Next Article
Previous Article Rosebay Index Next Article

Hybridizing Rhododendrons: A Recipe for Success

Noel Sullivan
Late of Burnie, Tasmania, Australia

Reprinted, adapted and taxonomy updated, with permission, from The Rhododendron, March 1989, v.28:3. Official Journal of the Australian Rhododendron Society Inc.

Why hybridize?

To arrange the genetic makeup by the natural reproductive process to create a new plant better suited for the needs we have in mind, whether it be a bigger truss to win at shows or a plant designed for survival in a different climate. As for the adage ". . . that a hybrid is two species spoiled," surely what we can do in a few years might well happen in the millions of years it might take for the species involved to be in sufficient proximity for the same result to occur naturally, and for the resultant hybrid swarm to sort itself out. The result would be called a species.

Rhododendron fortunei spp. discolor
'Rhododendron fortunei spp. discolor in Westport Point, MA

Why bother when there are already too many hybrids to even remember them all? Apart from the sheer pleasure in this "do it yourself world", what about new, brighter colored flowers on a smaller plant with better foliage to allow a higher planting density in our now smaller "do it yourself" gardens?

Fruiting apple trees are hybrids developed to produce quality fruit and we know that seedlings grown from their pips will rarely attain this standard of excellence. Rhododendron hybrids were made for their floral attributes and the seeds they set with the help of the bees are just as unreliable. To even maintain the standard, they need the correct partner. But how do we find it?

Rhododendron hybridizing could not be termed an exact science for, as amateurs, we lack the financial resources to study and tabulate the many millions of genetic influences at work within a genus of nearly a thousand species. We would wish to know which genes are dominant or recessive for any characteristic, whether it may be height, leaf shape, color and number of flowers, etc., when any two species are crossed; and it will be different for any other two.

We lack sufficient data to program a computer to answer our dilemma. It might be called an art and with any art form there are guidelines for successful results. To cross pollinate rhododendrons without some knowledge of their genetic programming is like buying lottery tickets and expecting to be a consistent winner.

Start By Studying and Observing the Species

The species are the basic building blocks; by definition they are a set of fixed individual characteristics. Study a species, preferably with the textbook in hand, and note all the things that make it differ from other species. Look at R. griersonianum, used extensively in the production of warm reds, free from the blue cast seen in the old hardy hybrids. Take note of the flower color that bleaches rapidly in strong sunlight and which will also release its yellow fraction for strengthening other yellows. This is seen well in the hybrid 'Lascaux'--the bud, the leaf, the plant height and habit.

Now, with all this printed indelibly in your mind, look at its hybrid progeny, first and subsequent generations, and you will detect the dominant genetic characteristics that its offspring have inherited. BUT--but modified or even suppressed by the dominant genetic characteristics of the other parent. This knowledge and its application leads to successful hybridizing, for the function of genes is to program the living cells to build and maintain an organism to a predetermined design.

Genetic engineering will no doubt be used extensively in major profitable undertakings such as food, but probably never in the redesigning of rhododendrons. We will have to use the genes as they exist within the species.

Toward Developing a New Hybrid

The aim of this article is to enable the hybridizer to predict with some mathematical accuracy the results of any intended cross.

In the interests of brevity, to illustrate this method we will limit our instruction to the breeding of a hypothetical smaller plant with upright, if not tight, trusses in orange- yellow tones. This is our aim. It is somewhat difficult in that very few species, and here we are concerned with elepidotes, are orange. If we attempt to simulate orange from red and yellow, we must remember that flower color is not like an artist's palette but an ever-changing three dimensional pattern of living cells in tissue and sap. We must eschew the many reds with blue tints and even more that will be too large. Among warm reds are many small species, but most have few flowers in a lax truss. The yellows are fewer in number, some too tall, and the smaller they are, the laxer the truss, e.g., R. caloxanthum. All this is counter productive to our aim.

