Apples and Oranges

Twenty years ago or so, I was talking with Tom Liggett after he returned from a visit with Ralph Moore. Tom held up a bloom of ‘Grey Pearl’ and told me that Ralph told him the color is due to a double recessive. That didn’t make any particular sense to me, so I let it slide.

Recently I uploaded an article on moss and minis that Moore chose to publish in the anti-science Creation Research Quarterly. Among other things, he wanted to share this “theory”.

So called “blue roses” are merely those in which the magenta has the ascendancy. Lavender color apparently is produced by the combination of magenta and yellow. Some will doubt this, but about 15 years ago in a conversation, I told Dr. Nisbet, then editor for the American Rose Society, that I believed such to be the case. A check of the pedigree of so called lavender roses will show that this is true.

That’s not true, by the way, but his belief still puzzled me.

Then I went back to re-proof and re-edit some of Lammerts’ articles. There it is, in the References:

Mehlquist, G. A. L. “Inheritance in the carnation, Dianthus caryophyllus” 1. Inheritance of flower color, Proc. Amer. Cos. Hort Sci. 37: 1019-1021, 1940

Lammerts assumed that the inheritance of flower color in roses works the same as it does in carnations. It does not. In carnations, the ivory and yellow shades involve various fancy flavonols and flavones that are not found in roses. The yellow of roses is carotene, to the contrary, located in the plastids rather than mixed in the vacuole with the water soluble pigments.

That’s confusing Apples and Oranges.

I first became disenchanted with Lammerts’ articles when I found that the diploid roses exhibit the same degrees of doubleness that Lammerts tried to explain by different doses of a “doubling gene”. Tetraploids might have as many as four copies of such a hypothetical gene. But if the same explanation cannot apply to diploids, which were parents of the tetraploids, then the his speculative model must be wrong.

The second clue came while I was scanning and proofing his articles years ago. In 1960 he claimed that the genotype of ‘Chrysler Imperial’ is MMMm. But in 1964 assigned MMmm to the same rose. This makes it pretty clear that he reverse engineered the (alleged) genotypes in order to “predict” the results he had already obtained in breeding.

The quadrivalent inheritance he described is rather dubious. I have read that Dahlia variabilis manages such a regular quadrivalent meiosis, but that species is rather peculiar. In roses, associations of four chromosomes during meiosis usually reduce fertility.

In conclusion, Lammerts concocted a very doubtful reproductive mechanism, and assigned “genotypes” to the parents of his crosses, in order to reconcile color inheritance in roses with the very different system found in carnations.

Le Grice (1968) also had some rather confusing views on the basis for the “blue” roses. He stated that there were four strains:

  1. Lavender Pinocchio “… a complete admixture of colour with a considerable bias towards bicolours which derive originally from R. foetida persiana as employed by Pernet Ducher.”
  2. Meilland “… strong admixture of colours with bicolours and R. foetida persiana playing an important part.”
  3. Peace “same pattern of parents bring the same results.”

Given that ‘Peace’ was bred from the Meilland group, I don’t see why it rates its own “strain”. And then, Le Grice gave ‘Peace’ credit for the color of ‘Sterling Silver’, ‘Blue Moon’, ‘Prelude’ and others that are obviously derived from ‘Grey Pearl’.

The fourth strain? Le Grice didn’t mention it by number, so can only guess that he meant, “In my own case a cross between Tantau’s Surprise and Marjorie Le Grice resulted in a mauve seedling the ancestry giving the same results.”

“From the foregoing paper it will appear that mauves, lilac and brown colours appear only when a bicolour with red inner petal and golden reverse is used at least once in the crossing.”

I do not doubt that Rosa foetida contributed to this color range. The yellow pigment is not relevant. Nor is the “bicolor” trait. These are only markers that suggest descent from R. foetida.

And we should not confuse the issues. There are at least three ways to achieve colors in the mauve/lilac/purple spectrum: co-pigmentation, AVIs, and rosacyanins. Two or more of these may be combined in one cultivar, but in some cases the first cross may counteract two mechanisms. As Le Grice wrote, “Purple crosses largely give reds.” This would be likely if one parent relied on AVIs while the other had co-pigments.

