PNAS 6(2): 66-70 (February 1, 1920)

Communicated by T. B. Osborne, Read before the Academy, November 10, 1919

The fact of self-sterility or self-impotency is well established in many species of plants and in at least one family of animals. It is also believed that even in the normal condition of self-fertility the germ cells from different, unrelated individuals in some cases are more efficient in accomplishing fertilization than the germ cells from the same or similar individuals. Statements to this effect are made in text books on biology.

A simple method of accurately testing this assumption is available in the use of pollen mixtures which carry such inherited characters that the different kinds of seeds resulting from the pollinations can be distinguished. By taking pollen from two distinct types of maize, designated A and B, in approximately equal quantities, thoroughly mixing, and applying the mixture to the plants which furnish the pollen, it is possible to obtain and to separate the two kinds of seeds on each plant. On the A plants the seeds are A × A self-fertilized and A × B cross-fertilized. Similarly, on the B plants the seeds resulting from the mixture are B × A cross-fertilized, and B × B self-fertilized. The numbers of individuals in each class form a proportion such that the A × A seeds are to the A × B seeds (produced on A plants) as the B × A seeds are to the B × B seeds (produced on B plants). The end terms of this proportion represent selffertilized, while the middle terms represent cross-fertilized seeds. If fertilization takes place at random the numbers form a perfect proportion, irrespective of the relative amounts of functional pollen present in the mixture and independent of the total number of seeds obtained from either the A or the B plants. If there is a deviation from a perfect proportion, this will be either in favor of cross-fertilization or self-fertilization.


X2 P
A × A A × B B × A B × B
1 3430   1738   46   1602   44   +0.045   0.027 0.994
2 5636   2133   145   3080   278   +0.955   7.572 0.063
3 1362   229   14   70   349   +12.715   146.196 0.000
4 3344   710   126   1856   652   +5.465   61.203 0.000
5 1956   589   6   1290   71   +2.105   28.718 0.000
6 424   40   71   187   126   -11.850   23.858 0.000
7 3459   23   89   1507   1840   -12.245   235.357 0.000
8 7783   2185   956   2619   2023   +6.570   144.108 0.000
9 8729   2550   1288   2922   1969   +3.350   42.061 0.000
10 5408   1084   1154   997   2173   +8.495   162.687 0.000
11 3314   448   264   1505   1097   +2.540   8.937 0.030
12 3561   1724   719   749   360   +1.790   5.313 0.053
13 736   185   391   95   6.5   -13.625   55.051 0.000
14 792   424   150   156   62   +1.155   0.533 0.894
15 3168   2609   47   14   498   +47.750   2889.561 0.000
16 2224   723   8   74   1419   +46.975   1965.981 0.000
17 1410   1303   3   4   100   +47.960   1298.993 0.000
18 1599   4   21   1   1573   +7.970   137.532 0.000
19 2606   528   392   343   1343   +18.525   376.394 0.000
20 2753   897   77   1174   .605   +13.050   283.002 0.000

One great advantage of using endosperm characters in this way is that the several kinds of seed develop under nearly as favorable environmental conditions as it is possible to obtain, so that selective elimination of zygotes, which may favor one class more than another, is reduced to a minimum. In those plants in which the cross-fertilized and self-fertilized progeny cannot be distinguished until they are grown to maturity there is considerable likelihood that unequal germination and differential viability will tend to favor cross-fertilization. Animals are subject to the same objection as material for investigating this problem.

The method of mixed pollination in reciprocal applications is the simplest means of accurately testing the efficiency of two kinds of pollen in competition, with each other. It is impossible to know definitely the relative numbers of the two kinds of fertilizing elements present in a mixture.

Even if it were possible to obtain equivalent amounts of pollen or seminal fluids there would be no way of knowing the ratio of the two kinds of effective germ cells, since in many species such cells rapidly lose their viability.

In many experiments performed with maize, pollen was collected from a number of plants of two distinct but uniform types of plants. Approximately equal quantities of each kind of pollen were taken, thoroughly mixed by shaking in a bag, and applied to a number of plants of each of the two kinds supplying the pollen. The aim was to have from 1000 to 2000 seeds in each of the two parts of the proportion. Twenty such mixed pollinations were made, using many different types of maize, and altogether there resulted 63,694 seeds. These have been classified, counted, and a sample of each lot grown to test the accuracy of separation, and the proportions have been calculated as percentages. See table 1. Seventeen of the twenty show a deviation in favor of the plant's own kind of pollen, while only three show the reverse effect. Of the 17 experiments which indicate a prepotency of pollen on the stigmas of the plants by which it was produced, 2 are not significant when compared with the deviations expected in sampling. In the remaining 15 the deviations are so large that there can be no question that in maize there is a pronounced preference of the plant for its own kind of pollen, even though the foreign pollen is perfectly capable of accomplishing fertilization when not in competition.