The obvious starting point is R. dichroanthum. It is a variable species; good in that it never exceeds two meters/six feet, and as the name suggests, has a flower of two colors--yellow overlaid in part by red--so that it appears orange. The drawback is the truss, very loose with four to eight flowers, usually less, and the color can be dull. All this is very dominant, so to achieve our aim we need some additive to lift the truss.

Lax trusses on larger plants can be very attractive for you look into the flowers. On smaller plants they can be disastrous for not many can be persuaded to lie on their backs so that they can fully enjoy all the floral details. Upright flowers, even in small trusses, are more desirable on plants below head height. An upright truss needs a strong and substantial rachis to support the overburden. The flowers should number ten or more, depending on their shape and size. The pedicels should be short and/or strong to hold the flowers upright, the more flowers the less this applies. These details become part of our genetic input.

What we need is a formula, a recipe as in cooking, listing the ingredients in proportions--one spoonful, fifty grams, two cupfuls, etc.--to make a cake. Our cake is that orange-flowered plant that we must construct from our building blocks, the species. But can we use a spoonful or a fraction of a species? The key lies in the hybrids.

To digress momentarily, Rothschild successfully crossed his 'Naomi' with yellow species to produce 'Idealist', 'Carita', and 'Lionel's Triumph'. A very simple and successful method of hybridizing is substitution of one parent with another, but be careful that the substitute differs only in degree. Thus 'Naomi' x 'Fabia' gave 'Ayer's Rock' and 'Naomi' x 'Champagne' gave 'Ripe Corn'. Therefore 'Naomi' x 'Lascaux' should be successful, provided you grow on sufficient offspring to exploit all the genetic possibilities.

Back to our theme. 'Lady Bessborough' is the union of two species: R. campylocarpum and R. discolor. Two units contributed genes to program a different unit or entity, which in turn can pass on those genes that it inherited from both parents. We could express this as a recipe. For 'Lady Bessborough' it is equal parts of both parents but, since the Lady is a unit, better expressed as R. campylocarpum 1/2, R. discolor 1/2.

Cross 'Lady Bessborough' with R. wardii. Again we unite two units and you finish with another unit made up of two halves, so the original halves on one side are now diluted to quarters. The formula now reads: R. campylocarpum 1/4, R. discolor 1/4, and R. wardii 1/2. This plant is 'Crest'.

Cross 'Crest' with R. yakushimanum and, following the rules above, the formula reads: R. campylocarpum ?, R. discolor ?, R. wardii 1/4, R. yakushimanum 1/2. We now use this formula to make an assessment of the genetic input in the areas that are of concern to us. Take color. Yellow genes have been provided: R. campylocarpum ? plus R. wardii 1/24 = total ? white, and R. discolor ? plus R. yakushimanum 1/2= total ? predicted color white with a trace of yellow. We could analyze the height in a similar manner and leaf shape, etc.

Writing this on New Year's Day [1989] in an unusual season, I go outside and examine this unnamed plant. It mimics the formula: the flower is a very pale yellow, the plant compact, and the leaf form follows the dominant parent, R. yakushimanum. Might this plant be a stepping stone towards our aim? What could we cross it with? We could cross it with anything provided that the total formula when analyzed tallied with the formula for our aim. This formula would show all the ingredients, the species, combined in the proportions to produce our hypothetical plant.

Let us look and see if there are any flowers out that may fill the bill. 'Dido'? No. 'Tortoiseshell Wonder'? No. What about this unnamed 'May Day' x 'Margaret Dunne' with bright orange but very open trusses of twelve flowers? Analysis of the total cross involving seven species gives a result of 7/16 for white and 9/16 for color, being 1/4 red and 1/4 yellow. Total analysis is a prediction that the average result of this cross suggests an orange flower in an open truss of twelve with the plant never exceeding three meters/nine and a half feet.

With any cross of this complexity, a large number of seedlings must be grown on to see all the genetic combinations. A more in-depth evaluation of the species involved in those proportions may reveal known dominant factors what may alter the final result, but what this percentage evaluation shows is that a visual possibility has become a mathematical probability. Did I make the cross? Of course. To honor the occasion, and if successful, I may even name it 'Day One'. But I won't grow on many seedlings for I have better irons in the fire and limited acreage, and I don't want to put my marriage into great jeopardy.

