JARS v50n4 - Breeding Cold-Hardy Rhododendrons

Breeding Cold-Hardy Rhododendrons
Peter M.A. Tigerstedt & Marjatta Uosukainen
University of Helsinki
Helsinki, Finland

In the search for cold-hardy rhododendrons suitable for Finland and other comparable climates, authors Peter Tigerstedt and Marjatta Uosukainen of the University of Helsinki are involved in a breeding program begun in 1973 involving thousands of plants. Through an explanation of Mendelian genetic principals, they show the importance of a breeding program that can be followed for at least 25 years and the crossing and observation of large numbers of plants.
Of particular importance in passing on cold hardiness to its progeny is Rhododendron brachycarpum ssp. tigerstedtii , used extensively in their program. The authors include a list of cold-hardy cultivars developed in their program.

Finland is in the far north of Europe, our south coast is on the same latitude as Anchorage and the north of the country coincides with the north coast of Alaska. Rhododendrons are grown in gardens at the same latitude as Fairbanks, perhaps even further north. Our coniferous boreal forests grow a lot of ericaceous shrubs but only one rhododendron, the tiny Rhododendron lapponicum . It grows in the far north, in subarctic Lapland. However, recently our Ledum palustre has been renamed as R. tomentosum (2), and, indeed, viable hybrids have been obtained from the cross R. lapponicum x L. palustre indicating the closeness of the two genera that perhaps should be one. In this article we describe a breeding program of evergreen rhododendrons which was started in 1973 (6). A similar program on deciduous azaleas was started in 1988 and has been reported in this journal earlier (7).
How then is it possible to grow here a number of introduced evergreen rhododendrons that are native to regions 20-30 degrees south in Korea, China and North America? The answer is the Gulf Stream, the warm ocean current in the North Atlantic Ocean that flows out of the Gulf of Mexico. Thus our extreme low temperatures in the winter are not much different from regions much further south, in USDA hardiness Zone 3 of North America. Yet, there is a considerable difference in the daylength. This tends to have a strong effect on the formation of flower buds; in fact some species from the continental Chinese Szechwan and Yunnan mountains are very recalcitrant in flowering when transferred to our latitudes, such as R. przewalskii . Daylength also upsets growth initiation and cessation, thus having a direct effect on spring and fall frost damage. In addition to daylength, the difference in humidity between the place of origin and the place of introduction also affects the introduced shrubs and trees.

Climates Similar to Finland
Nearly 90 years of experimenting with different introduced shrubs and trees at Arboretum Mustila in southern Finland has taught us where in the world one can find climates similar to ours (1, 3). In North America an analogous climate is found within Zone 3, particularly where the zone spreads far north in central British Columbia and Alberta. Also areas around the Great Lakes in Minnesota and Ontario are well matched with ours. But the Midwest has a lot of areas with a much more continental climate than ours, while the eastern part of the same zone is too maritime for us. In Europe, the climates matching ours are found at above 2000 meter elevation in the Eastern Alps and Carpathian mountains. High altitudes in the Caucasus where their rhododendrons grow are also well suited. Much of Siberia is too continental; in fact, you have to look for matching areas further east in the Amur, around the Sea of Okhotsk, Sakhalin and the Kuril Islands. Hokkaido, northern Japan, is too far south for a good match and appears to be a bit too maritime as well. We have learned through experience that introduced rhododendrons from different parts of the world behave unpredictably when transferred to our conditions. The geneticist would call this "genotype x environment interaction."

Bottleneck Years
In breeding rhododendrons we have found that large hybrid families are needed from which to select the extreme types that match our climate. The marginal climate that we have to breed for also features large variations from year to year. We have begun to talk about "bottleneck years" to define years that are particularly dangerous for newly introduced species or selections of new hybrids. Experiments at Arboretum Mustila, which started in the first years of this century, have shown a series of such test winters: 1926/27, 1939/40, 1956/57, 1972/73, 1984/85 and 1986/87 (6).

