Effects of the major plant color genes upon kernel weight in maize.

Brink (1934) has demonstrated that maize plants belonging to the anthocyanin series of color types differ significantly in their average production of grain. Comparison of the four anthocyanin types led to the conclusion that purple was much inferior to dilute sun red, while dilute purple and sun red exceeded dilute sun red in average yield per plant. Subsequent unpublished results indicate that there is probably no significant yield difference between sun red and dilute sun red. Two trials in successive seasons in which dilute sun red (A b pl) and triple recessive green (a b pl) were compared, suggest that dilute sun red has a significantly greater yield.

In 1938 and 1939 the writer conducted three additional experiments at Madison, Wisconsin, in an effort to clarify the status of those color types which had given inconsistent results and in order to include the brown class (a B Pl) which had not occurred in earlier trials. A number of ears resulting from the backcross A₁a₁ Bb Plpl x a₁a₁ bb plpl were obtained. Two experiments, the first including 12 backcross families in three randomized replications and the second, with 18 families in two replications, were grown in 1938. A third experiment (12 families, 3 replications) was grown in 1939. The heterozygous A B Pl plants used in backcrossing were not closely related to the a b pl stock and the segregating progenies exhibited considerable hybrid vigor. Five-eighths of the residual heredity in each family was derived from commercial strains of yellow dent corn adapted to Southern Wisconsin conditions.

The plants were classified as to color type and distinctively tagged. No

attempt was made to distinguish the a B Pl and a b Pl plants from a b pl in the green class. The frequencies of each type within each row determined; the mature ears from each color group in a row were harvested together. The samples were dried to a uniform moisture content, shelled, and the shelled corn weighed to the nearest ounce.

The mean shelled grain weights per plant for each plant color class in experiments I and III appear in table I.

Table I.

Mean grain weights per plant by color classes

	Experiment I (1938)		Experiment III (1939)
Phenotype	No. plants	Mean in lbs.	No. plants	Mean in lbs.
A B Pl (purple)**	674	.307(6)	600	.282(6)
A b Pl (dilute purple)	681	.361(3)	641	.323(2)
A B pl (sun red)	694	.355(4)	686	.318(4)
A b pl (dilute sun red)	688	.372(1)	682	.331(1)
a B Pl (brown)**	698	.344(5)	602	.305(5)
a B pl, a b Pl, a b pl (green)	2002	.368(2)	2000	.319(3)
Total	5437		5211
**Highly significant differences between this and other classes.

The analysis of variance for each of these experiments reveals that the low yield of purple is highly significant in both cases and that brown with a significantly greater yield than purple is significantly below the yields of the remaining four classes. The relative standings of the six color types with respect to mean grain weight are indicated by the numbers in parenthesis in table I. Dilute sun red has the largest mean in each experiment, the value being significantly (P = .01) greater than the pooled mean of the green, dilute purple and sun red classes in each case. In a combined analysis of experiments I and III the difference between dilute sun red and sun red is highly significant.

The results from experiment II are consistent with the other two experiments with respect to the purple and brown classes. The differences are again highly significant. The mean of sun red is second highest in the experiment instead of fourth as in I and III. This high value for sun red in experiment II is subject to question, however, for when the analysis is based upon kernels per ear instead of kernels per plant, sun red is fourth highest while the relative standings of the other are but slightly changed. In this experiment, also, sun red contributes disproportionately to the variance. The error term is larger than in the other experiments making it impossible to pool the results of experiment II with the others. A summary of experiment II and the total frequencies of each colar type are presented in table II.

Table II.

Mean grain weights per plant by color classes

	Experiment II (1938)		Total plants
Phenotype	No. plants	Mean in lbs.	I + II + III
A B Pl (purple)**	806	.320(6)	2080
A b Pl (dilute purple)	803	.370(3)	2125
A B pl (sun red)	884	.376(2)	2264
A b pl (dilute sun red)	920	.379(1)	2290
a B Pl (brown)**	848	.345(5)	2148
a B pl, a b Pl, a b pl (green)	2555	.367(4)	6558
Total	6816		17,465
**Highly significant differences between this and other classes.

A chi-square test for the corresppndence of the observed frequencies of plants in each color class to the expected 1:1:1:1:1:3 backcross ratio reveals that the frequencies shown in table 2 have a probability of .01. The largest deviations occur in the purple class which is smaller than expected and the dilute sun red class which is larger than expected. Since these are the classes which have the lowest and highest mean grain weights, respectively, it appears that the same genotypes which influence kernel weights also influence viability. Relatively large negative deviations also occur in dilute purple and brown, while the sun red frequency exceeds the expected. It seems probable that the dominant gene, Pl, has an adverse effect upon viability.

Plants with the purple phenotype carry the three dominant genes A B Pl and are much less productive than these plants in which one or more of these dominant factors is not present. The brown plants which have the genes B and Pl are at a similar but less marked disadvantage. The dominant genes were always present in heterozygous condition. Since the presence of a single gene A is the only known condition which differentiates the purple from the brown type within a given family, it appears likely that this gene acting in conjunction with B and Pl results in a decreased storage of starch in the kernels. In contrast it is found that dilute purple, dilute sun red, sun red, and green, all have higher mean grain weights than brown. In the three anthocyanin color classes A is present, but b, pl, or b pl are homozygous. The heterogeneous green class includes combinations of a with b, pl or both in homozygous condition. Therefore, it may be concluded that the B Pl gene interaction is effective in reducing the mean weight of grain per plant, presumably by affecting starch storage during development. The gene, A for anthocyanin pigment, in combination with B Pl increases the effect.

The relatively higher yield of dilute sun red in all three experiments is noteworthy because this is the genotype which is virtually universal among North American varieties of dent corn. While the evidence is hardly adequate to demonstrate that this genotype is always superior in grain yielding potentiality, the fact that A b pl yields are probably significantly greater than those of A B Pl is suggestive. In sun red as in purple and brown the development of deeply pigmented tissues must immobilize considerable quantities of carbohydrate which might otherwise be stored in the seeds.

The possibility that the results reported are actually caused by other genes, rather closely linked to the three segregating color genes cannot be entirely rejected on the experimental evidence now available. The foregoing conclusions are based upon a rather homogeneous sample of residual heredity tested in a single locality. Until further evidence is available on the point, however, it would be inadvisable to introduce B and Pl as markers in dilute sun red commercial breeding stocks.

Ben W. Smith