If you have been brave enough to read this far you will see that with this system you don't have to build with individual bricks. You merely locate two or more prefabricated units. With the ingredients in the correct proportions, slap them together and HEY, PRESTO!

Planning Your Cross

You should not expect to do this off the cuff. It should be planned like a military campaign; you need to assemble your troops or at least know where they are and when available, and you need a plan of attack.

It becomes obvious that you need textbooks or access to them for you will need details of the species. You must know the pedigree of the hybrids for you will notice that the order in which the components have been assembled dictates their proportions. The last added is always half of the total; if you need less it must be added earlier. For example, the extensive use of R. yakushimanum for dwarfing has shown that it invariably strips the color so the best use for this dwarfing keeps R. yakushimanum to below one quarter.

To make your plan, and having an aim, I find it easier to start with the main species involved--in our case R. dichroanthum. There may be more, but treat them separately, and from the Rhododendron Register [H. R. Fletcher. London: RHS, 1958, with annual updates.] list all the hybrids that are relevant to the aim. Make up a Graphic Display as illustrated on page 24. My Graphic Display for R. dichroanthum has many hybrids that have been left out of the illustration for simplicity. The Display, after photocopying, is kept as a permanent record and saves time, and wear and tear on the books when you have to look it all up again next week. The duplicates are used with colored pens in planning your campaign.

Tentative crosses are examined in various combinations and analysis shows that a certain combination has too much R. dichroanthum. We know from bitter experience that it will make the truss droop, so we attempt to find a better combination that sometimes does not exist. Knowing this, I have already made some of these halfway houses. While they won't be winners, they could be the vital link in the making of one.

Don't burn your experiments until you analyze them for future use. Don't use hybrids of vague parentage, e.g., 'Toucan', an R. eriogynum hybrid, orange-red. Obviously R. dichroanthum is in there . . . but what else? Displays may be made in any form. With many 'Bambi' crosses on the way, I have one with 'Bambi' on the bottom line.

The Graphic Display on page 24 suggests a battle plan and the analysis tells whether you have a winnable war. Let us, in brief, look at the Display but only the right hand side that deals with non-indumented plants. 'Fabia' x R. decorum has produced many plants, but which 'Fabia' and which R. decorum, and even which sibling?

'Tomeka' was used for its strong vermillion color. Bob Malone crossed it with 'Percy Wiseman' and we both grew on some seedlings. With some twenty flowered to date the results are pleasing. It has been given the provisional name 'Emu Valley' grex and predicted results have been confirmed. R. yakushimanum at 25% and R. decorum at 25% balance the height, lift the truss, and make 50% for white. The 50% color opens red and fades with the R. yakushimanum to yellow. The whites have cleared the opacity from 'Fabia' and the truss has good luminosity.

If we add 'Dido' to this blend, analysis shows that the color weakens as does the truss. 'G52' is an attempt to use the very dominant, low-mounded plant to get a better furnished truss than 'Cowslip'. We are assured of plant form and color, but will the truss stay up? Less chance with 'H41', for with R. dichroanthum at ?, its dominance for a lax truss will surely exert this influence.

Too little attention has been paid to the development of smaller hybrids with beautiful shapes and textures in the leaves. Such good-foliaged plants could be worthy rivals of dwarf conifers for pride of place in a garden, and flower as well. Better-furnished plants result when you use species that retain their leaves for more than one year.

An added bonus is indumentum, a beautiful but transient wooly coating on the new leaves that is permanent on the lower surfaces. This characteristic is genetically recessive and only passed on to the offspring when the other parent is similarly endowed. When listings of all the species with indumentum are assessed, we might group them in many ways.

Grouping Potential Parents

Perhaps the most meaningful assembly would be to group those from section Pontica, subsection Pontica with subsection Taliensia and the recently introduced R. pachysanthum from subsection Maculifera. Members of this group are smaller and all have pale or white flowers. There is trouble here with some having a short rachis, and crosses between members will not get you much color.

