Variation and Domestication

Karl King (2000)

It is great fun to compare modern theories of heredity with the actual results of plant and animal breeders who practised their craft without benefit (?) of Mendelian theory or the recent development of gene sequencing. Currently, I am enjoying Plant Hybridization Before Mendel by H. F. Roberts.

Roberts discussed a brief memoir on plant breeding published by B. Verlot in 1862.

Verlot: “the more plants are cultivated, the greater their variations are and, by the same token, the easier they are to fix.”

Roberts: “It now seems probable that the increased variation manifested by wild plants, when brought into cultivation, is due to the removal of the restrictive influences of competition, rather than to any actual increase in the heritable variability itself.”

Maybe yes, and maybe no. If we accept Mather’s theory of polygenic inheritance, we must concede that there is some truth in both Verlot’s and Roberts’ positions. Where a trait is influenced by a large set of genes — some dominant, some recessive; some enhancing the trait, others restricting it — it is true that selective pressure and competition will act on the expression of the genes, rather than on the genes themselves. When the plant is taken under cultivation and its development directed by artificial selection, rearrangements of the gene set and selection among the various combinations will lead to greater extremes of variation than would be encountered in the wild. These new combinations are just as hereditary as the originals.

So, it comes down to a matter of definition. Verlot was writing about actual hereditary variation available for selection; Roberts was referring to potential variation inherent in the gene-sets.

Verlot: “As we see, by the sole fact that a plant is cultivated it is forced to vary. The instability of a cultivated plant is even evident in certain cases, in such a way that it does not only manifest itself in the direct descendants of the plant, but also in the plant itself.”
Certainly Verlot is correct in this opinion. Cultivated soil differs from “wild” soil in a variety of ways, not all of them beneficial to the plants. Adaptation to soil type is typically polygenic. Thus a plant which is transplanted to a new type of soil, probably in a different environment than it has previously experienced, will no longer be optimally adapted. Natural Selection alone will produce a directed bias in the “fitness” of the possible offspring, which will continue in subsequent generations until the population is as close to an optimum adaptation as its gene pool allows.

Now consider the consequences. Suppose there are 100-300 genes associated with adaptation to soil type, each with 2 or more alleles. Suppose also that the species in question has 10 pairs of chromosomes. There may be 10-30 genes per chromosome affecting adaptation to soil type. While these genes are sorted and selected towards a better adaptation to local conditions, other genes affecting yet other traits probably will be dragged along by linkage. Also, some of the genes affecting adaptation may have effects on other traits. Therefore, it is nearly inevitable that directed selection for one trait will lead to increased variation in other traits which are only incidentally connected to the selected trait through shared or linked genes. Furthermore, because directed selection leads to a decrease in overall genetic variability, new variations arising under cultivation may be easier to fix than similar variations arising by chance in wild populations — where the gene pool is larger, and the chance occurence of a particular variation is not supported or reinforced by prior selection.

Roberts: “Another example of the older point of view regarding plant improvement is Vilmorin’s opinion, quoted by Verlot, which is here reproduced to show how thoroughly the primary idea concerning 'breaking up of the type' in order to bring about 'variation' entered into the thought and operations of pre-Mendelian breeders.”

Verlot: “To obtain from a plant not yet modified varieties of a kind determined in advance, I will first set myself to making it vary in some direction or other, choosing for the reproducing factor, not that one of the accidental variaties which would most nearly approach the form which I have proposed to myself to obtain, but simply that which would most differ from the type. In the second generation, the same care would make me choose a deviation, the greatest possible at first, the one most different, in a word, from that which I would have chosen in the first place. Following this direction for several generations, there necessarily ought to result, in the products obtained, an extreme tendency to vary; there then results again, and that is the principal point according to me, that the force of atavism, asserting itself counter to divergent influences, will have lost a great part of its power, or, if one ventures to make use of comparison, it will exert it always in a broken line.”

Genes are not perfectly free to segregate, but neither are they completely fixed in place. Selective pressure acting on a population for many generations will favor some arrangements over others. Genes will be temporarily locked into linkage groups. When we alter the conditions in which a plant grows, and begin selecting in some new direction, we cannot expect the genes to jump immediately into the ideal arrangement for our needs. The linkage groups combine genes which enhance and others which suppress the desired trait. Some will be dominant in their action, and others recessive. Verlot’s “atavism” can be found in the linkage of genes, and in the relational balance between paired linkage groups.

