New Rhododendrons through Hybridization and Selection
Gustav A. L. Mehlquist, Storrs, CT
Selection is the oldest known method of plant and animal improvement for one must assume that when man first began to domesticate plants and animals he selected the best specimens he could find. Selecting the best, whatever his criteria were, not only gave him a better start but also tended to result in greater genetic heterozygosity in the material which in turn would increase the variability in succeeding generations and provide further opportunity for selection. Also, natural hybridization among the selections would further increase the variability in the population.
Although it is known that man already in biblical times knew that there were two sexes in the date palm and of course was aware of the different sexes in domesticated animals, there is little to suggest that he was sufficiently aware of the sexual differences within the flowers of the plants that were cultivated to provide a basis for controlled hybridization. Apparently Rudolph Jakob Camerer, better known under the Latinized name of Camerarius (1665-172 1), was the first to indicate a definite knowledge of the sexual elements in plants. In a letter to a fellow professor dated August 25, 1694, he pointed out that in Mercurialis, spinach and hemp, all of which are dioecious, and in corn or maize where the different sexes occur separately on the same plant (monoecious) the pollen is indispensable to fertilization and consequent seed production.
It was nevertheless more than fifty years before an actual experiment in hybridization was recorded. Joseph Gottlieb Kolreuter (1733-1806) is generally given credit for producing the first plant hybrids under controlled conditions. In 1760 Kolreuter produced a hybrid between Nicotiana paniculata and N. rustica, and over the next six years he described 65 crosses involving 13 genera and 54 species.
During the interval between Kolreuters first hybrids in Nicotiana and Gregor Johann Mendels (18221884) work with garden peas, numerous hybrids were produced, both intraspecific and interspecific, but no one provided a satisfactory explanation for the various phenomena observed until Mendel published his account of segregation among seven different features in the garden pea, Pisum sativum, in the 1860's. His explanation based entirely on the mathematics of probability was lucid and simple. His contemporaries, however, failed to understand the significance of his work and it was allowed to sink into oblivion in the library of the monastery in Brunn where Mendel was first a monk and later abbot. The people who would have benefited the most from his explanation were not accustomed to solve what was then considered a horticultural problem by means of mathematics of probability. In other words, they did not have the appropriate training to appreciate this tremendous breakthrough.
It was not until the year 1900 that three botanical scientists, De Vries in Holland, Correns in German and Von Tschermak in Austria, independently rediscovered Mendel's paper. By that time a great deal of breeding data had been accumulated and botanical science had progressed to the point that Mendel's data could be appreciated. As a matter of fact, as is often the case with new discoveries, his work was over appreciated to the point that his explanations were assumed to hold in many situations where later data were to show that they did not. Only a few years after the rediscovery of Mendel's paper, the British geneticist, R. G. Punnett, found that in the sweet pea Lathyrus odoratus some hereditary traits tended to be inherited together rather than independently as Mendel had found in the garden pea Pisum sativum. The American geneticist, Thomas Hunt Morgan, furnished an explanation for this phenomenon by pointing out that if the hereditary traits or genes, as they were called already in those days, were located on the chromosomes, and there was a good deal of data to indicate that this was the case, then, whenever two genes are located on the same chromosome, they would tend to be inherited together rather than separately. This situation would certainly obtain whenever the number of genes exceeded the number of chromosome pairs. This phenomenon which is referred to as genetic linkage is now taken for granted in all organisms, for every organism that has been subjected to extensive investigation has been proven to have a greater number of genes than pairs of chromosomes.
To Punnett, also, goes the credit for discovering and explaining the effect of complementary genes. These genes have one effect when present in different individuals and quite another when present in the same individual. Such genes are widespread in the plant and animal kingdom.
Another example of over appreciation of Mendel's data was the assumption that the rather clear cut dominance of one character over another, the recessive, would nearly always be the case. To be sure dominance is a common characteristic of many kinds of crosses but clear cut dominance and independent segregation in the Mendelian sense are to be expected only in crosses between individuals of the same species, that is, in intraspecific crosses.
In the genus Rhododendron where we have a very large number of different species that are interfertile in varying degrees, one should not only expect more interspecific crosses than intraspecific ones but one should also expect deviations from so-called Mendelian results very frequently.
In the chromosomes, the genes are ordinarily aligned to linear order. Different forms or varieties of a species presumably have the same genes except for those which serve to distinguish these forms and varieties from each other. In two different but related species, the genes may be largely the same but some must be different or the two species would not be distinct and the genes may be aligned along the chromosomes in somewhat different order. In distantly related species, the genes are not only likely to be quite different, but the order of the genes on the chromosomes might differ as well.
Generally speaking, the more similarity in the genes of two species the more readily hybridization can take place and the greater the fertility of the hybrids, whereas the greater the differences among the genes and chromosomes the lower the fertility to be expected in the hybrids. In cases of extreme genetic diversity between the species, hybrids may be obtained with extreme difficulty or not at all.