Similar experiments have been carried out with another plant, the tomato. Two pollen mixtures were made, utilizing characters by which the seedlings could be separated. As with maize, the results give a deviation favoring the familiar pollen. However, the number of plants obtained was not large and classification was not so sure as in the experiments with maize.

Taking the results from maize alone, magnifying the experimental error to its fullest extent, giving due allowance to differential viability, and taking into consideration the differences arising from random sampling, the conclusion is inevasible that in this species at least there is a definite receptiveness of the plant to its own kind of pollen. This is notable in view of the great advantages which hybrid vigor gives immediately to the cross-fertilized seeds and the plants grown from them. The weight of cross-pollinated seed is increased as much as 50% in some cases. Both the embryo and endosperm are larger, the seeds have a higher specific gravity, and they mature faster, as shown by theft lower water content at the end of the growing period. The increase in weight, expressed as per cent, permits a comparative estimation of the amount of heterosis shown by the various combinations involving different materials. There is a significant correlation between the amount of heterosis and the preference shown by the plant for its own pollen. Since heterosis is roughly proportional to germinal diversity, the greater the genetic differences there are the greater is the handicap placed upon the pollen when competing with pollen from the plants on which it is to act. This is a remarkable result, because in proportion as the cross-fertilization benefits the progeny the less effective are the germ cells in accomplishing fertilization.

It should be clearly kept in mind that the selective action shown in pollen mixtures is not the same as a differential potency of unlike gametes produced by one individual. In the latter case there is no positive proof that such an effect is ever obtained, where the different gametes are all viable, and in most cases there certainly is no such action.

It has been found that the same selective action is shown when first generation hybrids are paired as when strains which have been long inbred are used. In other words, the effect is apparent, whether the plants have a long line of similar ancestors back of them or whether their immediate parents are diverse. It is also shown both in plants of weak growth or of full vigor, and whether the gametes of either type or both are alike among themselves or are exceedingly diverse in the hereditary factors which they carry. The only feature in common is that the cytoplasm is alike for all the germ cells of one plant, and this cytoplasm, in self-fertilization, is the same in which the pollen fulfils its function. This indicates that the tinequal fertilizing ability, is governed by the rate of pollen tube growth, although it may be determined after the male gametes are brought to the egg.

There is current in biological literature the assumption that heterogeneity in protoplasmic structure is favorable to developmental efficiency. This idea has been proposed again and again and applied in many different ways. Stated in general terms, it implies that the union of diverse elements and the resulting lack of balance stimulates growth. This is a heritage from Darwinism, and the writer believes that it is founded upon fallacious reasoning and is not supported by the facts.

The hypothesis has been used in theories of rejuvenation, explanations of hybrid vigor, and speculations concerning selective fertilization. The necessity for sexual reproduction at some time to maintain organisms reproducing asexually is no longer admitted. That the process of forming gametes which reunite to make a new individual may bring about a reorganization of the protoplasm with elimination of waste products resulting in increased growth is easily conceivable, but the significance of such a procedure is not necessarily to be found in the bringing together of unlike elements. The vigor of hybridization has now been put on a basis of pure inheritance, and the physiological stimulation hypothesis is no longer needed. Homozygous factor combinations, according to present theory, are more efficient than heterozygous combinations of the same factors. It now seems that self-prepotency, except in those cases where a definite process has been developed to prevent self-fertilization, indicates that unlikeness instead of favoring fertilization is a hindrance.

So valuable have been the evolutionary advantages of sexual reproduction in increasing variability that many contrivances have been perfected to insure the fulfilment of the function responsible for its creation. Self-sterility or self-impotency is one of the many special adaptations which serve this purpose. The evidence for selective fertilization favoring organisms of the same type, or self-prepotency, is limited just now to one or possibly two species. Will it not be strange to find it so restricted? Is it not more likely to be a general phenomenon manifested in some degree by many organisms? Even in those cases where cross-fertilization is made imperative by a physiological impediment to self-fertilization the same tendency may operate although overwhelmed by the special adaptation. One cannot insist that such is the case, with the evidence isolated as it is at present. Neither is it maintained that the reaction of the cytoplasm of the pollen tubes with the tissues of the host, preceding fertilization, has any relation with the processes which go on within the cells after fertilization. But the prepotency of germ cells acting upon the same or similar individuals which produced them is another indication that homogeneity, likeness, similarity, familiarity, or however it may be described, in protoplasmic structure is consistent with and favorable to the highest developmental efficiency.