Genetic Variation in Hardiness
Most of the hardy rhododendrons used in our breeding program were planted in Arboretum Mustila in the beginning of the 1930s and had thus shown their resistance through three crucial winter periods (test winters) before breeding began in 1973. Thus, we conclude that a successful breeding program for hardiness is a particularly long-term undertaking that starts with observations in arboreta and other plantations perhaps 30-40 years back in time. This experience was essential for our breeding program. Hence we were able to identify R. brachycarpum ssp. tigerstedtii [sometimes written R. brachycarpum (as ssp. tigerstedtii )] as THE source of cold hardiness. Had we based our breeding on so-called "hardy rhododendrons" such as R. catawbiense and its hybrids, our success would probably have been only marginal. When selecting the hardiest mothers we also found that within the introduced species you can find large differences among seed sources or seed origins. But even within a defined seed lot, plant individuals differ in many important traits, including hardiness. Thus an initial introduction should always include a population of plants, preferably several hundred individuals, from which to start selection.
We have learned that in crossing two species you may well be lucky with a small number of offspring from the cross; maybe 100 is enough. If, however, one or both of your parents in a new cross are hybrids, as is most frequently the case in rhododendrons, then you rely on genetic recombination among thousands of genes and the chances of finding the right combination is minute. Here the offspring numbers should preferably range between 500-1000! Let us look at a simple example: Say that a desirable trait occurs in the progeny of a cross between two hybrids at a frequency of 1 plant out of 20, 1/20. Say that another equally desirable trait occurs independently in that progeny with the same frequency 1/20. The chance to find a plant that has both of these independent and genetically segregating traits is then 1/20 x 1/20 = 1/400! As we look for more independently segregating traits, say three, we need a progeny of 8000. Here you can clearly see the importance of large numbers when combining valuable genes and traits in new hybrids. Fortunately for us, however, many of the traits that we look for, such as "growth profiles" (see explanation below) or hardiness, are under the control of great many genes that act additively. Thus, for such traits we can expect to find intermediate inheritance between both parents so that if you cross a continental "growth profile" with a maritime "growth profile" you most likely get a progeny which is intermediate and thus has best possible adaptation to a transitional climate between the two extremes. In fact, such a hybrid may be tolerant to climate on a much broader basis than local seed origins from typically maritime or continental regions.

Rhododendron brachycarpum 
ssp. tigerstedtii
Fig. 1. Rhododendron brachycarpum ssp. tigerstedtii ,
one of the original introductions to Arboretum Mustila in 1933.
The senior author inspects pollination bags.
Photo by I. Tigerstedt

The Breeding Plan
A breeding plan is basically a spreadsheet having maternal rows and paternal columns. Particularly the mothers we selected for the plan represented the most hardy material that could be found in old plantations at Arboretum Mustila, mostly planted in the 1930s. We especially refer here to R. brachycarpum ssp. tigerstedtii (Nitzelius 1970). We used 8 species or subspecies and 27 hybrid mothers and a total of 53 individuals as mothers. For hardiness our three best groups of mothers were 10 plants of R. brachycarpum ssp. tigerstedtii , 8 of R. smirnowii and 17 of R. smirnowii Seidel hybrids. Astonishingly, many of the Seidel hybrids appeared to be very adaptable to our climate although Rudolph Seidel used R. arboreum in many of the hybrids. This species contributed deep red flower colors to the hybrids, but it certainly caused frost sensitivity.
We think that our selection of "the ultimate in cold hardiness" on the maternal side gave us a proper background to experiment more widely with pollen donors or paternals. After all, the approach was largely an adjustment to realism; we choose mothers that we knew would flower well, set seed well and had shown their hardiness for 40-50 years. In fact we used the only reasonable mothers at our latitudes. Of course another practical aspect is that pollen can be collected anywhere in the world and transported to the site of hybridization. As we are far north, flowering commences much later than in most places where rhododendrons are grown, so even in the same flowering season we can use fresh pollen as our flowers are receptive roughly in the month of June.
We used a total of 23 species and 48 different hybrids as paternal parents. Pollen was collected in plantations in Sweden, Denmark and Holland in addition to our own sources. After the cold winters in the 1980s we really learned to appreciate the use of hardy parents of both sexes. It seems reasonable to say that frost sensitive species should occur in the parents only as genetic components in hybrids, i.e., their genes should contribute to hardy crosses only after recombination, as e.g., R. arboreum in the Seidel hybrids. At the moment we are in fact making a second round of hybrids, using R. brachycarpum ssp. tigerstedtii as the only mother in more than 30 crosses with yellow hybrids of which pollen was sent to us from the United States early in June 1995. We are thus making a large effort to produce a yellow rhododendron hybrid with outstanding hardiness. So far, the 1973 breeding plan has yielded a full color spectrum from whites to dark reds, but we miss the ultimate hardy yellow! Our earlier attempts to use R. wardii as a paternal has given us plants with insufficient hardiness and somewhat pale apricot flowers. It may well be, that a straight F 1 cross using R. brachycarpum ssp. tigerstedtii does not give us a clear yellow but rather an F 2 cross will be necessary.

