been very little used in citrus.

Problems and Needs in Citrus Breeding
      The particular problems associated with citrus breeding are reflected by the genetic and reproductive behavior of the genus (chap. 4 and previous portions of this chapter).   The most unusual feature is, of course, nucellar embryony.   This mechanism limits, and often essentially precludes, crossing and selfing in many varieties.   When first generation progeny are obtained, their degree of nucellar embryony may be greater or less than in their parents, depending upon the cross.   In addition, various varieties and hybrids such as the satsuma mandarin, the Washington navel orange, and the Morton citrange, are highly pollen-sterile.   Self- and cross-incompatibility are more common than had previously been suspected.   On the other hand, great gene diversity, frequent bud variation, and the overall interfertility among Citrus species provide a wide range of material for breeding purposes.
      Important breeding goals exist in citrus with respect to both scions and rootstocks.   Many needs are of long standing, and resemble those in other tree fruits.   Vigor and longevity of tree and sufficient amount and regularity of crop are important.   In scion varieties, fruit size is a critical factor and many hybrids of otherwise good quality are discarded because the fruit is too small.   With varieties to be eaten out-of-hand, seed content is also important.   Many of the varieties now grown for fresh fruit in the United States have few seeds, with the notable exception of many tangerine types and certain juice oranges.   Seedlessness is difficult to obtain by regular breeding methods, however, since mechanisms related to seed production often affect fruit development.   Fruit shape, rind appearance, and flavor are of major importance.   Season of ripening, storage life, and adaptability to local environment often determine whether a new variety will be grown.   In Japan, the satsuma is a major variety partly because the fruits can be harvested before cold weather begins.   In South Africa, storage life is important, since much of the fruit is exported; in the Coachella Valley desert area of California, high foliage density and fruits relatively resistant to sunburn are essential.
      Most prominent citrus varieties suffer one or more defects, even in growing areas to which they are well adapted; these defects are often difficult to remedy.   The Washington navel orange tends toward low productiveness under many conditions (perhaps as a consequence of its otherwise desirable seedlessness).   The Valencia can suffer from small fruit size and granulation of the flesh.   Grapefruit, except in a few favored areas, is often rather sour and bitter early in its season.
      Certain characteristics being sought in new varieties in the United States have recently been summarized by Cooper (1962).   They include new mandarin types suited to various seasons, new sweet oranges, increased cold hardiness in scions and rootstocks, and rootstocks resistant to fungi, nematodes, and virus diseases.   In Florida especially, orange varieties combining early maturity, high productivity, and high soluble solids are sought for processing purposes.   For use as fresh fruit, oranges attractive in appearance, easy to peel, and high in solids are needed.   Seedlessness is of less concern in some areas than in others.
      Unusually early-maturing or late-maturing new forms are always of interest, since they may fill a need in a pattern of production or marketing.
      In contrast to scion types, successful rootstock varieties must produce many seeds and be highly nucellar.   Stock-scion interactions are of critical importance.   Physiological reactions and disease reactions are relatively specific in relation to the stionic [sic] combinations used, and new hybrids must be evaluated for these reactions.

Breeding Procedures in Citrus
      Selection of Parents.—
The selection of parents in citrus breeding is governed by a number of considerations.   Varieties which are entirely zygotic are to be preferred as seed parents.   The pummelos are such a group, and natural crossing and selfing in the past have no doubt given rise to most of the existing varieties.   Many known or putative hybrids that are apparently entirely zygotic are now recognized among other citrus groups.   They include the Clementine, Kishiu, and Wilking mandarins, the Temple (tangor?), Clement tangelo, Umatilla tangor, and other unnamed crosses.   Varieties which can produce a useful proportion of hybrid embryos have been used as parents; among these are the King mandarin, Ruby orange, and Eureka and Lisbon lemons.   Highly nucellar varieties have been used, especially if they have certain desirable characteristics.   Most of the earlier hybrids which have gained some commercial usage, such as the Orlando and Minneola tangelos and the Kara and Kinnow mandarins, arose from the second and third types of seed parents.   However, three recent hybrids, Robinson, Osceola, and Lee (Reece and Gardner, 1959), were produced from a zygotic seed parent, Clementine.
