Selective Pollination
Donald Forsha Jones (1928)

Chapter 2


Mendel, in his experiments with peas, worked with seven contrasting characters. It so happens that there are seven chromosomes in Pisum. Apparently, with one possible exception, none of the factors that Mendel worked with showed linkage. Had he worked with many more factors in combination than the famous set of seven, he very likely would have found that segregation was not independent but, on the other hand, that the hereditary determiners were associated with each other in transmission.

In the first detailed investigations on heredity following Mendel—those of Correns, Tschermak, and De Vries, which led to the appreciation of Mendel's work—it is of interest to note that Correns found that in certain crosses of maize he did not obtain the expected proportion of one-fourth recessives. This was reported in 1902 based on the results obtained from a cross of pointed pop corn of the Zea mays everta type with a sweet, wrinkled variety. In a large number of trials this factor pair segregated in the proportions expected for a single Mendelian character, so that in the one case when Correns observed 16 percent recessives the deviation from expectancy was particularly noticeable.

Lock (1906) and East and Hayes (1911) tested the behavior of this endosperm character in many crosses between several different varieties and found independent segregation and recombination to be the rule. In one case, in which the material used was similar to that employed by Correns, a decided deficiency in the number of recessives was noted by East and Hayes.

There are five combinations of genotypes in which a differential action might be shown. The factor symbol for sweet or sugary endosperm is su for the recessive and Su for the dominant type. The five combinations are then:

1. F1 plant by F1 plant Su su x Su su
2. F1 plant by the recessive parent Su su x su su
3. Recessive parent by the F1 plant su su x Su su
4. F1 plant by the dominant parent Su su x Su Su
5. Dominant parent by the F1 plant Su Su x Su su

All of these pollinations have been made (Jones, 1924) with the following results:

1. Eight F1 plants self-pollinated gave 16 percent of recessives in a total of 3,681 seeds, a deviation from the normal 25 percent of 18 times the probable error. The ratios on the individual ears are shown in Table I. The deviation is exactly the same as obtained by Correns.

2. The F1 plants pollinated by the recessive parent showed no significant deviation from equal numbers of the two types of seed—the ratio found being 1,374:1,397, and the deviation less than the probable error.

3. Pollen from the F1 plants, when put on the flowers of the recessive parent, resulted in 885 seeds of the dominant type and 939 of the recessive, which deviates from the expected 1:1 ratio by 27 seeds, which is less than 2 times the probable error, the recessive seeds showig an insignificant excess.

The F1 plants pollinated by the homozygous dominant parent produced only starchy seeds, some of which were homozygous and some heterozygous. These could be distinguished only by growing plants from these seeds and classifying them as "segregating" or "not segregating." The plants were allowed to interpollinate naturally, and all showing any recessive seeds were classed as segregating. Any plants that resulted from Su Su fertilization would show no recessive seeds; while most of the plants of the combination su x Su would show some recessive seeds. Without selective fertilization one-half of the plants would be segregating so that one-fourth of the pollen in the field would be carrying the recessive factor. Segregating plants would be expected to have half of their ovules with the recessive gene, so that the segregating ears should have on the average one-eighths of their seeds of the wrinkled, sweet type. Differential fertilization, working against the recessive gametes, would reduce this proportion. Some segregating plants might have no recessive seeds as extreme deviations in a chance assortment, and for that reason would be wrongly classified. Seven percent of the plants with some sweet seeds had from 1 to 10 recessives. The percentage of segregating plants that had none would be less than this; and in view of the marked differences obtained, this source of error could not change the results appreciably. Moreover, the error would be the same in either type of mating.

4. From the backcross of pollen from the homozygous dominant parent on the heterozygous F1 plants, 207 segregating and 213 non-segregating individuals were obtained. These numbers differ from a 1:1 ratio less than the probable error.

5. The pollen from heterozygous F1 plants which were backcrossed on to the homozygous dominant parent gave 88 segregating and 353 non-segregating individuals, a deviation from a 1:1 ratio close to 19 times the probable error.

It will, therefore, be seen that no differential fertilization occurred in those matings in which the male gametes were all alike. Also, no difference in fertilizing ability was shown when the segregating pollen was backcrossed on to the recessive parent. A selective action took place only when the segregating pollen was used with F1 plants or the dominant parental type, both of which carried the normal allelomorph either in the homozygous or heterozygous condtion. The significance of this situation will be discussed later.