As the linkage groups are broken up, the new combinations of genes will produce traits which differ from the wild type. It is not possible to create new genes simply by selective breeding, but there will be an increase in the number of apparent genes as the linkage groups are rearranged. Suppose our initial selection carries two linkage groups: abcde and ABCDE. Because of close linkage the two groups will usually be inherited as units, giving only three types of offspring: abcde/abcde, ABCDE/abcde, and ABCDE/ABCDE. But as we select for anything differing from the “type”, we collect new linkage sets that are produced by crossovers. For example: abcDE, ABcde, ABCDe, and so on. The longer we continue the selection, the greater the range of variation in the linkage groups. Starting with two linkage groups with 5 genes each, we may eventually find 32 different arrangements, which can be combined in 528 ways. There will be no change in the number or type of genes, but there will be a considerable increase the range and type of variation produced by those genes. It is pointless to describe these new combinations as “mutations”, but it appears that some Mendelists, obsessed with genes having large effects, have done so. Many of the so-called mutations were already old when “discovered”, but were usually suppressed and hidden in linkage groups. In fact, the Lamarck/Darwin concept of evolution by imperceptibly gradual change fits polygenic theory rather nicely. The “mutations” described by Devries were also polygenic, mostly involving changes in linkage groups.

Verlot: “We ought then to recognize that it is necessary to take account for the choice of seed-bearers, not only of the external characters, but even of the idiosyncrasy of each of them. Now, since this does not manifest itself except by its effects, we shall, if a variation seems to present some difficulties in becoming fixed, have to examine separately the products of each of the seed parents, and make our choice bear upon those which present, in the least pronounced degree, atavism or the tendency to return to the primitive type.”
Aside from the writing style, this statement could have come from Mather’s discussion of polygenic inheritance. The plants with genotypes Abcde/abCdE and AbCdE/AbCdE may give a similar expression of the selected trait (assuming that capital letters represent dominant genes), but the former will produce at least three types of offspring while the latter is already true-breeding.

A simple case of this situation was described by Dr. Myron Gordon in breeding guppies. One strain of guppy had black tipped scales, another displayed small black blotches. When these were mated, all the offspring were half black. As it happened, the two “mutant” genes were located on homologous chromosomes, fairly close together. By comparing the progeny of the various specimens (accounting for “the idiosyncrasy of each of them”), it was possible to identify those in which crossovers had brought the two genes onto the same chromosome. These specimens became the parents of the true-breeding half blacks.

Roberts criticised Verlot improperly yet again.

Roberts: “Noting the well-known characteristic of augmented vegetative growth in hybrids, he is led to ascribe the frequent seed-sterility to this latter—a conclusion easily if naively arrived at, from the well-known inverse relation between undue vegetative luxuriance and seed reproduction.”
Hybrids of radish and cabbage can be enormously vigorous, but are typically sterile. However, Karpechenko found that the fertility of the hybrids was greatly increased when they were raised in reduced cultural conditions — in perfect accord with Verlot’s opinion, and contradicting Roberts’.

A final matter still needs to be explored. Can pollen from different stamens in a flower give different types of progeny? Verlot cited McNab “to the effect that the best dwarf varieties of Rhododendron are obtained by the use of pollen taken from the small stamens. ‘...the products of which, I am able to certify, are very different from those obtained by the use of the pollen of the large stamens.’” (I take it that Roberts was quoting Verlot quoting McNab.) Similarly, William Herbert insisted that in breeding double Camelias it was best to use pollen from petaloid anthers. Bradley, working with Hippeastrums, found that pollen from the stamen attached to the lower (and narrower) tepal tended to give offspring with narrower tepals; whereas pollen from the stamen associated with the upper (and broader) tepal tended to give more breadth to the tepals of the offspring.

Is there something to this, or were they all deluded? I suspect that cytoplasmic inheritance may be involved, and the ability of a plant to segregate cytoplasmic components into different tissues. It could be a fertile area for further experiments.

CybeRose note, Sept. 15, 2010: I have since learned that it was not McNab who discovered the difference in Rhododendron pollen. Rather, it was discovered independently by two other men—Beaton and Anderson-Henry—at about the same time, and both in Scotland.


Bernard Verlot, 1836-1897