JARS v63n2 - But I Don't Like Pink, Daddy!

But I Don't Like Pink, Daddy!
Paul Rogers
Aloha, Oregon

        Okay, we can change it. And now we're ready to talk about how. In Rogers (2008a), we saw genes are just the "parts- lists" of proteins, one or more of which makes a specific enzyme. Enzymes do the work of synthesizing other biochemicals such as pigments, growth hormones, etc. Using the flavonoid-anthocyanin color pigments as an example, we saw the consequences of mutated alleles that changed how those enzymes function. Malfunctioning enzymes shunt synthesis down specific pathways, to colorless and yellow flavonoids or three of the six anthocyanins that color red, purple, and "blue" rhododendron flowers generally. That explains why most rhododendron species, better known for reds, purples, and what we call blues, also come to have yellow and white members. Alas, it also provides evidence that our purple rhodies are not delphinidin derivatives, otherwise we'd have true blue rhodies. Pigments aren't the whole story of hue though. Genes that control petal pH are almost as important, enabling that chameleon cyanidin to produce red roses to blue cornflowers and Himalayan poppies. We wandered the garden and saw that story expressed in plants, like unimposing 'Peeping Tom'. We are better prepared to observe plants and evaluate what alleles they might bring us. If we're going to be creative about our hybridizing, it helps to be able to recognize and choose the best resources available!
        In Rogers (2008b), we finally looked at intensity, going from the color red to pink. Gene regulation showed us why there is a season to things, why we get picotee and eyes, spots and blotches. We also saw that genes function independently of each other. Concepts of dominance, recessiveness, co-dominance and incomplete dominance should be discarded in favor of understanding how enzymes are functioning. Dominant or recessive has nothing to do with whether both genes on a pair of chromosomes will be synthesizing their particular protein, just how we might perceive it. We also established why there is a unitary nature to genetics. Genes and their alleles are as specific as words and sentences. Genes are carried on chromosomes that are visible and can be watched being duplicated and dividing during meiosis, i.e., during the creation of gametes (female ovules and male pollen). Now we're ready to understand how what happens to those chromosomes can make the perfect rhody for us, at least within the options Mother Nature provides us, of course.
        Everything is discussed as it happens in diploid organisms, which includes virtually every animal and at least ancestral plant lines. If polyploids ring your chimes, understand the process is the same, one just has to account for two, three or more times as many equivalent chromosomes, meaning many times as many ways they would combine. That is relatively complicated, so it's not the best way to begin understanding genetics.

Four-square
The way we analyze and predict the results of a cross is this: with respect to (one or two) genes of interest, we determine all the uniquely different types of pollen or ovules that are produced, and then combine them in all possible combinations. It's called a Punnett Square. Here we apply it to one of Mendel's (1865) prototypical crosses, true-breeding white x purple peas. This is called "the first filial generation", or F₁, the first generation of progeny from the cross.

Table 1: F₁ cross
Punnett Square	Pollen parent Purple (PP)
Seed Parent White (pp)	P pollen
p ovule	Pp Purple flower

We have ignored all the other ways in which the parents may be alike, or different, i.e., tall or short, wrinkled or smooth, etc., and just concentrated on the purple color. We labeled the alleles of the purple parent "P", and the equivalent alleles of the same gene in the white parent "p". Because each parent came from true-breeding strains (and because all their seedlings are purple), we deduce the parents have both alleles identical. The F₁ seedlings inherited one of each allele from each parent. Mendel observed the heterozygotes were all purple, so he called purple dominant, white recessive. Typically, upper case letters are used for dominant factors, lower case letters for recessive factors. The condition where both chromosomes carry the same allele is called "homozygous." If they carry different alleles, they are called "heterozygous." (The seedlings themselves would be called homozygotes and heterozygotes.) The alleles specify somewhat different proteins, or enzymes, which are both read from the DNA and synthesized - they're on equal footing there. Their own performance differs. In the "bucket-brigade" to which I've likened pigment synthesis, one gets its bucket, does the right thing and passes it on. The other fumbles and drops it. We will see in some other processes, the pH of petals for example, the functioning of both alleles may be visible. Now then, Mendel crossed (or selfed, for that matter) these purple F₁ seedlings. This is called an F₂ cross, the second filial generation.