Le Grice stated confidently, “Even when some unlikely stray appears, a little detective work brings the same results and in Lilac Time (Golden Dawn x Luis Brinas) we find the strong admixture with white and the R. foetida bicolor background.”

I offer seven objections.

  1. Erinnerung an Brod
  2. L’Evêque
  3. Veilchenblau
  4. Deuil de Paul Fontaine
  5. William Lobb
  6. Superb Tuscan
  7. Reine des Violettes

To be fair, those are dark purples compared to the lavenders that appear to be the topic by Le Grice. Not that the mechanisms don’t have an overlap, just that it’s understandable how he’d ignore them given the very different visible colour.

Le Grice began by quoting Harborne’s example of ‘Reine des Violettes’, then ended with “From the foregoing paper it will appear that mauves, lilac and brown colours appear only when a bicolour with red inner petal and golden reverse is used at least once in the crossing.”

Though his statement may be true, it is not at all helpful to anyone interested in breeding roses of these colors. Just the following year (1969), Harry Murray wrote in The Rose Annual, “Without ‘Soleil d’Or’ the world would not have seen any of the deep yellow, orange or flame-coloured roses that have done so much in recent years to delight the eye and uplift the heart. Over twenty years ago there were no less than 513 varieties which were traced as direct descendants of this magnificent rose and today it is estimated that at least seventy-five per cent of all modern roses carry ‘Soleil D’Or’ in their ancestry.”

Three out of four “modern” (circa 1968/9) were descended from a red/yellow bicolor. I am pretty well convinced that the mauve roses, colored by rosacyanins, are derived from Rosa foetida, though I still don’t know what exactly that species contributed.

We might learn more if ‘Charles P. Kilham’ and ‘Capucine Chambard’ were still around.

Has anyone recently raised a “blue” rose from ‘Peace’ or "Mrs Sam McGredy’ without some other mauve parentage?

I guess I shouldn’t be surprised that the quaint musings on pigmentation of breeders like Moore, LeGrice, Lammerts, Marshall and others would continue to be seriously discussed in the century after they lived and died. The actual science of the matter is largely inaccessible to the layman even today by being hidden behind paywalls and by the sheer complexity of the biochemistry involved to say nothing of the molecular biology (and jargon - AVI’s for instance are anthocyanin vacuolar inclusions).

While I don’t disagree with Karl on the (at least) three ways to achieve colors in the mauve/lilac/purple spectrum it reminded me of a paper that describes those mechanisms in detail and using a slightly different terminology: “Three major mechanisms for blue flower coloration exist: self-association, co-pigmentation, and metal complexation”. See section 3 p 887 of:

"Blue flower color development by anthocyanins: from chemical structure to cell physiology in Kumi Yoshida, Mihoko Mori and Tadao Kondo, Nat. Prod. Rep., 2009, 26, 884–915.

A primer on the more-or-less current understanding flower pigments is “Biosynthesis of plant pigments: anthocyanins, betalains and
carotenoids”, The Plant Journal, (2008), 54, 733–749. See citation

Contrast these papers with the relative simplicity of the 1973 paper “Breeding Research on Rose Pigments” by DeVries, Euphytica 23 (1974) 447-457.
Breeding Research on Rose Pigments - (502 KB)
Blue flower color development by anthocyanins - from chemical structure to cell (2.16 MB)

I read about these mechanisms years ago, and soon learned of cases where all three mechanisms work together, as in the delightfully blue Dayflower, Commelina communis.

Rosacyanins are a different matter, since they don’t fit any of the above three. They are found elsewhere, as in wines, but I still can’t learn what is in Rosa foetida that might account for the production of pigments not found in that species.

It is still useful to check back on old research, now and again, to see what might have been learned in the interim. There was a time when some researchers insisted that the yellow-flowered dwarf bearded irises were not colored by flavones. The evidence was that such substances could not be extracted. Turns out that the flavones were bound to the protein matrices of the cell. This association kicked up the color several notches.