The other group contains the remainder, a mixed bag, but this is where most of the color and size is. Thus we include section Pontica, subsection Neriiflora and a few from subsection Campanulata, some subsection Arborea, and some big-leafed species. To save space, I will list but a few and hope that you get the message.

Group A. R. caucasicum, R. smirnowii, R. pachysanthum, R. adenogynum, R bureauvii, R. roxianum, R. elegantum, R. recurvoides, R. wasonii.

Group B. R. haematodes, R. mallotum, R. dichroanthum, R. lanatum, R. lacteum, R. griersonianum, R. macabeanum, R. falconeri, R. arboreum var. niveum, R. floribundum, R. simiarum.

Basically use Group A as the seed parent and cross with Group B to introduce color and change the leaf form. Select the best seedling and cross back to a different species in Group A. For example, using R. yakushimanum to reduce the size would surely strip out the color, so try R. roxianum. This is largely unexplored territory with little or no precedent, so we must make some rules to best investigate where the dominant genetic characteristics lie.

Seedlings of the union R. haematodes x R. falconeri show a dominance of the leaf form of the pollen parent. If the waxy red color of R. haematodes is dominant as it usually is, then this could be a good combination as big-leaf plants generally have paler flowers.

Seedlings of the R. yakushimanum x R. pachysanthum combination show a leaf nearer to R. yakushimanum but with a different indumentum. This cross won't indicate the degree of dominance of R. pachysanthum to color inheritance. Much better a trial with R. mallotum or R. pocophorum.

In general, single genes don't control single characteristics; their orchestration is complex. Thus, some factors are inseparable, such as the color and the hanging flower of R. dichroanthum--have one and you have the other. Cross it with R. haematodes and you reinforce. Cross it with something with opposite characteristics and you may neutralize it. Therefore in testing go to extremes. You often learn more.

As I explained before, however, we don't have to build brick by brick. To this end I have been using 'Bambi' as a prefabricated unit of three species. My oldest plant, at eighteen years from a cutting, is now 0.6 m/2 ft high and 1.4 m/4.5 ft wide, well-furnished with deep green shiny leaves and a beautiful orange indumentum. New growth is covered with a light tan wool. The truss of 12 to 14 flowers is well presented but the pedicels are somewhat long so the flowers, which open a coral pink and fade to a soft yellow, look downwards.

A pretty plant and a good plant for experimentation, it dwarfs the parent R. yakushimanum, a plant of younger age. I have crossed it with every indumented plant that I can find, especially species, provided that the input suggests tighter trusses and more color. Yes, even R. niveum and R. floribundum for who knows what may happen.

Back to the Graphic Display on page 24, the left hand side. 'Fabia' x R. smirnowii gave 'Hello Dolly' (Lem, Elliott, 1974). R. smirnowii carries some pink and, unlike R. yakushimanum, the color does not fade the color so 'Dolly' is a deeper yellow with pink overtones but a looser truss. 'Fairweather' (Brockenbrough, 1974) is a cross between an unnamed hybrid with the same genetic input as 'Bambi' (Shown here as "Bambi' for graphic simplicity.) and 'Hello Dolly'. This hybrid has outstanding foliage, heavily indumented, and the flowers are a delightful warm yellow tone. We think the color opaque and lacking in luminosity. This plant is so good that I have done the reverse cross and substituted 'Bambi' for the unnamed hybrid to see what other possibilities there may be.

'Fairweather' crossed with R. lanatum is purely speculative as it would be with R. lacteum or R. wasonii. R. lanatum is an example of a little-used species with great possibilities--an open but compact mound, good leaves, widely bell shaped flowers in trusses of six to ten, pale yellow or purple, red-spotted. Known crosses, but one, with 'Sir Charles Lemon', yields predictable results.

Scrutiny of the ARS Seed Exchange Catalog (a textbook of what is being done) shows awakening interest. The obvious cross with R. yakushimanum would probably strip out most of the color, so "Testing 1, 2, 3" might be R. mallotum for a tight truss of a warm red on a medium sized plant with nice foliage. Or, R. niveum for the same reasons but to check on that latent purple. Or 'Fairweather' to put in more yellow or luminosity. See how easy it all looks when you have the information at hand in graphic form.