Table 2: F₂ cross
Punnett Square	Pollen parent Pp purple F₁ seedling
Seed Parent Pp purple F₁ seedling	P pollen	p pollen
P ovule	PP Purple flower	Pp purple flower
p ovule	pP purple flower	pp white flower

        Since each of the F₁ seedlings were heterozygous, half the pollen produced by the pollen parent carried the allele we've labeled "P" which works, and half carried one that doesn't, "p". This also applies with the seed parent. There are four different ways those pollen grains and ovules can combine, shown in the field surrounded by the bolder outline. There are three purples to one white progeny. But one third of the purples, the one with "PP", and the white ("pp"), are homozygous like their grandparents and would breed true. Don't overlook that fact. In them, the other allele is lost forever! The other two purples are heterozygous, as the F₁ seedlings are¹. We know the white does not carry one of the "P" alleles because it's not purple. This is the 1:2:1 ratio Mendel discovered, and since mapped into the way chromosomes divide and join.
        Each diploid rhody has one pair of 13 different chromosomes. When gametes are produced, the pairs of chromosomes are divided and one of each is distributed to each pollen and ovule in random combinations. During fertilization, the pairs are restored. Fifty percent of genes are shared between parent and progeny, and between siblings. The probability that any particular seedling will have a particular allele of a gene on a chromosome inherited from its seed (or pollen) parent is 50%! The fraction of a parent's genetic alleles that will be inherited from a seed or pollen parent is also 50%. The probability that any particular pair of seedlings will have the same allele of any particular gene on a chromosome inherited from either parent is also 50%.
        Let me introduce some more new words. The "phenotype", i.e., the appearance of the seedlings with respect to this gene, is purple or white. The "genotype", i.e., the genetic makeup of the individual, is either homozygous or heterozygous with respect to this one gene. The way these particular enzymes work, all three purples have two different genotypes, but the same phenotype. That's not always true, though, as we saw in Rogers (2008b) with codominance and incomplete dominance. Importantly, what we see in the garden when we're selecting crosses is the phenotype. The genotype is generally hidden from us, even if we know a plant's parentage. Unless we have done well-constructed experimental matings, we're mostly guessing what any particular plant's genotype might be. But let's be serious for a moment. For most of us understanding the "science" isn't our goal. We just hope to use it to make our path to better rhodies a little easier. Good guesses, when grounded in the science, may be good enough.

How Did This Happen?
Often we'll use Punnet Squares to show how the alleles involved in a cross can recombine, thereby predicting results. They can also be used to help visualize what has happened when used with actual results from plant crosses. Returning to Dr. Bowers' observation in Craig (2006), commentary in Rogers (2006): "I have found that even white clones (most of which are pinkish in bud), such as 'Catawbiense Album', 'Album Elegans' and 'Boule de Neige', when inter-crossed give a preponderance of purplish or lilac colored seedlings in the F₁."
In Rogers (2008a), Figure 1, we saw how synthesis pathways are interrupted. In Rogers (2008b), we saw the process involves genes, alleles, and enzymes that create stacks of anthocyanins and co-pigments, increasing light absorption and pigment intensity. It is just a further progression of the same kind of processes we saw in Figure 1. The fact that these "white" clones are pigmented in the bud, albeit lightly, shows us they have functional enzymes for making anthocyanins. So our first, simplest hypothesis should be that there are several steps where the synthesis of the more intense colors could be interrupted. In anthocyanin synthesis, it only takes one functional allele at each step to allow synthesis to proceed. We should assume it only takes one functional allele of these "intensifying" genes to produce the necessary enzyme to do the job. These alleles will appear to be "dominant." Science makes this sort of prediction, and then seeks out confirmation of the hypothesis. The observation is explainable by hypothesizing that: 1) 'Catawbiense Album' is homozygous for both malfunctioning alleles of intensifying gene "I", i.e., has "ii", and functional alleles for intensifying gene "J", i.e., has "JJ"; and 2) 'Album Elegans' is just the reverse. The F₁ hybrids would then get one of each, and the synthesis of intensified pigment "sandwiches" proceeds.

Table 3: Dihybrid F₁
Punnett Square	'Catawbiense Album' White (ii JJ)
'Album Elegans' White (II jj)	iJ pollen
Ij ovule	iI Jj Purple flowers

If our hypothesis is right, we don't expect just a "preponderance" of the seedlings to be "intensely" colored, we expect them all to be colored. If not, maybe it's not as simple as just two genes. If the 'Catawbiense Album' were heterozyzous, with "Jj", then we'd expect half-and-half, not a "preponderance." I'll leave it to the interested reader to workout the simple Punnett Square results.
If we really wanted to solidify our hypothesis, how could we test it? One way is by producing F₂ hybrids.