New cultivar 
candidate, clone E13
Fig. 3. New cultivar candidate, clone E13, an open pollinated seedling
from R. smirnowii. Inheritance of the corolla blotch suggests a
Rhododendron brachycarpum
ssp. tigerstedtii father.
Photo by M. Uosukainen

The Selection Scheme
There is a German proverb saying, "Kinder machen ist nicht schwer - Kinder haben aber sehr." / To make children is no problem - To raise them though is hard enough! That proverb fits precisely to rhododendron breeding. Our first round of crosses brought us 148 combinations and a total of 496 seed batches and some 14,000 plants to be evaluated. They had to be raised through the greenhouse to the plant nursery, and finally the plants had to be planted out in hybrid orchards for testing and comparison and for selection of promising individuals (ortets). Ortets had then to be micropropagated into a large number of copies (ramets) to be tested in clonal trials. From an initial 22,000 hybrid seedlings about 14,000 were planted in the field. About 37% died a few years after planting leaving us with some 9,000 that survived to the two bottleneck winters of the '80s. The minimum temperatures during the crucial selection years 1984/85 and 1986/87 were down to about -35°C to -39°C (-31°F to -38°F). At that time our hybrids had reached a height well above the snow and were thus directly affected by the frost. Perhaps as many as 70% suffered bad frost damage. In some cases whole families were injured or killed, and in recombinant families variation among individuals was enormous, from complete killing to virtually no damage.
We are convinced, however, that had we started with a much smaller initial breeding material our chances would have been badly curtailed in finding new cultivars combining hardiness with other desirable traits. It may sound ludicrous, but in fact we were very glad to see so much of our material killed or injured during the cold winters. We knew then that what we had left was exactly what we were looking for: combinations of ornamental traits and winter hardiness. We underline again that had we reduced the program at the beginning the chances at the end of the day to find the right new trait combinations would have been reduced accordingly. This perhaps marks the difference between the activities of the amateur breeder, who has to make everything personally, including the field trials, and a professional, who can delegate the breeding program to many collaborators. This is not said to down rate the work that has and is done by rhododendron amateurs around the world. Together we contribute significantly to the breeding of new varieties. But it is also clear that very large hybrid populations must be established for selecting "the ultimate in hardiness." In our case, it was quite easy to produce the hybrid seeds, and the first real struggle against large numbers of plants came at the seedling and nursery stage, where our limited resources were stretched to a maximum. But already at that stage, we had gained the understanding and collaboration of the City Parks Division of the City of Helsinki, which was interested in assisting us with the nursery phase. We had told them in advance that they could get a nearly unlimited number of plants free of charge to be planted in their parks, as long as we were allowed to carry out selection in the hybrid populations. Our collaboration has worked very well, and we can say now that the 14,000 plants that were planted for selection and first evaluation would never have been done without such collaboration. We collaborated with a number of public organizations around the country to get this work done. At least three of our hybrid plantations have in the mean time become popular park areas for the public where people can learn something about plant breeding and enjoy the rhododendrons. One could say now that all parties in the effort have benefited from this collaboration.

Field Trials
Ortets selected had to be cloned by micropropagation to ramets. This operation continued for roughly 10 years, 1982-92. Usually we planted the clones in 4-5 test locations covering the south and central parts of Finland. It is at this stage that the plant breeder can separate genetic effects from effects cause by genotype x environment interaction. This means that the true inheritance of hardiness now becomes apparent. In this way, some 80 candidates have been cloned and planted, and eight cultivars have been named and released. Briefly the releases have following characteristics:
'Elviira'*: R. brachycarpum ssp. tigerstedtii x R. forrestii Repens Group/ compact dwarf, cherry red flowers, tolerance - 34°C.
'Hellikki': Open pollinated seedling from a Seidel R. smirnowii hybrid / medium dwarf, purple, -34°C.
'Peter Tigerstedt': R. brachycarpum ssp. tigerstedtii x R. catawbiense var. album Glass /tall, white with blotch, -36°C (Fig. 2; see front cover photo).
'Haaga': R. brachycarpum ssp. tigerstedtii x 'Dr. H.C. Dresselhuys' / medium tall, pink, -36°C.
'Helsinki University': open pollinated seedling of R. brachycarpum ssp. tigerstedtii / medium tall, pink, -39°C.
'Mikkeli': R. brachycarpum ssp. tigerstedtii x R. smirnowii / tall, white with pink tint, -37°C.
'Pohjola's Daughter': R. smirnowii x R. catawbiense var. album Glass / medium dwarf, white, -34°C.
'Kullervo': R. brachycarpum ssp. tigerstedtii x R. yakushimanum / medium tall, pink-white, -34°C (Fig. 4).

R. 'Kullervo'
Fig. 4. Rhododendron brachycarpum ssp. tigerstedtii x
R. yakushimanum
, released as 'Kullervo'. The view is
from one of the Helsinki city hybrid plantations.
Photo by M. Uosukainen

Our experience tends to indicate, however, that replicates at many locations within the same hardiness zone are really a waste of time and resources. Perhaps two locations with 4-6 ramets per location is a good compromise. However, ramets could have been sent to other parts of the world for testing. Had we known our colleagues across North America, particularly in the continental mid-western states at that time, we would probably have asked for collaborative field testing. It is clear to us that frost hardiness at 60 degrees north under our long days and relatively short summers, may be very different from hardiness perhaps 15 degrees further south where minimum temperatures may be even lower but where other climatic components are different.

