Let's Talk Hybridizing, Taboo or Not Taboo: That is the Question
Stephen Krebs
Painesville, Ohio

In a recent issue of the Journal (Winter 2000, "Let's Talk Hybridizing: Incest among Plants"), Pat Halligan raised some thoughtful questions about the use of inbreeding as a strategy for hybridizing rhododendrons. I too have been thinking about this issue and would like to offer some of my opinions in this forum.

Pat is certainly correct in his assertions that self-pollination of normally cross-pollinating plants such as rhododendrons can result in severe inbreeding depression of plant vigor. Lesser degrees of inbreeding, such as sib-mating (sibling mating), also result in loss of vigor but to a lesser extent (see Table1). As a consequence (if my interpretation is correct) Pat is advocating a breeding strategy that avoids the possibility of inbreeding by ensuring that the parents of each generation do not share identical ancestors. While this approach will undoubtedly assure vigor in his populations, I am less certain that it will optimize the probabilities of successful gene recombinations.

Hybridizers make crosses with the intent of swapping traits (recombining genes) possessed by a particular pair of parents. To cite a common example, a cold-tender yellow is crossed with a cold hardy white, and the desired recombinant offspring are cold-hardy yellows. These individuals result from specific parental genes appearing simultaneously at specific gene loci. The probability of obtaining recombinant offspring is determined according to Mendel's law of random gene assortment during sexual reproduction, assuming no linkage between hardiness and flower pigment genes.

Once F1 progeny are produced from a primary species cross (e.g., catawbiense x wardii using the example above), a number of options are available. One can work "inside" the parental gene pool by selfing or sib-mating, or go "outside" the gene pool by crossing the F1s with a third, unrelated parent. The inside approach creates inbreeding but maximizes the chance that particular sets of genes carried by the parents will recombine. The outside approach eliminates inbreeding, but by bringing in additional sets of genes from a third parent it can "dilute" the probability of recombination among the original parent genes.

Pedigree breeding of rhododendrons for the past 150 years has involved both of the above approaches, whether consciously or not. Breeders who have made primary species crosses (interspecific F1s) often hybridize those plants with additional species in the subsequent generations. Consequently, many contemporary hybrids have very complex genetic backgrounds, with multiple species present (some of David Leach's crosses contain up to fourteen species). At the same time, the pedigree method often generates some level of inbreeding, particularly when hybridizers use cultivars repeatedly in their populations (e.g., 'Mars', 'Crest', and 'Janet Blair'). This increases the probability of common ancestry in a given cross.

I agree with Pat on the general pitfalls of selfing rhododendrons, but it is a strategy that can be useful. For example, I am currently screening hundreds of selfed seedlings of 'Bali', a Phytophthora root rot tolerant clone, with the goal of identifying even higher levels of resistance in some progeny (others may lose the trait altogether). Other cultivars, such as 'Janet Blair' and 'Ceylon', have produced fairly vigorous arrays of selfed progeny growing in my field rows. The point is that while selfing severely reduces the average vigor of a population, there is usually some portion of the progeny (perhaps only 5-10%) that make it through the inbreeding depression "bottleneck." If one needs fifty viable selfed progeny, a minimum 500 seeds might need to be sown to enable selection for the most vigorous at transplanting.

There are "kinder, gentler" inbreeding methods for working inside a breeding population. These include matings between full siblings, parents and offspring (i.e., backcross), and half-siblings. From Table 1, the levels of inbreeding produced from these crosses are on average 50 – 75% lower than from selfing. In my opinion, this is an acceptable amount of inbreeding and a reasonable trade-off for the chance to maximize recombination.

I am betting heavily on a sib-mating approach in my own breeding program. By way of example, consider again the cross between catawbiense and wardii, made with the objective of producing cold-hardy, yellow-flowered recombinants. About fifty vigorous F1s are grown out, with expectation of intermediate attributes (moderately hardy ivories). At flowering, a select group of the F1 is sib-mated (I would choose about ten progeny with the best qualities for this purpose, but fewer could be used). For sib-mating, a bulk pollination is performed. Pollen is collected from each of the ten F1 selections in roughly equal amounts, physically mixed (bulked), and then used to pollinate each of the ten plants. With only a few pollinations required, the bulk method approximates the much more laborious approach of crossing the ten F1 individuals in all possible combinations.

F2 seed from the plants would be combined and grown out in large quantities. For this type of cross, large numbers of progeny are required. This is partly due to the expectation of some depression of vigor among offspring (25% inbreeding level) but also takes into account the likelihood that cold hardiness and flower pigment are complex traits controlled by multiple genes. I would want to have 2000+ seedlings in germination flats to select out the most vigorous seedlings and at least 1000 F2s to evaluate at flowering.

Like Pat, I don't have proof to date that this approach is effective. Nor have I actually made the cross described above (I have used the sib-mating strategy to produce F2 seedlings, not yet flowered, in other populations). But I offer this example of producing a cold-hardy yellow in stark contrast to the pedigree method. The hybrid 'Capistrano', which is David Leach's hardiest yellow, has a pedigree encompassing twenty-two crosses and at least seven species. Could a plant like 'Capistrano' only have resulted from these combinations? I don't believe so, and it would be very instructional to attempt a re-creation of 'Capistrano' using the two-species, two-generation approach described above.

While Pat and I differ on the potential of inbreeding for hybridizing rhododendrons, we both appear to be urging a change in traditional breeding methods. Management of genes in populations, selection for groups of individuals carrying those genes (as opposed to the pedigree method of selecting single individuals), and large population size are important considerations in breeding, particularly for addressing complex traits such as plant adaptation and stress tolerance.

I sense, however, that our proposed methods lack the romantic and biblical appeal of pedigree breeding that has engaged rhododendron hybridizers for over one and one-half centuries. That is, it doesn't involve scrutinizing the studbook by the winter fireside, learning who begat whom, and forming hunches about who should be doing the begetting next spring. For the record, while I am a proponent of change, I am not advocating abandonment of methods that are time proven and...well, just plain fun to do.

Stephen Krebs, Ph.D., is the director of the David G. Leach Research Station of the Holden Arboretum. He is a member of the Great Lakes chapter.