Table 4: Reversion to type
Punnett Square	Purple flowered F₁ Ii Jj
Purple flowered F₁ Ii Jj	IJ	Ij	iJ	ij
IJ	II JJ	II jJ	iI JJ	iI jJ
Ij	II Jj	II jj	iI Jj	iI jj
iJ	Ii JJ	Ii jJ	ii JJ	ii jJ
ij	Ii Jj	Ii jj	ii Jj	ii jj

Each genotype shown in the outlined cells possesses at least one of both functional alleles, thus is "intensely" colored. Each genotype in the white cells possesses two malfunctioning alleles of the genes I, J, or both, stopping the progression, so these plants are nearly white. Note that the ratio of white to colored is 7:9.
Another way we could provide evidence for our hypothesis would be crossing F₁ hybrids back to both parents.

Table 5: Back cross to parent
Punnett Square	Purple flowered F₁ Ii Jj
'Catawbiense Album' ii JJ	IJ	Ij	iJ	ij
iJ	Ii JJ	Ii jJ	ii JJ	ii jJ

        This time we expect an even 1:1 ratio of white to colored, but there are two genotypes of both white and colored phenotypes.
        Again, we only need to go to these extremes if our goal is in fact to establish the presence of certain intensifying genes in these varieties. In our world of hybridizing for beauty, we should understand how this works because it allows us to follow alleles of interest through generations. It may be that one of those is "recessive," and we want to know where it could have gone when we can't see it so we can bring it back with suitable crosses. That's very doable.
        Let's say it's important to determine if an F₂ seedling in Table 1 was homozygous or heterozygous in light of further plans. Cross it to a white, and then just one resulting white progeny will establish that it's heterozygous! This is the general technique for discovering a hidden recessive allele. How many does it take? In the example above, if we raised one seedling, we'd have a 50% chance of producing a white and getting an answer: it's heterozygous. If it's purple, we still wouldn't know. If we raised two seedlings, each with a 50% chance of being white, only 25% of the time would both be purple and we'd still be uncertain. Each additional seedling halves the probability. If we're observant and lucky, we might find one as the result of someone else's work!

Submerge & Resurface
The point has already been made that genes are unitary structures. They may mutate, but they are not "contaminated by association." In reference back to Table 1, suppose we wanted to get the "recessive" allele, "p", from "Species 1" and "transplant" it into "Species 2" which normally has allele "P", but we wanted most of the other genes and characteristics of "Species 2" to remain. We would make our F₁ cross, then cross some of those seedlings back to parental "Species 2", as in Table 6. These seedlings have the same apparent phenotype, but only half are heterozygotes carrying the allele we want. We need to separate them. We do that by backcrossing to "Species 1", using a table quite similar to Table 6. The heterozygotes are then identified by the production of "pp" seedlings. The heterozygotes take the place of our first F₁ hybrids in a backcross to "Species 2," and we go around again, producing a new generation with ¾ of their alleles from "Species 2." We can take this merry-go-round as many times as deemed necessary. However, it's a decade per cycle, and these intermediate seedlings are sooner or later likely discards, unless we like them. Then, to bring back "pp" homozygotes that otherwise carry mostly "Species 2" alleles, just self two of the latest heterozygotes, as in Table 2. This works because genes retain their structure.

Table 6: Back-crossing
Punnett Square	F₁ hybrid Pollen parent Pp
Species 2 Seed Parent PP	P pollen	p pollen
P ovule	PP homozygote	Pp Heterozygote