In a genus such as Rhododendron where there are from 600 to 1,000 species, depending on whose authority one chooses to accept, it is rather surprising that practically all the species can be referred largely to only three different genetic groups, that is, the Vireya section which appears to be quite distinct and the lepidote versus the elepidote sections among the remainder of species.
It probably will not be possible to obtain hybrids between every two species within each of these sections but ordinarily the species of any one series, as presently constituted, will hybridize with species of related series within these sections. Very few crosses have been reported between the lepidote and the elepidote groups and between either of these and the Vireya section. It may be that as our techniques are refined and improved we shall be able to obtain successful hybrids here also.
In this connection, it should be pointed out that interspecific hybrids are, generally speaking, obtained more readily when the environmental conditions are more favorable for the growth of the pollen tubes down through the stylar tissue. Ordinarily the temperature is the most influential factor but light, moisture and nutrients may be determining as well. This clearly implies that a greenhouse where favorable conditions can be more readily obtained is a better place for obtaining difficult crosses than the out-of-doors. Where the probability of success is very low, as determined by repeated failures, it may be necessary to resort to insects for cross pollination. A hive of honey bees or even a few bumble bees can make many more cross pollinations within a given time span than any human operator. Of course, such techniques should be employed only where the two species are quite distinct and the eventual hybrids can be expected to be distinguishable from each parental species.
Insects of course were responsible for the numerous natural hybrids that occurred after man brought together different forms and species which in their natural environment had not been growing within pollination distance of each other. Hybrids are particularly common when seeds are obtained from private or botanic gardens where numerous species are grown within pollination distance of each other. Because most Rhododendron species appear to be at least partially self incompatible, the probability of hybrids when different species and forms are grown side by side is greatly enhanced. In fact, many botanical gardens now warn that seeds of many genera including Rhododendron may contain hybrids due to their inability to provide self pollinated seeds grown under protection.
To the experienced taxonomist, these accidental hybrids present little problem as he usually has the means by which to ascertain what should properly be regarded as the species and what should be excluded, but the average horticulturist or nurseryman does not. As a consequence, the trade is full of hybrids being sold as species simply because they originated from a packet of seed with a valid species name. Of course, the average horticulturist is more interested in plants which by virtue of appearance or easy cultivation are more useful to him than he is in preserving what might be considered the typical form of the species. Often further selection may further modify the appearance of the so-called species so that in time it bears little resemblance to the wild species. This is not necessarily a bad thing but one should bear in mind that what is available in gardens or nurseries under a valid name is not necessarily representative of the species as originally described.
Given this background situation, what should the plant breeders select in order to accomplish his objectives? This, of course, depends partly on what his objectives are and partly on what materials are available. I used to tell my students that the cheapest plant breeding is done in the library. With that I meant, of course, that one can often save a good deal of work by finding out what others have already done. Some people are reluctant to publish what they know, others draw the wrong conclusions and may thus produce misleading information but by scanning the literature and by talking to the people already doing breeding work, one is likely to gather a good deal of information as to what might reasonably be expected from available material. Let us examine what is known from such sources and from my own work.
Albinism expressed as white flowers is of two kinds, pure white and tinted white. Pure white is characterized by complete absence of anthocyanin pigments in both flowers and leaves. It is usually due to a single recessive gene, meaning that when white is crossed to a colored form, the F1 is colored and in the F2 there is segregation of 3 colored to 1 white or about 75 and 25% respectively. At times, there may be somewhat less than 25% white due to less viability of the whites in the seedling stage. Of particular interest to the plant breeder is the fact that because there is no anthocyanin in the leaves, the plants destined to become white-flowered can be recognized in the seedling stage. To my knowledge, this type of white occurs only in certain species of the Azalea, Dauricum, Ferrugineum and Lapponicum series.
Tinted white is characterized by pinkish buds and newly opened flowers fading to nearly pure white within a day or two. In cool weather the color may be stronger and last longer. The foliage may vary from nearly complete absence of anthocyanin to normal foliage color. In R. carolinianum album I have found individual plants with very dark foliage color especially in the fall and winter though the flowers are as white as any. It is not always easy to get exact ratios in populations segregating for this feature due to variation in both flower and foliage color but a single recessive gene is indicated. My own data are based on intraspecific crosses in R. carolinianum and R. catawbiense but this type of white is apparently widespread in the genus, including some species of the Azalea series where white is also found.