Another endosperm character in maise, known as "waxy," behaves in somewhat the same way as sweet endosperm. A deficiency of recessive seeds in the F2 endosperm generation from crosses with normal starchy were noted by Collins and Kempton (1911) and later studied extensively by Kempton (1919), Brink and MacGillivray (1924), and Kiesselbach and Petersen (1926). Waxy endosperm is of particular interest in that the carbohydrate reserve is of a different chemical nature than that of other varieties of maize; it stains red with iodine, while other types of endosperm stain blue (Weatherwax, 1922). This difference in staining is also visible in the pollen grains and in the embryo sac.

Fig. 1.—Pistillate flowers of Melandrium, showing stigmas and arrangement of seeds in the ovary. (From Correns in Hereditas.)

All of the available data on the inheritance of waxy endosperm have been summarized by Kiesselbach and Petersen (1926), to which they have added considerable new data of their own. All the results combined show a deficiency of 1.1 percent from the expected 25 percent when the heterozygous plants are self-fertilized, and 0.7 percent from the expected 50 percent when segregating pollen from the heterozygous plants is applied to the recessive type. The deviations in the two instances are 15 and 5 times their probable errors and, although small, are undoubtedly significant. Unlike the results with sweet endosperm, a deviation is obtained in the backcross of the F1 plants on to the recessive parent. Similar results have been obtained with another endosperm factor in maize. These will be considered with the discussion of pollen-tube factors. The dioecious red-flowered Campion, Melandrium rubrum Garcke (Lychnis dioica L.), commonly produces more pistillate than staminate plants. When a large amount of pollen was applied (from 25,000 to 50,000 grains), Correns (1921) counted 68 percent pistillate plants. This species produces from 300 to 400 seeds in a capsule. When no more pollen grains are applied than the number of seeds usually contained in a capsule, the proportion of pistillate plants was reduced to 56 percent. Over 2,000 plants were grown in each lot, so that the difference of 12 percent is undoubtedly significant.

The sex in Melandrium is determined by the pollen, and Correns considers the difference in the two classes to be due to a faster rate of growth of the pistillate-determining pollen tubes, as indicated in the foregoing results and also in the following experiment. The seeds of this plant are borne on a common placenta in which the pollen tubes grow. The first tubes to reach the ovary fertilize the ovules in the upper part of the seed capsule; while the later tubes have to pass on down and fertilize those in the lower part. This was proved to be the case by pollinating a white-flowered plant with pollen from a red-flowering sort. Later, pollen from a white plant was applied. The upper third of the capsule yielded 45 percent, while the lower two-thirds gave 7 percent, red-flowered plants. The greater proportion of white flowers in both lots indicates a selective action in favor of the plants' own kind of pollen, although this is not the question at issue. The difference in position of the seeds shows that the first pollen tubes to enter the ovary tend to fertilize the ovules in the upper part. For that reason the difference which Correns obtains of 67 percent pistillate plants from the upper part and 55 percent from the lower part indicates also that the female-determining pollen grains produce faster-growing pollen tubes. More pistillate plants were also obtained by pollinating at the tip than at the base of the pistils. Apparently the rate of growth of the pollen tubes is not the only factor involved, since an excess of pistillate plants is obtained even when every pollen grain has a chance to function. A selective elimination, sufficient to account for this difference, may occur after fertilization.

The age of the pollen profoundly affects the proportion of pistillate plants. Starting with approximately equal numbers of the two types obtained from fresh pollen, the percentage of pistillate plants regularly decreased to zero as the pollen was aged up to 100 days before being applied, the staminate-determining grains being more viable (Figs. 2 and 3).

Fig. 2.—Curve showing the reduction in seeds produced with the age of pollen; the line below represents well-developed seeds only, the line above, all seeds. (From Correns in Hereditas.) Fig. 3.—Curve based upon data from three different plants, showing the reduction in percentge of female plants with the age of pollen. (From Correns in Hereditas.)

A similar situation exists in the dioecious garden sorrel, Rumex acetosa L. Correns (1922) finds a greater proportion of pistillate plants. The staminate plants are heterogametic; and the difference, as in Melandrium, is apparently due to a faster growth of the pistillate-determining pollen tubes.

In Oenothera Lamarckiana Heribert-Lilsson (1920, 1923) finds an excess of red-veined plants when the pollen of a heterozygous red plant is backcrossed on to an uncolored plant. Homozygous red-veined plants do not exist on acount of this combination being lethal. The unequal fertilization by the two types of male gametes from the same individual is attributed to differences in pollen-tube growth. Renner (1919, 1921) also explains deviations from expected ratios by the assumption of unequal growth of pollen tubes carrying different factors.