I thought this association of flavones with proteins might have something to do with AVIs, but apparently not.

My web page is in the process of being migrated (so I’m told), so I’m going back to see whether some of the papers I scanned and uploaded 20+ years are still worth keeping. On the one hand, there is historical interest in what the old guys did and how they thought they were doing it. On the other hand, some are not helpful because no sensible person alive today should start crossing among dozens of red/yellow bicolors in hopes of getting a mauve.

The number of available cultivars are quickly diminishing such that I doubt it would even be possible to assemble a stable having dozens of red/yellow bicolors.

One has to wonder how many breeders of the early Pernetianas found mauve seedlings among the more brightly colored type, and simply discarded them.

However, it may be possible to infer some useful facts scattered in the old comments.

in a newspaper interview, Gladys Fisher commented that her ‘Morning Mist’ was in the paternal lineage of ‘Sterling Silver’. The patent record gives the parentage of ‘Morning Mist’ as a self-seedling of an unregistered, unnamed and undescribed seedling of Fisher’s breeding. I guess (danger alert) that she would have mentioned if the parent had been of the same color. This implies that the “unidentified factor” is recessive.

Le Grice described ‘Fantastique’ as “brownish yellow, flushed carmine”. The one I saw as the San Jose Heritage was about the same color as ‘Peace’, but in a very different form. Perhaps the expression of the “unidentified factor” is influenced by temperature.

Le Grice also wrote, “Browns intercrossed with salmons give wide variations in red, pink, yellow and white.” If we consider that brown/tan is mauve over yellow, we have further evidence that the “unidentified factor” is recessive.

One test might be to outcross a mauve variety to a white that is not descended from the pernetianas. Does the color “break down” and give an assortment of pinks and reds?

Another: Peace x mauve. Are there about as many mauve seedlings as non-mauve? One to three?

A quick check of the F1 offspring of ‘Sterling Silver’ revealed, i’m not surprised, that most are mauves. Naturally, most people breeding with SS are aiming for similar colors, even when the other parent seems unlikely to give the desired results.

Orangeade x Sterling Silver - Blue Danube
Orangeade x Sterling Silver - Flighty

Then there are some non-mauves that turn up:

Baccara x Sterling Silver - Glen Artney (dark red)
Sterling Silver x Honor - Pearl Essence (light pink)
Tropicana x Sterling Silver - Saul (deep pink)

And finally at least one that doesn’t seem to fit. Maybe I have overlooked some other variable.

Sterling Silver x Prélude - Russkij Suvenir (White, light pink center)

I think that Morning Mist is a cross between Lavender Pinocchio and some one of the work horse florist roses from the 60’s like Tiffany. In fact I think Tiffany is a good candidate based on scent and morphology.

I wonder if anyone at Meilland or Suntory has figured it out yet based on genetics. Should be easy to do.

‘Morning Mist’ was patented in 1950, ‘Tiffany’ in 1954. ‘Lavender Pinocchio’ is also too late. That seems to leave ‘Grey Pearl’, 1945, which Fisher could have crossed with one of her own seedlings. The best seedling from the cross was then selfed to get ‘Morning Mist’.

It might have turned up from her own breeding efforts, although ‘Rapture’, ‘Better Times’ and ‘Pink Delight’, mentioned in a newspaper article as being in the ancestry of ‘Morning Mist’, do not seem likely sources for the color. That leaves ‘Grey Pearl’, I think, as the most probable “missing” ancestor.

Last evening I was walking my dog and got to wondering how it happens that ‘Paradise’ and ‘Angel Face’ (among others) manage to have the red border that expands with exposure to light. Why doesn’t that pigment get tied up in rosacyanins?

I have searched for info on the biosynthesis of rosacyanins, but so far haven’t found what I need. But while strolling with Mickey, I wondered whether the synthesis starts by knocking off the two glucose molecules on 3,5-glucoside of cyanidin. The “chameleon gene” that adds red over the ground color after the flower opens and is exposed to light, produces the 3-glucoside only. No rosacyanins from that source.