In general, lighter flower color is recessive to darker ones though there are many exceptions. In the case of R. yakushimanum, where white is the normal color for the species, the white behaves as a dominant feature. For example, in crosses between R. yakushimanum and R. smirnowii, which is pink, the F1 color is usually the same as that of R. yakushimanum. That is, the bud may be rose colored but as the flower matures it becomes nearly white. Even when R. yakushimanum is crossed to reds, the F1 color is greatly reduced though not necessarily to white. When R. yakushimanum and other species are crossed to R. wardii and yellow hybrids in general, the F1's are usually intermediate. When reds such as 'Vulcan' or 'Mars' are crossed to different color forms of R. catawbiense, the F1's are more or less rose pink. When so-called red R. catawbiense is involved, the hybrids have been deep rose whereas when the white form R. catawbiense album is used, the F1 color is light rose pink. With some forms of R. catawbiense the color has been magenta pink. In no instance have I obtained any real reds in the first generation from such crosses.
Since the reason for involving R. catawbiense is to obtain hybrids with red flower color combined with the hardiness of R. catawbiense, it becomes necessary to grow F2 population from such crosses. To date large F2 populations where several hundred plants have bloomed have failed to produce any hardy real red. There have been reds but not sufficiently hardy to survive out-of-doors. The reason for this failure is probably genetic linkage. That is, the genes for hardiness and the undesirable magenta color of R. catawbiense are located on the same chromosome and probably very close together so that the probability of obtaining a hardy red segregate is small. Since many plants have been eliminated before flowering due to cold, it is assumed that the expected reds are included in that group. However, there is still hope that among the thousands yet to bloom there will occur one or more that combine the red color of 'Mars' or 'Vulcan' with the heat and cold hardiness of R. catawbiense.
Indumentum is a recessive or nearly recessive characteristic. My data indicate that it is due to a single recessive gene, but David Leach has suggested that two recessive genes are involved. It is difficult to get reliable date for this characteristic as to my knowledge there is no species of Rhododendron where both indumented or non-indumented forms occur. All data pertaining to this characteristic, therefore, must come from interspecific crosses which, as I have already indicated, are not always reliable when it comes to determining the number of genes involved. When highly indumented species such as R. yakushimanum and R. smirnowii are crossed to other non-indumented members of the Ponticum series, the F1 leaves are quite puberulent on both sides but not indumented. The puberulence on the upper side washes off toward the end of the summer but the puberulence on the lower side and the petioles may persist for a year or more. Curiously, when these species are crossed to members of the Fortunei and Thomsonii series, the F1 is practically glabrous.
The indumentum may vary in color from almost white as in R. smirnowii through orange in R. yakushimanum to red-orange in R. metternichii and certain forms of R. arboreum. It adds an attractive feature to the foliage especially when the leaves move because of wind or the plants become tall enough to permit some of the leaves to be seen from below. Presumably the indumentum also protects the plants from the effect of drying winds. It is also said to increase resistance to insects especially the lacewing fly. I have not noticed any greater resistance to other insects.
The flowering season, as far as I have been able to get data, tends to be intermediate between that of the parent species in the first generation but, presumably in succeeding generations it might be possible to obtain favorable recombinations.
Cold and heat hardiness or the lack of them are probably determined by a number of genes influencing various diverse physiological processes influencing hardiness directly or indirectly.
Inheritance of size of plant or general stature is of utmost importance to the Rhododendron breeder as there is increasing interest in plants of moderate size. It is generally assumed that the F1 size of plants from crosses of parents that differ in size is going to be intermediate. This may be true but much depends on the nature of the cross or the kind of intermediateness one is talking about.
In intraspecific crosses large size appears to be dominant or nearly so in the first generation. What happens in succeeding generations is not well known since, to my knowledge, little has been published on this subject. My own data from R. dauricum indicate that dwarf and intermediate habit of growth are recessive to the normal tall.
In interspecific crosses, on the other hand, the situation may be different. In crosses between the ordinary tomato Lycopersicum esculentum with a fruit size of about 100 grams and the currant tomato L. pimpinellifolium with fruits of about 1 gram the fruit size of the F1 was not 50 grams as might have been expected but only about 10 grams.
At this time it might be well to bear in mind that there are two kinds of averages, the arithmetic, the usual one, and the geometric one. The arithmetic mean or average is obtained by adding the two parental values and dividing by 2 which in this instance would be (100+1)/2= 50.50. The geometric mean or average is obtained by extracting the square root of the product of the parental values (100+1), that is, the square root of 100 which is 10. Obviously the data from the tomato cross fits the geometric mean but not the arithmetic one.
In orchids where crosses of species that differ greatly in size or number of flowers per spike are common, the geometric mean appears to be a better measure of the F1 than the arithmetic one. When Cymbidium eburneum which rarely has more than one flower per spike is crossed to Cymb. Lowianum which averages 25-30 flowers per spike the hybrids average 5-6 flowers per spike instead of 12-13. When Sophronitis grandiflora, an orchid with bright red flowers about one inch in diameter on a plant 2-2 1/2 inches high, is crossed to Cattleya species with flowers of 8-9 inches in diameter on plants 12-16 inches high, the F1 hybrids more nearly agree with the geometric average in flower and plant size than with the arithmetic average.