      Selfing and genetically narrow crossing are expected to produce progeny somewhat similar to their parents; however, as previously discussed, these progeny are likely to be weak or otherwise inferior.   Selfing of more heterozygous F1 hybrids may be useful, but it has not yet been adequately investigated.   Wider crosses and moderately wide, second-generation intercrosses have so far proved the most promising.
      Citrus forms showing characters to be desired in the progeny are commonly used as parents, but there is no assurance that such characters will regularly appear in the hybrids.   A quantitative range of phenotypes is the usual pattern in F1 and to a great extent also in advanced crosses.   Unfavorable characters such as bitterness, unpleasant rind or juice oil, or tough segment membranes may appear when the parents show little of these faults.   Since hybrids often show many differences from their parents, it is sometimes difficult to classify them under a given parent group, such as orange or grapefruit.   An occasional F1, however, does show very predominantly the characters of one parent; an example is the fruit of the Orlando tangelo (grapefruit X tangerine), which in rind and flesh color, flavor, size, and length of season resembles the tangerine parent much more than the grapefruit.
      When a seed parent produces some nucellar embryos, use of a pollen parent with markedly different leaf characters is helpful, since it permits early identification of a part of the hybrids.   Many doubtful cases, however, must be carried to fruiting.   Seedling tests involving in vitro color reactions of bark or leaf tissue (Furr and Reece, 1946) have not been fully dependable.
      Techniques for Controlled Pollination.—Controlled pollination in citrus is relatively easy to achieve.   When certainty of parentage is required for genetic or taxonomic studies, seed parent and pollen parent flowers should be protected against contaminating pollen.   Trees and branches especially likely to set fruit should be used; those which have just borne a heavy crop may not flower, especially in varieties which are alternate bearing.   The earlier and larger flowers of a cluster are preferable, but only those with full-sized pistils should be used.
      Emasculation is easier and less injurious if the flowers are nearly ready to open (fig. 5-20); flowers with dehiscing anthers must be avoided.   Dehiscence usually occurs soon after the petals separate, but in varieties such as the pummelos it often occurs in the closed bud.   Several flowers may be emasculated on each selected twig.   Setting of fruit is perhaps favored by removing other blooms from a considerable surrounding area.   Emasculation is accomplished by gently separating the petals with forceps and pulling off the anthers, avoiding contact with stigma.   Forceps may be periodically sterilized in alcohol.   Emasculated twigs may be enclosed by flat-bottomed paper bags, tied with twine or pliable wire.   To exclude small insects and minimize motion of the bag, cotton pads are often wrapped around the twig at the level of tying.   Unemasculated buds must be eliminated from bagged twigs.   For selfing, bagging alone is often sufficient, without manipulation of the flowers.
      Special storage of pollen is seldom necessary for crosses within the genus Citrus, except for long distance transportation, since most varieties have similar and rather long blooming periods.   Fortunella, however, blooms much later than Citrus in many areas, and Poncirus may bloom earlier or later, depending on climatic conditions.   Kellerman (1915) reported that stored Citrus pollen, collected in Florida and packed in Washington, D.C., showed some germination about six weeks later in Japan.   Toxopeus (1931) stored mature, unopened buds over a drying agent; he obtained 60 per cent germination after three weeks and produced hybrids after four weeks.   Soost and Cameron (1954), using Poncirus pollen, obtained hybrids after 20 and 36 days storage of dehisced anthers at 4° C, with and without a drying agent.
      Pollination may be carried out immediately after emasculation, or up to several days later, depending upon the condition of the seed parent flowers.   Receptivity of the stigma is indicated by an abundant, sticky secretion; unpollinated flowers remain receptive for several days.   Flowers to provide pollen are collected just before opening, and the anthers are allowed to dehisce.   Pollen may be applied by use of the entire flower (with stigma removed) or dehisced anthers may be placed in a vial and the pollen applied with a brush.