Genetic load
As we proceed to second and third generations, using seedlings that were "closer, but not quite there yet," we should understand "genetic load." This is the accumulation of deleterious mutations in a population, often 'masked' or not expressed because they are recessive and carried at low frequency in a heterozygous condition.
        Mutations are happening all the time, and mutation rates vary. Frequently, one in a million is said to be a general estimate. That doesn't sound like much. It isn't, if you're watching for one specific allele to mutate in one individual. But Mother Nature isn't watching just one allele or just one individual. She'd be just as happy if any mutation in any individual turned out to be superior! Of course, sometimes a mutation in one individual can make a big difference! Examination of her family tree shows Queen Victoria had a spontaneous mutation of a gene that produced hemophilia, which, being who she was, was passed to the royal houses of Spain and Russia with serious consequences. In Rogers (2008a), I mentioned a single mutation in a human hemoglobin gene causes sickle-cell disease.
        Although some mutations are deleterious, or in some cases even lethal, over time the most advantageous alleles tend to accumulate in organisms by "out-competing" less advantaged ancestral alleles. Long ago, when I was in college, I read an estimate that the average human carries something like ten severely deleterious or lethal genes. It's another way of expressing genetic load. It's generally not as much of a problem as it might seem because the frequency of these alleles in the population at large is very low. Because humans are not hermaphroditic and have social taboos against consanguineous marriages, the probability of two rare alleles coming together (i.e., being homozygous) in any of us is low, being the product of two numbers close to zero. That's even closer to zero. So it's highly likely that where we have a deleterious allele (either inherited or freshly mutated), the companion chromosome we inherited from our other parent will provide the appropriate allele to produce the needed functional enzyme.
        Hermaphroditic plants can use a very different life strategy. All they need to start a new colony is just one seed without deleterious genes and good fortune. Each plant can produce a lot of seed each season, and selfing isn't unlikely as pollinators go from flower to flower on the same plant. Selfing tends strongly to favor homozygosity. When that happens to deleterious alleles, natural selection tends to remove the allele from the population, favoring the frequency of non-deleterious alleles. Consequently, deleterious genes tend to get "weeded out," and genetic loads are generally lower than in mobile, bisexual, higher animals like us that can tolerate higher genetic loads.
        Heursel, in Appendix C of Galle (1985), tells us "hose in hose" is either lethal or sterile; also that it is "sex-linked." Sex linkage is known in certain genera of animals, birds most notably, but would certainly be curious in hermaphroditic plants. Nevertheless, we should be aware that deleterious mutations could be an issue for us.

Selfing vs. Sibbing
Are sibbing (have the same parentage) and selfing (self-pollination) equivalent? No. Let's say we have a pair of sibling seedlings. Each has one chromosome with all its genes, of each pair from each parent. They have four grandparents. A pair of appropriately chosen siblings can carry between them four different chromosomes inherited from each of their four grandparents, i.e., four possibly different alleles of each gene. An individual plant you might self can only have two different alleles.
As an example, let's say we crossed two rhodies that were each the result of two different primary (homozygous) species crosses. Let's consider just one gene, but assume each of the four original species have, or could have, different alleles. Just to make things clear, say the seed parent inherited alleles "A" & "B" from its species parents, and the pollen parent inherited alleles "C"&"D" from its species parents. So the cross produces different combinations of those alleles in Table 7.

Table 7: An F₂ cross of four species
Punnett Square	Pollen parent CD
Seed Parent AB	C	D
A	AC	AD
B	BC	BD

Now, if we chose siblings to cross without being exactly sure what their genotype is, the pairings could be any of Table 8. The variability of seedlings would be considerable.

Table 8: Random Sibling pairings of F₂
AC x AC	BC x AC	AD x AC	BD x AC
AC x BC	BC x BC	AD x BC	BD x BC
AC x AD	BC x AD	AD x AD	BD x AD
AC x BD	BC x BD	AD x BD	BD x BD

Along one diagonal we have pairings equivalent to Table 7 once again, mixing all four alleles. Along the other diagonal we have pairings nearly equivalent to selfing.

Table 9: Selfing one of the F₂
Punnett Square	Pollen parent "BD"
Seed Parent "BD"	B	D
B	BB	BD
D	BD	DD

Which is better sibbing or selfing? It depends.
In our selfing example, while we may be concentrating on just one gene and its "B" & "D" alleles, this tendency to homozygosity is happening in all the other genes as well. We should think about the possibility some of those could be deleterious. While crossing two "BD" seedlings, the outlined box in Table 8, might be equivalent to selfing one of them with respect to all other genes there is still the possibility of some genetic diversity.
If that's not an issue, the questions might be: Are we trying to capitalize on, i.e., "purify", something we've already got a piece of, say allele "D", i.e., trying to produce "DD", in the outlined box in Table 9, or less certain what we want will occur by creating homozygosity in a particular allele? By selfing, we increase the probability of producing homozygous seedlings, but only if the alleles we want are present. If the alleles we wanted were "CC" and its presence wasn't obvious, i.e., it's "recessive," then selfing this "BD" isn't going to do it. However, a quarter of the randomly chosen sib-pairings in Table 8 have the "C" allele in both parents, the upper left boxes. That's equivalent to a Mendelian F₂ cross in Table 2, of which a quarter of the seedlings would be homozygous for "CC." That's certainly better than the zero we'd get from selfing this "BD." I suppose one could say selfing is a way of focusing on something in particular, and sibbing is a way of "not burning our bridges" when we're not quite sure.