The reason for crossing the bright red-flowered Sophronitis to Cattleya species was to obtain large-flowered Cattleya-like flowers of the bright red color that characterizes Sophronitis grandiflora. To date, after more than 75 years of more or less continuous breeding along these lines, we still do not have large flowered red hybrids from this parentage. When the flowers are red they are not large, when they are large they are not red. That is, the gene for red in Sophronitis and the gene or genes for small size must be located on the same chromosome very close together so that these features tend to be inherited as a unit. However, progress has been made in these 75 years and some day we shall have large-flowered reds from these crosses. We just do not know how soon.
Results similar to the above were obtained when I pollinated R. impeditum with pollen from R. dauricum album, R. mucronulatum 'Cornell Pink' and R. racemosum. Since the R. racemosum used was the dwarf Rock form which generally does not grow more than twice the height of R. impeditum the two means are so similar that it would be difficult to ascertain which mean is the best measure in this cross. On the other hand R. dauricum and R. mucronulatum grow about four times the height of R. impeditum. It may be argued that the hybrid plants have not yet reached maturity in the 10 years which have lapsed since the crosses were made but this objection can be met by comparing the hybrids with cutting propagated parent plants of about the same age. From these comparisons it is evident that the average size of the hybrid plants is more nearly that of the geometric mean rather than the arithmetic one.
It should be obvious that if one wishes to study this problem, one must choose parent plants that differ so greatly in size that the arithmetic and geometric means can be distinguished without difficulty.
There is another confusing aspect to this problem and that is the chromosome numbers involved. I know that R. dauricum and R. mucronulatum are both diploid. I do not know the chromosome numbers in the forms of R. racemosum and the R. impeditum that I used. Furthermore some botanists say that what is in the American trade as R. impeditum is in reality a form of R. fastigiatum and this species is listed in the Chromosome Atlas as having either 26 or 52 chromosomes. That is, both diploid and tetraploid forms were found. The problem is further befuddled by the fact that we do not know precisely on what plants Dr. Janaki Ammal made her chromosome determinations.
I regard a reexamination of the chromosome numbers in the genus Rhododendron of utmost importance to taxonomists and breeders. The determinations should be made on clones and selections now being used by taxonomists for reference and by breeders in their effort to develop useful hybrids. The family Ericaceae is not the easiest one in which to determine chromosome numbers but right now it is one that needs it.
Plants with twice the normal chromosome number occur in most plants spontaneously from time to time and the genus Rhododendron is no exception. At least one such plant has appeared in my plantings and several hybrids in the trade look like they might be tetraploids. Tetraploids crossed to diploids might produce triploids which presumably would be sterile. Such sterile triploids would represent dead end lanes as far as further breeding is concerned but presumably would not require dead heading, that is, the removal of dead flowers to prevent seed production.
Conceivably some sterile hybrids could be made fertile through induced chromosome doubling through the use of colchisine but chromosome doubling in woody plants is tedious and time consuming at best and not likely to yield results commensurate with the work involved except in certain cases. Dr. Kehr's 'New Era,' the colchisine-induced R. carolineanum album may prove to be useful in crosses with tetraploid species. As yet I have not had the chance to do much with it except to propagate it for the members of our chapter.
I am interested in the report by Ellis A. Jones to the effect that 'Windbeam' is a triploid with 39 chromosomes and that he succeeded in doubling it by colchisine to produce a 78 chromosome fertile plant. I have not yet had time to confirm the count in 'Windbeam' but I can state definitely that 'Windbeam' is not sterile, partially sterile maybe, but older plants will produce enough seed from open pollinations to provide one with as many seedlings as one can handle. I have presently about 900 seedlings from such seed. They are nine years old and most of them have bloomed for about five years. Some of them look like they might be crosses with a nearby R. carolinianum album but all are interesting. The reason I planted this seed was that one nurseryman told me he grew 'Windbeam' from seed to get more compact plants than one did from cuttings of this hybrid.
One question facing every breeder is to determine how many seedlings to grow from a given cross or self in order to get from the population all that is worthwhile. In intraspecific crosses it is comparatively easy to determine the number of segregating genes from which one can determine the minimum numbers of seedlings required to give the breeder a reasonably good chance for obtaining at least one plant of a given genotype. In interspecific crosses on the other hand; it is more difficult to determine the number of plants that should be grown in the F2 generation because one does not know the number of genes involved and the segregation is likely to be erratic due to chromosomal and genetic aberrations. To get around this difficulty many breeders raise relatively small populations of 50 to 100 seedlings at first and later raise larger populations if results warrant it. Suffice it to say that nearly everyone of my friends who raise rhododendron seedlings raise too few to get everything desirable out of a given cross.