      Cross-pollination can be accomplished more rapidly if the twigs are not covered with bags.   Furr, Carpenter, and Hewitt, (1963) found that bees seldom visit citrus flowers whose petals have been removed; they emasculate and strip the petals from receptive, unopened flowers, and pollinate immediately.   Most later, accidental contamination should thus be avoided.
      Artificial Induction of Genetic Changes.—Little has been reported on the use of irradiation or chemicals for the induction of mutations in citrus.   Haskins and Moore (1935) treated seed of five species with X-ray.   Various seedling abnormalities were observed, including bud fasciation, leaves with split midribs, and a high incidence of albinism.   Occasional plants with bi- or trifoliate leaves were noted.   These variations were at least partly temporary effects due to injury to cell processes.   First-season flowering in grapefruit, which is known to be high as compared with other citrus, seemed to be increased.
      Hensz (1960) obtained several thousand plants from irradiated seeds and budwood of grapefruit and of Valencia orange.   In this material, the LD50 (lethal dose for 50 per cent) for X-rayed seeds fell between 5,000 and 10,000 roentgens; for seeds treated with thermal neutrons, the LD50 lay between 6 and 15 hours exposure.   With X-rayed budwood it was near 6,000 roentgens.   Leaf variegations and distortions were observed.   One seedling from Webb Redblush grapefruit had trifoliate leaves.   Another (Hensz, unpublished) produced an albino branch.
      In 1962, Cameron (unpublished) obtained growing shoots from nearly all of 300 Valencia buds irradiated in a nuclear reactor.   The exposures were from 5,800 to 8,300 roentgen equivalents man, derived from a mixture of neutrons and gamma rays.   After five years there was little indication of mutation in the resulting plants.
      Colchicine was used by Tachikawa, Tanaka, and Hara (1961) to produce tetraploids of Miyagawa Wase satsuma and of Natsudaidai.   The latter tetraploid was then used in crosses that produced triploids.

History and Progress of Citrus Breeding
      Work of the U.S. Department of Agriculture.—
The United States Department of Agriculture has carried on citrus breeding programs for more than seventy years, beginning before 1900.   The work may be conveniently divided into three periods.
      The earliest period.—The first recorded artificial hybridization of Citrus was carried out by Swingle and Webber in Florida in 1893, in relation to disease problems, but a severe freeze in the winter of 1894-95 destroyed most of the seedlings.   Impressed by the freeze injury to citrus plantings, Swingle proposed using the cold-resistant trifoliate orange as a parent in crosses to produce hardy hybrids with good fruit.   This project was undertaken, together with a program to obtain loose-rinded fruits resembling orange or grapefruit and intermediate varieties.
      Seeds from crosses made from 1897 through 1899 were planted in greenhouses in Washington, D.C., and the hybrids obtained were tested in Florida.   About 1,780 seedlings were grown, but the number of hybrids was much less since the greater part of the seedlings were nucellar.   The principal results were reported in various papers (e.g., Swingle, 1910; Webber and Swingle, 1905; Webber, 1900, 1906).   Several important features of citrus genetics, especially the general interfertility of wide crosses and the high variability among F1 hybrids, were discovered in the course of this work, and some distinct new varieties of horticultural value were produced.
      Two tangelos, the Sampson and the Thornton, hybrids between the grapefruit and the Dancy tangerine, at once indicated promise for this type of cross.   These varieties have since been grown commercially on a small scale, although they are now seldom planted.   The Sampson has also been used as a rootstock because of its vigorous growth, resistance to gummosis, and high degree of nucellar embryony.
      The crossing of the trifoliate orange, Poncirus trifoliata, with the sweet orange, Citrus sinensis, produced many nucellar seedlings, and some hybrids.   These were highly variable, and several produced fruit with certain good characters, but none was satisfactory in flavor.   The best of them proved rather sour and all had the disagreeable rind oil of the trifoliate parent, as well as an unpleasant flavor in the oil droplets of the juice vesicles.   They were hardy enough, however, to succeed hundreds of miles north of the main Florida citrus region.   Several of these citranges, including the Rusk, Morton, Savage, and Cunningham, have been used as rootstocks.
