Forbes W. Robertson
Articles written in Journal of Genetics
Volume 50 Issue 3 February 1952 pp 414-448
Volume 51 Issue 2 January 1953 pp 276-316
This paper describes experiments on a strain of
Under selection for long wings, the strain has remained for more than fifty generations in unstable genetic equilibrium, characterized by a phenotypic variance 50 % greater, and size 8% greater than in the unselected (control) stock.
When selection is relaxed or reversed, variance declines sharply and the deviations of wing and thorax length decline by 50 and 20 %, respectively.
No second-chromosome lethals were found in the selected stock, but some flies were homozygous for a third-chromosome gene-or block of genes-which was lethal in the genetic background of the lethal test.
Progeny tests indicated that about 50% of the variance of wing length was due to additive genetic effects, compared with 40% after a generation of relaxed selection, and 30 % in the unselected stock. Selection appeared to have increased the additive genetic variance 2 1/2 times, leaving the remaining variance (other genetic plus environmental) unchanged.
Additional tests indicate that, in spite of the appearance of additiveness given by the progeny tests, artificial selection has been favouring the flies most heterozygous for genes affecting body-size.
A genetic asymmetry in the relative changes of wing and thorax length during selection for large and small size is discussed.
The genetic correlation between wing and thorax length is about 0.75 in the unselected stock, and even higher (0.86) after selection. Other evidence supports this conclusion. Nevertheless, wing length declines relatively 2 1/2 times as much as thorax length, when selection is relaxed.
This apparent contradiction may be due to a postulated effect of continued selection in modifying the effects of unfixed genes on the selected character. Comparison of the genetic variances of the two characters before and after selection lends some support to this hypothesis.
There is a negative correlation between parent size and viability, in progeny tests on the selected strain, which appears to be due mainly to a correlation between size and larval mortality; but this alone could not account for the failure to advance under selection. Neither sterility nor mating behaviour can account for this failure and some further genetic mechanism must be at work.
Volume 51 Issue 3 July 1953 pp 586-610
Volume 52 Issue 3 September 1954 pp 494-520
A crossing method is described for creating all possible combinations of major chromosomes from pairs of inbred lines of
Complete chromosome analyses have been carried out on two pairs of contrasted lines of different size, descended from the Nettlebed and Edinburgh wild stocks. Each such pair comprises a small line, descended from a strain selected for small body size, and an approximately normal-sized line, inbred without selection from the same stock. Three unrelated lines inbred without selection have been studied in a similar way, except that twenty-one out of the twenty-seven possible combinations for each pair have been studied in females only.
The accuracy of the method of combining chromosomes was demonstrated by the agreement between preparations of the same genotype by different means, and also by the level of the within-culture variance, which was generally of the same order as that for untreated inbred lines.
The within-culture variance is not constant for all genotypes, but tends to decline with an increase in the number of heterozygous pairs of chromosomes.
When the unselected and small lines are crossed, a highly non-additive situation is revealed by the size of the
In the analysis of the unselected and small Edinburgh lines the size of the different types could be accounted for by aggregate dominance of the chromosomes of the larger
In the Nettlebed combinations, aggregate dominance and additive combination of non-homologous chromosomes account for the size of the majority of the types. But there are also a number of striking interactions which increase or decrease size, leading to different effects of particular substitutions and different dominance relations in different genetic backgrounds. Most of the larger interactions occur in genotypes carrying several chromosomes from the small line. The behaviour of the X-chromosome of the small line is exceptional in being incompletely recessive in all backgrounds.
In the combination of chromosomes from the unrelated, unselected, inbred lines, interactions between non-homologous chromosomes are much more frequent and striking. The substitution of a single chromosome or of a homozygous pair may increase or decrease size, according to the genetic background.
Inter-crossing these unrelated inbred lines always leads to heterosis in the
The results are discussed in relation to the mechanism of heterosis, inbreeding decline and possible ways in which selection has changed the genotype to produce small size.
Volume 55 Issue 3 December 1957 pp 410-427
The paper deals with genetic variation in the number of ovarioles, which make up the pair of ovaries in
Mass selection in a strain for high and low ovariole number led to an asymmetrical response, since selection for lower number led to about 14% reduction, whereas selection in the other direction increased ovariole number by more than 50%, and the response was apparently continuing when the experiment was stopped after ten generations.