Helsinki city main 
hybrid rhododendron orchard
Fig. 5. General view of the Helsinki city main hybrid orchard
established in collaboration with the breeders. This collaboration
has been very successful and has brought much public
goodwill to the rhododendron breeding program.
Photo by P.M.A. Tigerstedt

Lessons Learned
Frost hardiness is a complex trait. There is always strong interaction between genotypes and environments in the hardiness performance. Average annual minimum temperatures, as given in the USDA hardiness zones, may be a convenient measure for zoning, but many other climatic factors play a considerable role. Of these the humidity or continentality/maritimity gradient plays an outstanding role in shaping the genetically determined "growth profile" of the plant. Rhododendron brachycarpum ssp. tigerstedtii , the "ultimate in hardiness" under our conditions, is known to behave erratically in typical maritime rhododendron environments where it flushes early and may be injured by spring frosts. Thus it was ignored by most earlier breeders. Our extremely northern location, at the level with Anchorage, Alaska, and in addition our geographic position between the north Atlantic and the vast land mass east of us over the Ural mountains to Siberia, puts us in a climatic transition zone between maritime and continental. We have found that plants originating in such transition zones generally perform well over a wide range of conditions. Under natural conditions they have been "disruptively selected," to use the terminology of evolutionists. Plants can be "tailored" to such conditions by hybridizing between continental and maritime "growth profiles" that are inherited quantitatively.
Selection from within large hybrid swarms of plants seems to be the only way to attain the "ultimate in hardiness." But selection for hardiness should be performed in the zone where the new material is to be used. We are convinced that had we distributed our hybrid swarms to Midwestern Zone 3, selection for hardiness at our two locations would have picked very different genotypes. Thus in future efforts we suggest that hybrid recombinant material be produced and selected jointly in collaborative breeding programs. In essence this means breeding for wide adaptation, just as it is today performed in many international crop plant breeding institutes around the world. It appears that breeding for wide adaptation in rhododendrons is a worthwhile enterprise, particularly when considering global changes in climate that we are expecting in the future.
Finally, we like to point out that a well planned breeding program is a long and costly enterprise. It is relatively easy to produce the hybrid seed needed. However, the real work burden starts right there through the nursing of plant material through the lab-greenhouse-field. Also the selection of appropriate and large enough test locations for selecting hybrids is an important step, not forgetting planting, rearing and managing of such hybrid orchards. To make reliable hardiness selection the material has to pass through at least one, but preferably two, "bottleneck" winters that clearly wipe out part of the material, perhaps a major part. Such winters must come at a time when the hybrids are already well above the protecting snow level. As we know, such years come at irregular 10-15 year intervals in Finland. Then comes the cloning of selected ortets to ramets that should be evaluated further on at least two locations before deciding on candidates for new cultivars. We started in 1972 by preliminary selection of parents. By 1995 we have named eight cultivars and perhaps a few more may come soon. So the total tenure of this breeding program is about 25 years. On the other hand, a carefully planned breeding program can be continued almost indefinitely. Our hybrid orchards consist of material where further breeding and selection could be done simply by collecting open pollinated seed that would segregate for many traits. Also the hybrid orchards can be used for another round of hand pollination. In this way a breeding program may continue for several rounds of rhododendron breeding.

1.  Hmet-Ahti, L. The geobotanical status of Finland, especially of the arboretum Mustila, in the global scale. Proc. 90th Ann. Symp. Mustila Arboretum; 1993: 45-48.
2.  Harmaja, H. Taxonomic notes on Rhododendron subsection Ledum ( Ledum , Ericaceae), with a key to its species. Ann. Bot. Fennici 28: 171-173; 1991.
3.  Tigerstedt, P.M.A. Adaptability of seed sources across geographic zones - 90 years of experimenting in Finland. Proc. IUFRO Working Parties S2. 02-05, 06, 12 and 14. WA, U.S.A; 1990.
4.  Tigerstedt, P.M.A. Selection for wide tolerance in shrubs and trees. Proc. 90th Ann. Symp. Mustila Arboretum; 1993: 58-61.
5.  Uosukainen, M. Arboretum Mustila - a gene bank for rhododendron breeding. Proc. 90th Ann. Symp. Mustila Arboretum; 1993: 70-80.
6.  Uosukainen, M.; Tigerstedt, P.M.A. Breeding of frost hardy rhododendrons. J. Agric. Sci. Fin. 60: 235-254; 1988.
7.  Vinl, A. Breeding of winter hardy deciduous azaleas in Finland. J. Amer. Rhod. Soc. 48 (2): 94-96; 1994.

* Name is unregistered.