      Various other combinations of varieties were used in cross-pollination (Webber, 1900), but none of the resulting hybrids proved commercially valuable.   For example, hybrids between orange and lemon were obtained, but they were not especially good either for the uses of oranges or of lemons.
      The second period.—Beginning in 1908, W. T. Swingle and associates produced large numbers of hybrids in Florida, from many combinations of varieties and species (Swingle, Robinson, and Savage, 1931; Traub and Robinson, 1937).   Several hundred citranges were produced, and citranges were backcrossed with sweet orange, giving citrangors.   None were obtained, however, which combined a high degree of hardiness with good orange fruit quality.   The most promising citranges appeared to be a partial substitute for the lemon (Swingle and Robinson, 1927).
      Some citranges have a defect with respect to hardiness, in that early spring leaf flushes may be injured by frost.   Kumquats (Fortunella) were therefore also used in crossing because of their exceptional winter dormancy, which extends over a very long period.   Hybrids between the kumquat and the trifoliate orange appeared unpromising, but crosses with both citranges and limes were named and released (Swingle and Robinson, 1923).   The Thomasville citrangequat is hardy and has little of the unpleasant oil flavor of the citranges.   Glen citrangedin, a cross of the Calamondin with the Willits citrange, is likewise hardy, and produces acid fruit similar to that of the Mexican lime, and is free from bitterness (Swingle et al., 1931).   Three limequats, the Eustis, Lakeland, and Tavares, were introduced.   Their fruits are sometimes classed as limes, but the trees are much hardier.
      Another acid hybrid, the, Perrine "lemon," was obtained from a cross between the Mexican lime and the Genoa lemon.   It appeared hardier than common lemon and was resistant to citrus scab.   Because of other disease problems, it is no longer grown (Cooper, Reece, and Furr, 1962).
      Because of the good qualities of the Sampson and Thornton tangelos, additional hybrids were produced between the grapefruit and mandarin groups.   Several of these were named and described (Swingle, et al., 1931).   Two of them, the Orlando and the Minneola, which, interestingly, arose from sister fruits of grapefruit pollinated by Dancy mandarin, have been notably successful.   They are considerably different from one another in shape, flavor, and time of maturity.   Both are grown in Florida and to a smaller extent in California.   Another hybrid of related background is the Wekiwa.   It is a backcross of Sampson tangelo to grapefruit, with sweet, pink-fleshed fruit.   Other tangelos from this period of crossing have been used in further breeding.
      Reciprocal crosses among sweet orange varieties were also attempted.   Several hundred of the resulting seedlings were selected as possible hybrids, but many were not carried to fruiting (Cooper, et al., 1962).   Since most orange varieties are highly nucellar, it is probable that many of these seedlings were not hybrids.
      The recent period.—In 1942, an extensive series of crosses was made at the U.S. Department of Agriculture Horticultural Field Station at Orlando, Florida.   The parents included mandarins, tangelos, grapefruit, oranges, and other varieties.   Progeny from one cross in particular, Clementine by Orlando, yielded a number of promising hybrids.   In a population of 327 seedlings, a wide range of forms occurred, most of them predominantly mandarin in type.   Three of these, recently introduced (Reece and Gardner, 1959), are Robinson, Osceola, and Lee.   Under Florida conditions, all have large, sweet, early-maturing fruit of orange or red-orange color.   Robinson and Osceola resemble tangerines, while Lee has some characters suggesting tangelo or orange.   All are earlier than the commonly grown Dancy tangerine.   A fourth hybrid, called Nova, was obtained from this same cross (Reece, Hearn, and Gardner, 1964).   It resembles the Orlando in size and shape, but is sweeter and earlier-ripening.   The Page, a cross of Minneola X Clementine (Reece, Gardner, and Hearn, 1963), is early and has many orange-like characters.
      Furr and Reece (1946) in an effort to distinguish between hybrid and nucellar seedlings, applied a color test adapted to leaf extracts to 3,851 seedlings of the crosses referred to above.   Where Clementine was seed parent, all seedlings seemed to react as hybrids.   These workers then crossed Clementine by the trifoliate orange, so that hybrids could be identified by the dominant trifoliate character.   All seedlings produced trifoliate leaves, indicating that nucellar seedlings are apparently not produced by Clementine.   With the same crossing procedure, Temple (tangor?) also yielded only zygotic seedlings.