This seems to explain some of the “mysteries” of mauve inheritance, though that does not mean that it’s the correct model. ‘Peace’ and ‘Fantastique’ have little or no 3,5-diglucoside pigment, only the 3-monoglucoside that appears after opening. Thus, they might carry the gene to produce rosacyanins, but lack the necessary precursor.

Stoddard (1980) reported a seedling from Orangeade x Angel Face that “…was soft gray with an orange edging, unique in my experience”. This could have been a pelargonidin-based rosacyanin, with the “chameleon gene” finishing off the bloom with an orange edge.

PS: the Sterling Silver x Prélude - Russkij Suvenir (White, light pink center) I mentioned earlier could fit my hypothetical model if the pink center of ‘Russkij Suvenir’ is colored by chrysanthemin.

Whomever still grows Orangeade and Angel Face should make a ton of that “gray with orange edge” cross to see if something like it might result again.

I want pictures!

Me, too! And, none of those grainy, black and white images printed in newspapers and earlier books!

Not the same but similar lines, I did Gemini x Blue Moon…just deep pinks/cool toned reds and a lot of things with bad roots that died…that probably isnt helpful info.

All info is helpful.

I assume your ‘Gemini’ is the orange Polyantha rather than the pink HT. If so, do you have a picture to upload to HelpMeFind? The picture on the patent is useless, and I could not find a better one.

If any of your seedlings seem worth keeping, they should carry the relevant factors for both orange and blue, these traits being (mostly) recessive.

I had hoped to cross ‘Margo Koster’ with ‘Violet Mist’, but that plan has been put on hold. Among a couple of other things, I wanted to get the rosacyanin trait into some diploid lines, if possible.

I should mention that I’ve had time to rethink my suggestion that the 3,5-glucoside of cyanidin is a precursor of the rosacyanins. It really doesn’t make sense.

It is possible, and maybe more plausible, that the enzyme responsible for connecting gallic acid to the cyanidin skeleton is active only before the flowers open. The “chameleon gene” is active only after anthesis, so the pigment it spreads across the petal would not be affected by the “rosacyanin gene”.

No, I meant the Zary HT, it being a descendant of orangeade (a couple generations back)…part of that probably not useful info disclaimer.

Pulling rosacyanins down to triploid then diploid is something I’ve been attempting. Unsure of success/failure. None of the seedlings showed any obvious sign of those tones, maybe in the f2. Possibly need to try with better pigmented diploids than what I currently have and with better seed production.

I do have this multiflora nana hybrid (came from the angels wings seed strain) that starts deep pink and fades/blues to lavender. Likely a different mechanism but may help down the line.

My first cross (Sweet Chariot x Margo Koster) was somewhat similar. I had a wacky hope that I might get something purple with orange undertones, or orange with purple-bronze shadings. Nope! What I got was a charming plant with flowers that opened rose pink then changed to lilac pink by the second day. I wish I still had it.

‘Ernie’ (Blue Nile x Blue Mist) was (is?) a triploid step in the right direction. Some of the pictures show uniformly mauve blooms, like small ‘Sterling Silver’. Other pics show buds that are as irregularly colored as ‘Blue Mist’. So, the “Mauve Trait” seems to be not exactly dominant, but not entirely recessive.

There appears to be some overlap in the materials involved in making roses mauve/purple. According to Harborne (1976), the purple of ‘Reine des Violettes’ is due to the ordinary cyanidin 3,5-diglucoside co-pigmented with large amounts of gallotannin. Mauve roses, to the contrary, have the gallic acid bonded to the cyanidin (no glucose) rather than polymerized into gallotannin. And then, the Rosacyanin B is “decorated” with one or another ellagitannin (polymerized ellagic acid).

I uploaded a diagram of Rosacyanin B next to cyanidin and gallic acid to show a bit of what is happening. Can a single enzyme connect cyanidin and gallic acid?

Hi Karl, did you count Ernie’s ploidy? It isn’t listed on HMF as triploid and that should be something we’d like to add if it is known to be. Thank you.