By cutting off the styles at different times after pollination, Davis (1926) found that in a cross of short- and long-styled Oenotheras there was 1 short- to 5 long-styled segregates 23 hours after pollination, and 1 short- to 2.8 long-styled segregates 24 hours after pollination, indicating a slower growth of gametes carrying the short style.

In Datura one of the 2n+1 mutants, known as Globe, has been found by Buchholz and Blakeslee (1922) to differ in the rate of pollen-tube growth. Those pollen grains, which contain the extra chromosome, grow at a slightly slower rate than those grains which have the normal allotment of chromosomes. Half of the pollen grains contain the extra chromosome, as shown by cytological examination; but less than 3 percent of the offspring receive an extra chromosome from the pollen.

Stigmas of Globe plants self-pollinated were prepared and stained in such a way as to permit the measurement of the lengths of the pollen tubes after about 14 hours' growth under comparable conditions. Their results are shown in Figure 4, together with counts from normal plants self-pollinated. The Globe pollen tubes are clearly bi-modal in length. The pollen from both types of plants germinate equally well, approximately 95 percent in each case; but part of the Globe pollen tubes clearly form a slow-growing group in the same stigmas in which the other part grows at the same rate as in normal plants. Sirks (1926) has also obtained deviating ratios in Datura. By cutting off the styles at different times after pollination, he obtained varying percentages of plants having unlike factors for color and spines.

Fig. 4.—Differential pollen-tube growth in Datura. The area designated by horizontal lines is the distribution with respect to to length of the pollen tubes of the Globe type: the vertical-lined area is the distribution of pollen tubes of normal Datura. (From Buchholz and Blakeslee in Science.)

A dwarf type of barley which is described by Hallqvist (1923) segregates in a deficient ratio giving about 18 percent recessive plants. When the heterozygous plants are pollinated by normal plants, a good 1:1 ratio is obtained of segregating and non-segregating individuals. In the reciprocal combination of pollen from heterozygous plants on homozygous normal, 11 segregating and 35 non-segregating individuals were obtained. The marked difference in the reciprocal crosses indicates that the results are not due to an elimination of certain types but, as in maize, are more easily attributed to a reduction in the proportion of gametic combinations. In this case the abnormal plants produced no flowers, so the combinations with the recessive parent could not be made.

In all the cases of preferential fertilization just considered in Zea, Melandrium, Datura, Oenothera, and Hordeum, differences in fertilizing ability have been shown only by the pollen. There is at least one clear example of a difference shown by the ovules as reported by De Vries (1924) in the evening primrose. Oenothera lucida produces two kinds of gametes, one of which always carries a lethal. When pollinated by four different homozygous varieties, more than 3 times as many ovules of one type were fertilied than the ovules of the other. The female gametes possessing the lethal were the ones that were avoided. This lethal factor did not cause an elimination as, in the hybrid progeny, both combinations were viable; in one the lethal was not present, in the other it was in the heterozygous condition. The results are given in Table II.

Gamolysis of Female Gametes of Oenothera lucida
(Data from De Vries in Botanical Gazette)

Oenothera lucida
Pollinated By—
No. of Individuals Resulting From—
Lucida Ovules Deserens Ovules
Albida    5 Lamarckiana   95 rubrinervis
Elongata   14 empty seeds   86 lucida
Blandina   33 Lamarckiana   67 subrobusta
Deserens   42 lucida   58 deserens
Average    22 78

Pollen of Oenothera lucida was put on the flowers of other varieties. One lot with short styles showed no appreciable difference in the number of fertilizations by the two types of pollen. When put on three different varieties with long styles, the same type of gamete that was fertilized in greater numbers here also fertilized more than 3 times as many ovules as did the other gametes which carried the lethal (Table III). De Vries considers the selective fertilization to be due in some way to the presence of the lethal factor. Also in croses of Oenothera blandina by O. Lamarckiana the self-fertilized F2 progenies showed an excess of one combination above expectation, owing to a differential influence of some sort. Renner (1921) has found that there is a competition between the megaspores in certain types of Oenothera not due to egg lethals apparently but to an ability of some female gametes to grow at the expense of the others having a different germinal constitution.

The transmission of heritable characters, instead of following the rule of Mendel in being a chance recombination of independent units of inheritance, is dependent upon many things: the position of the genes in the chromosomes, the usual and unusual assortment of chromosomes, the elimination of gametes due to lethal factors, and the elimination of zygotes. Evidence has been given to show that there are still other influences which bring about the union of particular gametes in greater numbers than are expected on the basis of independent segregation and chance recombination. In the following chapter it will be shown that there are differences in the rate of pollen-tube growth because of factors located in the chromosomes which control the development of the male gametophyte and that these factors are responsible for most of the deviations from expected ratios.