      Beginning in 1948, the U.S. Department of Agriculture undertook an expanded program of citrus breeding, centered at its field stations at Indio, California, and Orlando, Florida.   Goals and accomplishments of this program are described below.
      Breeding studies reported by Furr et al. (1963) involved 155 different crosses made to produce edible fruit types.   Thousands of seedlings were obtained and grown to fruiting in close field plantings.   Mandarins, sweet oranges, tangors, and tangelos were the parents most often used.   The size of progeny families varied widely; as expected, the degree of nucellar embryony of the seed parents affected the proportions of hybrids obtained.   Table 5-7 shows the principal seed parents used and the numbers of hybrids selected for second test.   Clementine and Temple, both having many good characters and producing only zygotic seedlings, were most often used as seed parents and yielded most of the second-test selections.   The bulk of these had mandarin or tangelo as the pollen parent.   Umatilla, a zygotic tangor, also produced some promising hybrids.   Some hybrids of good flavor were obtained from Kao Pan pummelo crossed by Minneola and Orlando.   This is the same pummelo clone which produced a favorable hybrid, the Chandler, in other breeding work (Cameron and Soost, 1961).   Several thousand flowers of the highly female-sterile Washington navel orange, pollinated by four other varieties, yielded no hybrids.
      Red grapefruits as seed parents were highly nucellar.   However, a few hybrids between them, and between them and red-fleshed pummelos, all had some color in the flesh.   One hybrid of red grapefruit X Foster pink grapefruit had deep red flesh, and two hybrids of this same seed parent X Thompson pink showed slight pink color.   The regular presence of pigment in these hybrids suggests that the color may be rather simply inherited.
      Three tangerine-type hybrids obtained in these studies were described by Furr (1964).   They are Fairchild, Fremont, and Fortune, crosses of the Clementine mandarin by Orlando tangelo, Ponkan mandarin, and Dancy mandarin, respectively.   All seem as coldhardy as the Clementine or Dancy and at the same time are tolerant of heat.   Under hot desert conditions, Fairchild is the earliest ripening, Fremont is second, and Fortune is the latest.
      Resistance to chloride injury was studied in some 500 hybrid seedlings by the U.S. Department of Agriculture (Furr et al., 1963).   In several crosses, at least one parent such as Rangpur lime or Shekwasha mandarin was chosen as somewhat salt tolerant.   The other parent was sometimes a cold-resistant variety, such as trifoliate orange or sour orange.   Original F1 plants were subjected to water high in chlorides, visual symptoms of injury were rated, and the chloride content of dried leaves was determined.   A wide range of injury and of chloride content occurred within crosses, but there was a tendency for more progeny plants to show tolerance when both parents were tolerant.   Thus, among fifty-two seedlings of Rangpur lime X Cleopatra mandarin, thirty had less than 1 per cent leaf chloride and showed rather little plant injury.   A group of the most tolerant selections, including one each from 18 crosses, showed relatively low leaf chloride (from 0.11 to 2.22 per cent) and low defoliation (from 0 to 25 per cent), although the two characteristics were not always closely correlated.
      Testing of varieties and hybrids for tolerance to Phytophthora root rot was begun in 1956 as part of a program to obtain better rootstocks (Carpenter and Furr, 1962: Furr et al., 1963).   Seedlings were exposed to water suspensions of P. parasitica Dastur and the plants were then held in growing beds for selection of survivors.   Water-culture inoculation is a severe treatment, and in most seedling lots only a low percentage of plants survived.
      Among open-pollinated seedlings of existing rootstock varieties there was evidence of tolerance from widely varying sources, although reinoculation did not always give the same results.   Carrizo and Yuma citranges, some pummelos, citrangors, and citrumelos, and some selections of trifoliate orange were among the more tolerant.   Severinia and a Microcitrus hybrid were highly tolerant.   In most of these tests the seedlings were no doubt nucellar, and represented the parent genotype, but with the pummelos, and apparently with some Severinia selections, they were genetic segregants.