During selection, the variability in the high line rose above the initial level and then declined and fluctuated around the same level as the low line, which showed no consistent change. The behaviour of the variability, together with the results of intercrossing the high, low and unselected strains, and also back crosses, suggested that a good deal of the response to selection for high ovariole number was due to the increase in frequency and eventual fixation of a recessive gene, which, when homozygous, increases ovariole number by about 25 % over the heterozygous and alternative homozygous combinations. Further evidence of genetic variation tending to cause striking increases in ovariole number was provided by comparison of groups of genetically identical individuals which were created by combining haploid sets of chromosomes from the wild stock with those of an inbred line. Thus the origin of asymmetry in the present selection experiment appears to be quite different from the asymmetry noted in the results of two-way selection for body size.
Inbreeding leads approximately to 6 % decline in ovariole number, which is of about the same order as that observed for body size. The variability of ovariole number between individuals is greater in inbred lines than in the crosses between them.
When body size is reduced by under-feeding the larvae, there is a proportional reduction in ovariole number and also egg production, but the number of eggs produced per ovariole is unaffected even by striking changes in body size. Lack of correlation between body size and ovariole number or egg production provides a sensitive check on the uniformity of environmental conditions in experiments with genetically identical flies.
Provided the larvae have been reared under optimal conditions, individual variation in ovariole number is unimportant with respect to egg production. There is an appreciable ‘chance’ variation between genetically uniform individuals, but this does not involve a correlation with egg production. Selection for low number was accompanied by a slight reduction in the rate at which eggs were laid, while selection in the reverse direction led to a slight increase, but the differences in performance were slight, and somewhat uncertain in origin, compared with the great differences in ovariole number.
Independent selection for either high or low ovariole number and large or small size leaves the average value of the other character unchanged so there is no evidence of genetic correlation due to additive effects.
The variation in number of ovarioles between the two ovaries of a fly presents a further example of the regional indeterminacy in development which has been referred to as ‘asymmetry’. Under optimal conditions, variation in numbers on the two sides of an individual is uncorrelated, so that most of the non-genetic variation between individuals of a wild strain can be attributed to such chance variation. The level of such variation is the same in inbred lines and wild stocks ; it is a little lower in crosses between inbred lines, but it is uncertain how far this represents a real difference. Also, provided a logarithmic scale is used, the level of such variation turns out to be the same in the high and low selected lines, inbred lines and also wild strains, providing internal evidence of the validity of the log transformation.
It is suggested that the high level of ‘chance’ variation and also the considerable genetic variation between individuals of wild strains is related to the capacity of the ovary for physiological regulation, so that, within limits, actual number of ovarioles is of secondary importance. The possibility of detecting striking effects of segregation in such characters may provide otherwise unsuspected evidence of balanced polymorphism.
Volume 55 Issue 3 December 1957 pp 428-443
The variation of body size and egg production between individuals of a laboratory population of
From the results of two-way selection for body size and also egg production, the heritability of the two characters is estimated as respectively 0.43 and 0.18, suggesting that most of the genetic variation in body size behaves as additive in the statistical sense, while non-additive effects predominate in the variability of egg production.
There is a clear-cut phenotypic correlation between size and egg production among the individuals of the wild population and this is mostly genetic in origin. Selection for large and small size leads to comparatively little change in egg production. A series of different genotypes, each represented by a number of genetically identical individuals; were prepared by combining haploid sets of chromosomes from the wild population with a haploid set from an inbred line—and these provided no evidence of genetic correlation between size and egg production. Hence the genetic correlation in the wild stock is essentially non-additive in behaviour.
Gene-environment interaction has been demonstrated for both egg production and body size by comparing estimates of variance and heritability on the live medium with those found on a synthetic medium which permits normal growth. It is suggested that gene-environment interaction will be particularly important under the suboptimal condition in which the population usually lives.
Reduction of the body size of genetically uniform flies by varying the nutrition of the larvae leads to a proportional reduction in the rate at which eggs are laid.
The variability of development time has also been examined in preliminary experiments which indicate that about half the variance is genetic in origin. Under favourable conditions there is a high correlation between length of development and body size, while egg production tends to be negatively correlated, suggesting that egg production and growth rate are genetically associated.
The results are discussed in relation to the stability of average body size in. the population and the ease with which selection either way leads to striking change. There are strong indications that the apparently continuous variation of attributes such as size or egg production are heterogeneous with respect to the effects on physiology and development which contribute to the final variation, while the relative contribution of different sorts of effect may vary according to environmental conditions.
Volume 83 Issue 1 April 2004 pp 17-32
Volume 100, 2021
Continuous Article Publishing mode
Click here for Editorial Note on CAP Mode