Purple x Yellow
A painter would never suggest mixing yellow and purple. That would make brownish paint. However, genes are different, and we can plan crosses informed by the facts science has shown us. In Rogers (2008a), for reasons explored there, I speculated one might cross a yellow on a bluish-purple seeking a more golden yellow. How about bright yellow 'Sunspray' and deep purple 'Smokey', a.k.a. 'Smokey #9'? It's the higher pH genes we see in the purple of 'Smokey' that we'd like to combine with the yellow of 'Sunspray'. Bulgin (1986) surprisingly didn't list any history of such a cross, so we're on our own here. Let's break this down. Our analysis will necessarily be somewhat speculative, based on our best guesses with reasonable rationales, so let's start simply. First we'll just look at the anthocyanin synthesis path shown in Rogers (2008a) Figure 1. Genes for anthocyanin synthesis are one thing, those for pH another. Unless they are in the same linkage group, they are inherited independently.
'Smokey' shows at least one functional allele for each of the anthocyanin enzymes. Both parents, 'Burgundy' x 'Moser's Maroon', being purplish-red, also synthesize anthocyanins. (For demonstration purposes I'll assume 'Smokey' is homozygous.) 'Sunspray' has a broken synthesis chain for anthocyanins. We don't know which enzyme isn't functional, though we might suspect F3'H (Rogers (2008a)) because then no flavonoids would be sequestered further down a broken anthocyanin chain, leaving them for the production of yellow anthoxanthins. Let's use the symbol "a" for the uncertain allele in 'Sunspray' that breaks the chain to anthocyanins (leaving flavonoid anthoxanthins), and "A" for the allele present in 'Smokey' that works. It is obvious 'Sunspray' is homozygous for the malfunctioning allele "a". Assume we've already done the F₁ cross of 'Smokey' x 'Sunspray', and are now selfing or sibbing those heterozygous F₁ "Aa" seedlings. Because they have an "A" allele that allows synthesis to continue, "dominant" in Mendel's terms, the F₁ seedlings have anthocyanin pigments and are red or purple, but not yellow (if 'Smokey' in fact is homozygous, as it seems). In Table 10 we see how the alleles for the malfunctioning gene of 'Sunspray' recombine in F₂.

Table 10: Anthocyanin genes in F₂
Punnett Square	Pollen parent Aa anthocyanin F₁ seedling
Seed Parent Aa anthocyanin F₁ seedling	A pollen	a pollen
A ovule	AA anthocyanin pigment	Aa anthocyanin pigment
a ovule	aA anthocyanin pigment	aa anthoxanthin pigment

        Now let's consider the pH genes we see in 'Smokey'. This is a bit more complicated, but it's a reasonable example of what a hybridizer might do.
        The inheritance of pH alleles is important in this cross, yet we know far too little about it. My own selfing of deep purplish red 'Dan's Early Purple', roughly centroid 256 on the NBS color names chart (Anon. 1955), produced seedlings more like itself, vivid purplish reds, roughly 254, to the reddish side and vivid reddish purples, roughly 236, to the purplish side. That suggests 'Dan's Early Purple' is a heterozygote for an incompletely dominant allele, and it's homozygous for at least one other that separates its vivid purplish reds from true reds. 'Smokey' is still purpler than the vivid reddish purples, suggesting it has at least one more set of alleles of a different gene creating a higher pH, and thus more purple flowers. We can reasonably hypothesize at least three pH genes that separate 'Smokey' from true reds like 'Taurus'. For our purposes here, I'm going to use the symbols "pH1", "pH2" and "pH3" for the genes I think are in evidence. Based on my experience with 'Dan's Early Purple', I suppose they're incompletely dominant, so I'll use "pH1-" for the allele of the first leading to lower pH, and "pH1+" for the allele leading to higher pH, and similarly for the other two. So based on the phenotype we see, hypothetically, 'Smokey' is homozygous for "pH1+", "pH2+", and "pH3+."² 'Sunspray' doesn't give us good clues about its pH alleles. Let's assume worst-case, that it's homozygous for aa "pH1-", "pH2-", and "pH3-". My 'Dan's Early Purple' would be: "AA pH1- pH1- pH2+ pH2- pH3+ pH3+".
        We're assuming 'Smokey' produces ovules with alleles "A, pH1+, pH2+, pH3+". 'Sunspray' produces pollen with alleles "a, pH1−, pH2−, pH3−". So the F₁ cross would be as shown in Table 11.
        Our first answer comes from the color of the F₁ seedlings. Our hypotheses suggest those F₁ seedlings would all be a somewhat purplish red, perhaps something like 'Nova Zembla'? If we get any variations, those could suggest alleles we didn't anticipate, and so should be noted.