      New hybrid seedling populations showed wide differences in survival.   Backcrosses of Clementine hybrids to Clementine showed poor survival.   Suen Kat and Sunki mandarins, certain pummelos, and the Shekwasha, crossed by trifoliate and by certain citranges, often gave high tolerance; in these progenies, hybrid seedlings were identified either by contrast to the seed parent or by the trifoliate leaf conditioned by the pollen parent.
      Other recent hybrids were grown to fruiting, and their open-pollinated seedlings were evaluated by water-culture inoculation.   Again, hybrids involving trifoliate orange were notable for seedling survival.   A few seedlings from hybrids of Clementine X pummelo showed tolerance, but many were susceptible.   These seedlings were presumably all sexual, while those involving trifoliate appear to have been both sexual and asexual.   As with salt tolerance, inheritance of Phytophthora tolerance appears quantitative.   Resistance to trunk and bud-union infection in the field is not always closely related to root tolerance after water-culture inoculation, nor is tolerance to P. parasitica the same as to P. citrophthora (Sm. & Sm.) Leonian.
      Breeding for cold hardiness is being emphasized in the U.S. Department of Agriculture program (Furr and Armstrong, 1959; Furr et al., 1963).   This goal may be approached from at least three standpoints: hardiness of the scion, hardiness of rootstock, and reduction of fruit loss by production of early-ripening fruit which escapes freezes.   Crosses being studied often include one of these approaches together with efforts to incorporate other favorable characters.   Trifoliate orange may offer a source of long range improvement in scion hardiness, but it may be that only advanced crosses can provide acceptable fruit quality.   The satsuma and Changsha mandarins, the Meyer lemon, the kumquat, and existing kumquat hybrids are being used as more direct sources of hardiness.
      Varieties with early-maturing fruits, being used as parents, include the Clementine, Honey, and satsuma mandarins, Hamlin orange, and Pearl and Orlando tangelos.   Preliminary results indicate that Clementine X Hamlin, and Clementine X satsuma produce some early hybrids.   Crosses for cold hardiness in rootstocks have involved the trifoliate orange, the sour orange, and several citranges.   Segregation for cold hardiness was observed among hybrid seedlings of Rangpur lime X Brazilian sour orange.
      Work of the University of California Citrus Research Center.—Citrus breeding studies were begun at the University of California Citrus Research Center in Riverside, California, soon after its establishment.   These studies may be described under two periods.
      The first period.—From pollinations made between 1914 and 1916, H. B. Frost obtained nearly 5,000 seedlings, including about 500 from selfing and the rest from cross pollination (Frost, 1926b).   Most of the first group were asexual; about one fourth of the second group were hybrids, belonging mainly to the tangelo, tangor, tangemon, lemelo, oramon, and mandarin groups.   The parent varieties most often represented were King, Willowleaf, and satsuma mandarins, Dancy tangerine, Imperial grapefruit, Eureka and Lisbon lemons, and Ruby, Valencia, and Maltese Oval oranges.   The King proved to be an outstanding parent for the production of good-flavored hybrids, especially in crosses with other mandarin types.   Three hybrids, the Kara (satsuma X King), Kinnow (King X Willowleaf), and Wilking (King X Willowleaf), all obtained from rather small populations, were introduced (Frost, 1935); they are now being grown commercially to a limited extent.   All have high soluble solids and good flavor.   Kinnow and Wilking have unusually long seasons of use, but are strongly alternate in bearing.   Wilking, especially, can have small fruit when crops are heavy.
      The Pearl tangelo (Imperial X Willowleaf) was also produced from these early crosses (Frost, 1940).   It is early-ripening, with a mild, sweet flavor, although seedy and sometimes small in size.   An early-season orange, the Trovita, obtained from a nucellar seedling presumably from a Washington navel fruit, was described (Frost, 1935).   Under the climatic conditions at Riverside, it is the earliest, low-seeded juice orange yet tested, but it often has objectionably small fruit.