'Lascaux' (Hanger, 1947) is ['Fabia' x R. litiense (R. wardii)], flowers bell-shaped, barium yellow with a crimson blotch and no indumentum. It disappeared because the other parent, R. litiense, has no indumentum but some genes for indumentum are still there and reappear when R. yakushimanum is crossed with 'Lascaux'.

When the G23 seedlings flower, the best will be crossed with 'Hello Dolly'. The analysis shows that the average result will have more color than 'Emu Valley' grex.

Straight up the center of the Graphic Display shows the effect of adding R. arboreum for color and truss improvement, and then adding R. adenogynum to reduce the plant size and improve the leaf. The best of G102 seedlings show a leaf nearer to R. adenogynum and the predicted color, a light orange pink.

I have long since passed the one thousand cross mark, and have been privy to at least a thousand more. I have had to get my copy of the Rhododendron Register rebound because of constant usage. Except experience, I have little to show for my efforts, but, hopefully, some expectations for the future. Hence, my desire to save others from the hardship of starting from scratch.

I think I first saw the light at the end of the tunnel when I belatedly read the article by Art Childers in v. 31:3 of the ARS Quarterly Bulletin [Summer 1977, "Breeding for Yellow and Indumented Rhododendrons". P. 138.]. Thank you, Arthur, for being my mentor.


To recapitulate, with a method to guide you, it's never too early to have a definite aim, for you will get all the practice you need on that without any warm-ups on random crosses. Do your planning in advance by finding out what will be in flower next season; you don't have to own the plant, only use it, but it is much easier if it is on your own doorstep. Remember that pollen is easily stored.

Plan your crosses for the coming year, evaluate them, and reject them if they don't show probability. If good, still plan some alternatives for there are many occasions when a late frost really nips your plants in the bud. Keep accurate records including dates and label the plants, for memories fade as the numbers increase.

Finally, if the plants you produce after all this time and effort are to be popular, they must be commercial in that they must be easy to propagate, bush up well, and flower at an early age. Therefore, remember the rules of inheritance and don't use parents that don't have these properties. Lastly, don't make them too small. An advanced specimen of R. aureum x R. proteoides would be worth its weight in gold but still not be a marketable commodity.

P.S. There is another but little-known species with orange flowers. R. viscidifolium, which does not get a mention in the old [RHS] Handbook where they gave ratings. Descriptions may vary between "of little garden merit" or "a curiosity". The dependant truss of one or two flowers, often hidden by the foliage, is a bright copper-orange. On the brighter side we might consider it a larger form of R. williamsianum. Careful manipulation may pass on this color and, hopefully, leave behind less distinguished properties.

Meet the Author: Dr. Noel Sullivan

Noel Sullivan passed away in 2001, having long been a member of the American Rhododendron Society. He was born in Sydney in 1921 and served during World War II for Australia. After the war, he qualified in dentistry and practiced in Sydney. During a visit to Tasmania, Noel fell in love with the island State and decided to stay.

He became a member of the Australian Rhododendron Society, then based in Melbourne. Noel served several terms as President or Secretary for Rhododendron Society Branches in Tasmania. Noel was instrumental in developing the Emu River Valley rhododendron garden project, now a feature attraction of the City of Burnie.

Noel Sullivan contributed papers about hybridizing ornamental plants and rhododendron propagation for specialist journals, and frequently spoke on those subjects internationally. The Australian Rhododendron Society recognized the late Dr. Noel Sullivan in 1989 for his many contributions to the Society with the award of the Society Medal, their highest honor.

Adapted from a note by Mr. Hilary O'Rourke, Emu Valley Rhododendron Society, July 2000. Mr. O'Rourke is carrying forward Dr. Sullivan's hybridizing work.

Previous Article Rosebay Index Next Article The Rosebay Volume XXVIII Fall 2001 official journal of the American Rhododendron Society Massachusetts Chapter