Table 11: 'Smokey' x 'Sunspray' F₁
Punnett Square	Sunspray pollen
Smokey ovule	a pH1- pH2- pH3-
A pH1+ pH2+ pH3+	A a pH1+ pH1- pH2+ pH2 - pH3+ pH3-

        Now then, we will self or sib these heterozygous F₁ seedlings, but let's assume one more hypothesis: that the four genes we're dealing with are all in different linkage groups, i.e., on different chromosomes. Then when these seedlings produce ovules or pollen and the chromosomes divide, one of the alleles of each of the four genes will combine randomly with the other three. There will be one of two choices, each of four times, or 2 x 2 x 2 x 2 = 16, different "kinds" of pollen or ovules (Table 12).
        Sixteen different kinds pollen crossed on sixteen different kinds of ovules makes 16 x 16 = 256, different ways of producing the F₂ seedlings! But wait! As we saw in Table 4, many of them are just different ways to the same result. There certainly isn't space here to show the whole Punnett Square, but let's look at a few interesting combinations in Table 13.
        Let's examine the genotypes and phenotypes in just a fraction of what we should expect, as shown in Table 13. The ones numbered 3, 20, 34 and 49 are like their F₁ parents, probably purplish reds. Numbers 19 and 50 are similarly colored, but don't produce yellows. Number 18 is like its 'Smokey' grandparent, bluish purple, and breeds true. Numbers 2 and 17 are similar purples, but they're heterozygous for the functional anthocyanin allele. If they were selfed then a quarter of their seedlings would be yellows like Number 1. Numbers 35, 51 and 52 have all the pH alleles of 'Sunspray' but make anthocyanins. We'd expect these to be as red as we can get from this cross. The pH alleles in 'Sunspray' weren't obvious from its phenotype because anthoxanthins have a different structure than anthocyanins. Notice how red tells us something about the pH of yellows - the anthocyanin acts as an "indicator". Number 36 is like 'Sunspray', yellow, as 18 is like 'Smokey'. Numbers 4 and 33 are yellow, but midway in pH between 'Smokey' and 'Sunspray'.
        How yellow might the seedlings be? Number 1 is the one I speculated might be a golden yellow. It has all the high pH alleles of 'Smokey' but doesn't make anthocyanins, so is yellow. But it tells us about the sensitivity of anthoxanthins to pH. While this cross isn't an obvious one to make, we can see it tells us a lot about the pH genes of yellows, and the effect of other pH genes on yellows. It might also produce a golden yellow!

Table 12: Some F₁ ovule and pollen Combinations
A pH1+ pH2+ pH3+	A pH1+ pH2- pH3+	a pH1+ pH2+ pH3+	a pH1+ pH2− pH3+
A pH1+ pH2+ pH3-	A pH1+ pH2- pH3-	a pH1+ pH2+ pH3−	a pH1+ pH2- pH3-
A pH1− pH2+ pH3+	A pH1− pH2− pH3+	a pH1− pH2+ pH3+	a pH1− pH2− pH3+
A pH1− pH2+ pH3−	A pH1- pH2- pH3-	a pH1− pH2+ pH3−	a pH1- pH2- pH3-

Should we be put off by the prospect of 256 different ways these alleles might combine? As Table 13 has shown, there are many different ways of producing the same combination. We do not necessarily have to raise 256 seedlings to flowering, since every seedling has equal chance to be the golden yellow seedling. However, raising thousands of seedlings would not guarantee us a golden yellow. We may not need to find a golden yellow, as a quarter are expected to be yellows (see Table 10). Many may be phenotypically similar, but some are genotypically "interesting." We could self or sib some of these yellows. Selfing number 36 (Table 13) unfortunately produces all low pH yellow seedlings, but selfing number 4 and genotypically identical number 33 would produce the entire range of pH in their seedlings, including perhaps the golden yellow we hypothesize may be present. To be sure, the more seedlings we raise, the better our chances of finding something interesting. But that is true of any cross. We should not overlook that the existence of just one example proves something is possible, no matter how unlikely! Probabilities are as they are, but don't overlook luck!