      Two additional hybrids, the Frua tangerine (King X Dancy) and the Dweet tangor (Maltese Oval X Dancy), were described later (Frost and Cameron, 1951).   Frua trees, however, have shown low vigor; fruit of the Dweet is good-sized, juicy, and of good flavor, but too fragile for eating out-of-hand.
      Several other hybrids from this early work were given tentative names, but were not officially introduced because of defects.   One of these, called Honey (King X Willowleaf) is unusually early but commonly too small.   Others include Mency (Maltese Oval orange X Dancy) and Ruddy (King X Maltese Oval).   Both have several good characters, including heavy yielding ability.   All of these hybrids have been used in further breeding.   Hybrids of lemons with mandarins and with oranges were obtained, but as with similar early crosses made by the U.S. Department of Agriculture, they have not found commercial use.
      The later period.—Between 1928 and 1945 several thousand seedlings, mainly hybrids, were obtained from three principal groups of crosses to produce scion varieties.   The parentages were: (1) mandarins and mandarin hybrids such as tangors and tangelos; (2) tetraploids crossed by diploids, which yielded the first triploids reported from such planned crosses; and (3) a series of pummelo varieties pollinated by mandarin types, grapefruit, and other pummelos.
      Within the first group, King, Dancy and Willowleaf again indicated value in producing hybrids of good flavor; Clementine was also promising in this respect.   Among recent hybrids used as parents, Wilking, Dweet, Mency, and Honey proved useful.   Wilking has produced several hybrids with deep orange flesh color, oblate shape, and high flavor.   A few crosses of Temple by Frua have combined attractive, reddish rind color with relatively early maturity, but poor flavors have often occurred when Temple was a parent.   Two mandarin hybrids obtained from crosses of the 1928 to 1945 period were recently introduced (Cameron, Soost, and Frost, 1965).   They are Encore (King X Willowleaf) and Pixie (an open pollination from a cross of King X Dancy).   Both have unusually late seasons of use, and Encore is especially good until late summer under inland southern California conditions.   Pixie, a smaller fruit, is almost completely seedless.
      Triploids were obtained from crosses of tetraploid Lisbon lemon and tetraploid seedy grapefruit as seed parents; the pollen parents were Kinnow, Dancy, and certain lemon hybrids.   Triploidy or near triploidy was verified by chromosome counts in some of these plants (Frost and M. M. Lesley, unpublished; see also under Triploidy).
      From the crosses with pummelos, about 800 hybrids were grown to fruiting.   The pummelo seed parents included ten selections, ranging from several with rather high acidity to one completely acidless form.   Since pummelos ripen several months earlier than grapefruit at Riverside, one aim of these crosses was to produce large fruits that might be used like early grapefruits.   Fruits of good size were often obtained, although they were usually seedy, and many were bitter or of poor flavor.   Most of those with an acid pummelo as a parent were highly acid, but mild, early-ripening hybrids were obtained from crosses with the acidless form (Soost and Cameron, 1961a; see also [above]).   A cross of this form with a more acid pummelo produced the Chandler, which is pink-fleshed, early, and non-bitter (Cameron and Soost, 1961).   Unusually early, juicy hybrids were obtained from the acidless pummelo crossed by Frua.
      Secondary testing of certain hybrids from the crosses discussed above is still under way.   Among the most interesting ones are those with unusual seasons of ripening.   One hybrid of Clementine X Honey is earlier than either parent.   Several hybrids of Wilking X Dweet combine large size, firmness, and good color.   A King X Dancy hybrid gives high yields in the hot desert.
      Since 1947, the Citrus Research Center breeding program has included several goals.   There have been additional crosses to produce orange hybrids and mandarin hybrids.   Narrow crosses between Lisbon and Eureka lemon selections are under study to determine whether vigorous hybrids, which maintain typical lemon characters, can be obtained from these parents.   Expansion of crossing to produce triploids from several parent varieties has been carried out (see [above]).   Because of the need for vigorous, virus-free Washington navel orange sources, several new nucellar seedling selections were originated from a series of seed parents in 1949.   Hybrids for possible new rootstocks, involving trifoliate orange, Clementine mandarin, Ruby orange, Citrus ichangensis Swingle, Citrus taiwanica Tan. & Shimada, and other parents, have been produced.   The evidence that a very low-acid pummelo can contribute to low acidity, and therefore to earliness in its hybrids (see above) has led to advanced crosses of the present F1 plants with mandarins and with grapefruit.   Crosses of an acidless orange with mandarin types and with other oranges have been made.   The early-maturing grapefruit hybrid, Sukega, which is entirely zygotic, has been used in several groups of crosses.
