1980 American Rose Annual
Inheritance of Pigments
D. P. de Vries
Lidwien A. M. Dubois
Institute for Horticultural Plant Breeding
Wageningen, The Netherlands
|Königin der Rosen (Colour Wonder)|
|Super Star (Tropicana)|
As we have shown in Part I in this annual, the colours of rose petals are based on the combined action of two through five separate pigments.
Attempts to explain inheritance of rose flower colour in terms of Mendelian inheritance, i.e., a system based on dominance and recessivity of certain colours, have been made by some authors, but were not quite satisfactory.
Considering pigment inheritance in other crops, it seemed unlikely that in roses colour, rather than the constituting pigments, would be inherited. Particularly for the breeder, it would be useful to have an insight into the genetical basis of flower colour.
To elucidate this problem, in a five year programme, we have made a large number of crosses between 18 varieties of known pigment constitution. Subsequently, by means of paper chromatography, we have determined the relative quantities of the flavonoids: pelargonidin, cyanidin, quercetin and kaempferol in several thousands of their seedlings.
In several species. e.g. Mirabilis, flower colour depends on monogenic differences. When there is one dominant gene for red and one recessive for white, there are two flower colours: red and white.
This system is known as Mendelian — major gene — or qualitative inheritance. In general, in qualitative inheritance there are clear cut differences between the individuals of seedling populations, which can simply be counted.
In other species, flower colour appears to be determined by many genes, each with a small effect.
Additive gene action is also known as action of minor genes, polygenes or multigenes. Their effects, which are of degree, cannot be counted but should be measured.
Because in rose seedling populations the pigment contents could not be classed into clear cut groups, additive gene action was supposed. With the help of our biometrical department in using a statistical method called combining ability analysis of variance, it was possible to show that this supposition was correct.
To the breeder, additive gene action for a certain character offers the possibility, after crossing of two parents, to approximate the seedling population mean for this character. This is done by assuming that the population mean value agrees with the mid-parent value, i.e., the mean of the two parents used.
When, e.g., the cyanidin contents of two parental varieties are 2.8 and 3.6, the mid-parent value and approximately the mean value of the seedling population are (2.8 + 3.6) : 2 equals 3.2. However, the larger the difference between the parents, the larger will be the range of pigment contents in the seedling population.
It should be kept in mind that in case of flower colour, as a composite of several pigments, these population means apply to each separate constituting pigment.
In the Table, the 18 varieties used in our experiment were arranged according to the decreasing pelargonidin content. It can be seen from this arrangement there is apparently no relation with or between the other pigments.
The following is intended to illustrate how a breeder could use the Table in breeding seedling populations with, e.g., mainly orange flower colours. From our previous article, we know the pigment means for the orange colour group are: 3.8, 2.6, 3.6, 1.4 for pelargonidin, cyanidin, kaempferol and quercetin, respectively.
Next, two parental varieties whose pigment mid-parent values approximate these means should be selected. It appears that Korp and Numéro Un, with mid-parent values of 3.9, 2.4, 4.3 and 1.9 for our four pigments, make a good choice. Other varieties, not listed here, may give the same results.
When the same mid-parent values are built up from parental contents that are much wider apart, a correspondingly wider range of seedling colours may be expected. This is particularly so in the case of large differences in anthocyanidin content.
When Amica and Sonia are mated, which have about the same mid-parent value (3.7, 1.7, 3.7, 1.7) as Korp and Numéro Un, much more medium pink, deep pink and even orange-red colours are expected.
It is a misconception to speak of the inheritance of flower colour unless one refers to the individual pigments, which inherit quantitatively.
The breeder chemical analysis of flower colour appears to be imperative for a correct choice of progenitors. Once the pigment constitution of a number of varieties or of one's own selections have been determined, a straightforward breeding strategy for pigment content and hence for flower colour can be adopted.
Yellow carotenoids were not involved in our trials, but for several reasons it may be assumed these pigments too are quantitatively determined. This means that in determining mean population colours, five pigments should be taken into account.
In our present study, we have emphasized flower colours only. However, in rose breeding many other, most probably also quantitative determined characters such as vigour, stem length, thorniness, flower production, shape, size, petal number, vase life and fragrance, should be considered also.
Each of these characters should and can be treated in similar ways as described for pigments. Although in this way chances of finding the best plants enormously increase, combining the favourable part of the variation of each character in one plant, seems rather a matter of luck than of statistics.
The pigment constitution of varieties used in a five year breeding experiment
|Duke of Windsor||4.2||1.5||3.5||1.5|
|Königin der Rosen||2.6||1.3||1.0||1.0|