The Evolution of Genetic Systems (1958) pp 151-155
C. D. Darlington

iii. Sterility and Balance

We may see the effect of segregation on sterility most simply in a triploid plant. Spores are formed with all numbers of chromosomes between the haploid and the diploid. Those with intermediate numbers are unbalanced. They develop on the female side to produce egg cells. On the male side however, on account of the longer life of the spore, a proportion usually die before the pollen grain germinates. When they survive the balanced and unbalanced grains have to compete in growing down the style; only a small proportion succeed in fertilising the egg cells, and these are likely to be the balanced ones. When a triploid is crossed as a female with a diploid as a male the result is therefore a higher proportion of unbalanced progeny than in the reciprocal cross. This is notwithstanding a certain differential mortality among the young embryos which also reduces the proportion of unbalanced ones. When the triploid is the male parent very few progeny except diploids and simple trisomics are usually produced.

Sterility of a triploid is thus due to unbalance in the progeny. Now there are occasional plant species which do not show any serious effect of unbalance. This is sometimes due to the basic set being itself polyploid in origin, and sometimes to there being so much translocation and duplication of segments of chromosomes that a mechanical diploid is physiologically a polyploid. This is evidently true of Hyacinthus orientalis, n=8, for in this species different plants with all chromosome numbers from 16 to 32 are equally vigorous. In keeping with this lack of depression from unbalance, they are also almost equally fertile: how fertile exactly we still need to know. It is also in keeping with this situation that vigorous diploid plants deficient of chromosome segments have been found both in cultivation and in wild populations.

5 Darlington and Mather, 1944;
Darlington, Hair and Hurcombe, 1951.

How are we to describe the Hyacinthus situation? There are various ways of representing it. But for the present the easiest is no doubt to say that each chromosome is nearly self sufficient. Each chromosome has its own balance. Differentiation there must be; but it is as much within as between chromosomes.5

Absolute deficiency is, in the normal situation, an even more serious cause of sterility than unbalance. Rhoeo discolor having a ring of twelve chromosomes can produce, through errors in the orientation of the ring, pollen grains with five and seven chromosomes instead of six. Those with five never reach the first mitosis. Those with seven may germinate and they may perhaps grow down the style. They never give rise to offspring. Or so it seems for the seedlings all have the same uniform number and appearance as the parent.

These examples show us why a hybrid like Raphanus-Brassica is sterile. Owing to lack of pairing, pollen grains and embryo-sacs are produced with all numbers and combinations of chromosomes; none of the parental types are reproduced except by a rare chance, and a balanced combination will arise only by complete non-reduction; that is by omission of one of' the two sexual processes.

In an entirely opposite way, as we saw, following complete pairing and crossing-over in every chromosome of the diploid Primula kewensis the original parental combinations are even less likely to be produced. In consequence likewise the hybrid is absolutely sterile. In the tetraploid through pairing and segregation of similar chromosomes, uniform and balanced gametes are produced and the plant is fertile.

Fig. 28. Diagram showing the alternative conditions of segregational sterility (a) in a hybrid diploid, (b) in a non-hybrid tetraploid. (After Darlington, 1932a.)

The position of an autotetraploid is significantly different. Its chromosomes, as a rule, change partners at pachytene and, forming chiasmata in these different associations, they remain quadrivalents at metaphase. These quadrivalents are, except perhaps under very special conditions, incapable of regular orientation and segregation in every cell. With linear orientation three-and-one segregation often results. In this way a tetraploid cherry with eight potential quadrivalents has given a segregation of 19:13 instead of 16:16. With indifferent diamond-shaped co-orientation of the four chromosomes two may be left on the plate at anaphase to divide as univalents. Moreover, trivalents and true univalents often occur (Fig. 28). Thus there is irregularity of segregation in the autotetraploid and infertility is the result.

iv. Selection for Fertility

6 Upcott, 1939a.

There are two kinds of exceptional circumstance under which these rules do not hold. The first is that of the undifferentiated chromosome complement, as in Hyacinthus, where tetraploidy and even triploidy fail to destroy fertility. The second is that revealed by comparison of a number of species of Tulipa which are evidently autotetraploids. These vary in the number of quadrivalents they form at meiosis subject to two conditions: the numbers of changes of partner at pachytene and the frequency of chiasmata. Since the frequency of chiasmata per chromosome is always reduced in a tetraploid owing to the larger nucleus and the slower pairing some tetraploids such as T. chrysantha have hardly more than the minimum of one chiasma per bivalent. Quadrivalents are therefore almost entirely excluded.6

7 Bremer et al., 1954.

Thus, if sexual fertility is important for a new tetraploid, selection in meiotic behaviour should readily improve it. Experiments with tetraploid rye have shown what can be done. In the course of four years of selection for fertility improvements have been made in the proportion of good seed set and in the proportion of this seed which had the balanced tetraploid number and therefore gave good plants. Moreover this improvement was correlated with a reduction of laggards and of unequal segregation at meiosis in the pollen mother cells.7

8 Mashima et al., 1955.

Rye is an outbreeding plant. In rice, an inbreeder, parallel results have been obtained but only where the diploids have been produced by crossing varieties. Inbred diploids give tetraploids in which selection has no effect: there is no variation in the genotypic control of meiosis from which the breeder can select.8

9 Lawrence, 1931.

By such processes of selection we may suppose that Dahlia variabilis which seems to have arisen as an autotetraploid garden plant in pre-Columbian Mexico has come to combine regular quadrivalent formation and seed-fertility with free tetraploid segregation.9

The new unselected autotetraploid however always forms univalents. It thus yields unbalanced gametes. and its fertility is reduced. Now here is the contrast and, if you like, the paradox. The fertile diploid gives an infertile autotetraploid. The sterile diploid gives a fertile allotetraploid. There is a negative correlation between the fertility of diploids and that of the tetraploids they give rise to. Hence autotetraploids in nature do not usually establish themselves as new species unless sexual fertility can be to some extent dispensed with.

10 Darlington, 1956a.

If we enquire into their occurrence among plants we are at once led to discover how this happens. We find that the autotetraploid forms nearly always arise in individuals or varieties which differ from the average character of the species in having a greater propensity for vegetative reproduction. They are, we may say, pre-adapted to polyploidy. Or, better still perhaps, we may say that polyploidy and vegetative propagation mutually select one another.10