Table 13: Some; F₂ seedling crosses
Punnett Square	Pollen parent "F₁"
Seed Parent "F₁"	a pH1+ pH2+ pH3+	A pH1+ pH2+ pH3+	A pH1- pH2- pH3-	a pH1+ pH2- pH3-
a pH1+ pH2+ pH3+	(1) a a pH1+ pH1+ pH2+ pH2+ pH3+ pH3+ (golden?)	(2) A a pH1+ pH1+ pH2+ pH2+ pH3+ pH3+ (purple)	(3) A apH1+ pH1- pH2+ pH2- pH3+ pH3- (magenta?)	(4) a a pH1+ pH1- pH2+ pH2- pH3+ pH3- (magenta?)
A pH1+ pH2+ pH3+	(17) A a pH1+ pH1+ pH2+ pH2+ pH3+ pH3+ (purple)	(18) A A pH1+ pH1- pH2+ pH2+ pH3+ pH3+ (purple)	(19) A A pH1+ pH1- pH2+ pH2- pH3+ pH3- (magenta?)	(20) A a pH1+ pH1- pH2+ pH2- pH3+ pH3- (magenta?)
a pH1- pH2- pH3-	(33) a a pH1+ pH1- pH2+ pH2- pH3+ pH3- (yellow)	(34) A a pH1+ pH1- pH2+ pH2- pH3+ pH3- (magenta?)	(35) A a pH1- pH1- pH2- pH2- pH3- pH3- (reddish?)	(36) a a pH1- pH1- pH2- pH2- pH3- pH3- (yellow)
A pH1- pH2- pH3-	(49) A a pH1+ pH1- pH2+ pH2- pH3+ pH3- (magenta)	(50) A A pH1+ pH1- pH2+ pH2- pH3+ pH3- (magenta?)	(51) A A pH1- pH1- pH2- pH2- pH3- pH3- (reddish?)	(52) A a pH1- pH1- pH2- pH2- pH3- pH3- (reddish?)

Maybe we can use a cross like this to produce golden yellow elepidotes, maybe not. All we have are hints that it might be possible. However, my real purpose in detailing what might happen with such a cross is to show that a yellow and purple cross isn't such a wild idea, and it can tell us something we don't know about pH in anthoxanthins.

How Many?
Notice that with one gene, there were four genotypes produced in F₂ hybrids (Table 6). With two genes, there are 16 genotypes produced in F₂ hybrids. With three genes, there would be 64 genotypes, etc. We should therefore try to limit our ambitions to dealing with one or two genes! Remember, there is a hypothesis that six different genes control pH in flower tissues. Even with just two alleles per gene, that's a very large number of mathematical combinations. This should not be too discouraging, though. Within the genus or a species, some of those genes will certainly be homozygous, i.e., the same in all members, which is part of what makes them members of the same species. With respect to each gene in individual plants, diploid plants only have two chromosomes carrying the gene, i.e., two possible alleles. To put many genes together in a certain way, it's best to break the effort up into many "sub-projects" of producing homozygous individuals of each gene. They can then be combined pair-wise with F₂ hybrids. With any luck, we'll find something really nice along the way and we can stop there. That's not entirely facetious, because there's really no way to predict what a particular combination of many genes would be like, only a dream.