      Work of the University of Florida Citrus Experiment Station.—The University of Florida Citrus Experiment Station undertook limited citrus breeding in 1924, with emphasis on the production of acid varieties (Traub and Robinson, 1937).   Recently, new studies have been initiated (A. P. Pieringer, unpublished).   Nucellar seedling selections from commercial varieties are being established from monofoliate seedlings after pollination with the trifoliate orange.   Early-maturing, non-bitter grapefruit varieties are being sought in a cooperative breeding program with the U. S. Department of Agriculture.   Basic studies on means of identification of species and hybrids are under way.
      Studies in countries other than the United States.—Citrus breeding programs were under way in several areas of the world by about 1930, but World War II caused the interruption or abandonment of many of them.   In Java, Toxopeus (1931, 1933, 1936) conducted hybridization and related studies, with the primary aim of obtaining better rootstock varieties.
      In the Philippine Islands, Torres (1932, 1936) attempted to produce superior scion varieties resistant to diseases and insects.   He reported weak hybrids from certain mandarin crosses and from crosses of Calamondin with mandarins.
      Large-scale crossing was carried on in the Transcaucasian region of Russia during the 1930's, with the main objective of producing hardy and early-ripening varieties (Luss, 1935; Sukhenko, 1936).   Considerable evidence obtained by Lapin (1937) on polyploidy and on nucellar embryony has already been cited.
      Frequent reports on citrus breeding and selection have appeared in Russian journals since World War II.   Various hybrids have apparently been obtained, but interpretation of methods and results is difficult.   Kapcinelj (1948) mentioned a hybrid orange, called Pervenec.   Zorin (1949) reemphasized the importance of hardiness; he stated that several hardy hybrids had been obtained from Kin Kana crossed by Unshu (Unshiu) and that Unshu X Siva Mikan had produced a hardy mandarin.   He referred to a combined method of vegetative and sexual hybridization.   Later, Zorin (1959) stated that hybrids of three parents had been obtained by this method.   Mamporija (1960) described a vegetative hybrid (synthetic chimera?) which arose at the junction of Unshu budded on trifoliate.   He stated that neither the sexual nor the nucellar seedling progeny of this plant segregated for new characters, but that they were tetraploid, with thirty-six chromosomes.
      Programs of citrus improvement in Japan were long centered on selection among satsuma variants, but in recent years new breeding programs have been initiated.   At the Horticultural Research Station at Okitsu, the satsuma, Washington navel, and certain seedy varieties have been crossed with each other and with Poncirus (Nishiura and Iwasaki, 1963).   Bud pollination was used to increase seed production in part of these crosses (Iwasaki, Ohata, and Nishiura, 1954); some 8,000 seedlings, including nucellar ones, are currently being studied.   Marked differences in parthenocarpy and in degree of seediness were observed among seed parents after cross-pollination.   Among a few hybrids obtained with satsuma, Iwamasa (1966) found segregation for plants with undeveloped anthers in a ratio suggesting single gene inheritance (see also Nishiura, 1964).   Twenty hybrids, involving nine pollen parents, included twelve with normal-appearing anthers and eight with undeveloped ones.   Among the twelve, some produced very little viable pollen, as does satsuma itself.   Figure 5-21 shows flowers of a hybrid with undeveloped anthers, contrasted with flowers of its parents.
      Segregation for male sterility was also observed by Iwamasa among forty-six hybrids of satsuma crossed by trifoliate orange.   In these plants, anthers were developed, but in twenty-six of the forty-six the pollen mother cells degenerated and no mature pollen grains were present.