Side Issues
If we're going to be hybridizing, it might be a good thing to not be too fastidious about deadheading. We've all known plants where if we can waggle our fingers in a spent truss, all the peduncles fall off, and others that do not, and later will be covered with plump seed pods. Just on general principles, whatever our attraction to qualities of flower, plant size, hardiness, habit, etc., we ought to pay some attention to fertility. Genetic processes similar to what we've already explored control fertility. If we don't preserve fertility, we could lose it. It's really disappointing to have a nice seedling we want to use, but find out its barren. The ability to self-pollinate can be very helpful in hybridizing. Some genera have self-sterility genes to enforce cross-pollination.
        I've discovered many rhody people need to know parentage, and disparage open-pollinated plants. But if we're using sexual propagation, we have to relinquish some control. Meiosis, when pollen and ovules are produced, is a randomizing process. The only control we have is in what is there to be randomly assorted, by our selection of parents. That isn't necessarily a bad thing. I know it's often said "Five years of anticipation for five seconds of disappointment." In my opinion, once we make the cross we should be prepared to let Mother Nature show us what she's got. The successful cross is the one that brings us something good, not necessarily what we anticipated. We should evaluate our seedlings, not only against our goals, but on their own merits as well. I've done very well by open-pollinated, but obviously naturally selfed, seedlings.
        There is often a question about whether one parent or another has more or less influence over certain characteristics. When it comes to sexual propagation involving pollen and ovules, and processes in cell nuclei, the answer is no. However, DNA is in another cell structure too. Plant cells have DNA in their chloroplasts, as animal cells have DNA in their mitochondria. Chloroplasts and mitochondria are among the cell's extra-nuclear organelles, but being outside the nucleus do not participate in the whole nuclear fertilization process. Pollen and sperm are not much more than packaged nuclear chromosomes. So these chloroplast genes are transmitted down only the female line of descent. Since chloroplasts and mitochondria deal with energy transformation, their genes tend to do so as well. Generally we're dealing with nuclear genes and fertilization, so it doesn't matter which parent carries the allele.
        In Rogers (2008b), I speculated that the gene for the CHI enzyme might be "switched off" during petal development to produce picotee. The other night it suddenly occurred to me that if that were true, then we ought to see whites picoteed with yellow lips, or with yellow eyes or throats, not just with spots and blotches like 'Belle Heller', but full-on continuous yellow. I can't remember seeing any, nor have any of the friends I asked. I no longer suspect CHI, and that makes some biological and evolutionary sense too.
        With respect to the purpose of this trio of articles, rather than being a scientific exposition, my goal has been to show practical application of the fundamental processes underlying hybridizing by our dabbing pollen on a pistil. Rather than document a few discoveries, as many scientific papers do, I've been trying to demonstrate how it all works, even the genes we haven't discovered yet. It has been a review of 150 years of what has come to be understood as the general processes of genetics as they might be applied to hybridizing rhododendrons, and is far from complete. I've tried to show how observations can lead to reasonable hypotheses. These hypotheses can be used to plan crosses with some expectation of predictable results, and that surprising results can help us refine our hypotheses. Indeed, it is in the surprises that Mother Nature reveals some of her heretofore-unexpected secrets. Fundamentally my articles have been about applying "the Scientific Method," and this method is not at odds with producing more beautiful rhodies. Aware of it or not, this is the sort of thing that's going on. Since it happens anyway, let's use it intelligently!

Acknowledgements
I would like to express my gratitude to Editors Sonja Nelson for the first two parts, Rogers (2008a,b), and to Dr. Glen Jamieson for this article, and to the anonymous reviewers for ensuring I did not stray far from the facts as we know them.

Correction
In Rogers (2008b) I mistakenly stated that all seven of the genes Mendel studies were in different linkage groups. The seven characteristics studied by Mendel are members of four linkage groups. They seemed to be from different groups, as I said, because the linked ones are far enough apart that crossing-over occurs 50% of the time, as if they were independent. Crossing-over is not unusual.

References
Anon. 1955. NBS/ISCC Color Names Dictionary, Centroid Color Charts Supplement. Office of Standard Reference Materials, National Bureau of Standards, Washington DC, 1955.
Bulgin, L.W. 1986. Rhododendron Hybrids, a compendium by parent. Published by the author, 192 pp.
Craig, D.L. 2006. Nothing New Under the Sun. J. Amer. Rhododendron Society 60: 30-32.
Galle, F.C. 1985. Azaleas. Timber Press, Portland, OR: 519 pp.
Mendel, G. 1865. Versuche über Pflanzen-Hy briden (Experiments on Plant Hybridizing). Proc. Nat. Hist. Soc. Brünn. (www.mendelweb.org/Mendel.html)
Rogers, P. 2008a. Why Is It Pink, Daddy? J. Amer. Rhododendron Society 62: 91-96.
Rogers, P. 2008b. But Why Is It Pink, Daddy? J. Amer. Rhododendron Society 62: 188-194.
Rogers, P. 2006. Commentary: Dr. Clement Bowers' Letter. J. Amer. Rhododendron Society 60: 155.

Footnotes
¹ Is there an F cross? No. If we choose to cross the two heterozygotes from the F₂, it's just the F₂ cross all over again. Crossing the two homozygotes is the F₁ cross again. Crossing one of the heterozygotes with a homozygote, or one of the ancestral pure-breeding lines, is generally called a "back-cross" (see Table 6: Backcrossing).
² If one of those were heterozygous, then selfing 'Smokey' should produce seedlings that were homozygous, and an even bluer purple. Are there bluer purples than 'Smokey'? In any event, this demonstration has other goals in mind.

Paul Rogers, BS Chemistry, 1967, California State College - Long Beach, has been seriously raising animals, which must be bred and cannot be "vegitatively propagated," since he was 8 years old, using genetics since a high school student. He made the "mistake" of trying to identify some rhodies at a new home, and came under the influence of Tualatin Valley Chapter members. His main "project" is the production of a fragrant bright yellow.