| In order to facilitate the handling of the maximum number of progenies
of treated plants several methods of early selection have been proposed.
For kernel yield in cereals the possibility of a direct early selection
was tested by RAWLINGS, HANWAY and GARDNER (1958) and BOROJEVIC (1965,1966a)
by screening for a high kernel number and by GAUL and MITTELSTENSCHEID (1961)
by selecting for superior 1000-grain weight. To avoid a simultaneous negative
shift of other yield components, e.g., a decrease of kernel number by the
selection for 1000-grain weight, an indirect early selection via indicator
characters was tried; for example, GAUL and MITTELSTENSCHEID (1961)
selected for early types, and BHATIA and SWAMINATHAN (1962) or BOROJEVIC
(1966b) for awned mutants, of which the yield was determined after the necessary
multiplication of the original variants. But so far little is known about
the correlations between such indicator traits and the desired yield components.
By these kinds of early selection, one therefore runs the risk to lose rather
than gain the desired variability. In the following investigation on mutation breeding for kernel yield in spring wheat we abandoned any selection towards definite characteristics. For the reduction of the population size in M2 and M3 we used an early selection on the basis of variability as such, to keep the chances high for the following direct selection on yield in M4. The variability was measured with the aid of the characteristic "ear length", which is easily accessible, sufficiently indifferent to yield, and having relatively high heritability. Experimental results Seeds of a line of spring wheat (3880/48, OTTO BREUSTEDT, Schladen, Germany) were soaked in different concentrations of EMS (for details see TRUJILLO, 1968). The M1 was raised under glass and the M2 planted ear to row. The progeny of an M1-ear was called "family", that of an M2-plant "line". 1888 families were grown in the M2-generation. The efficiency of the mutagenic treatment was measured phenotypically by means of the frequency of viable drastic mutations in M2 which ranged from 6.3% to 8.5%. But for the present purpose only those 14.650 families were maintained which did not show drastic mutants, but were phenotypically normal. However, even these were far too many to be included into a yield test, which needs conventional field plots in replicate. Of the M2-families two groups were selected consisting of 100 families each with 10 plants (lines): Group I was characterized by a high variance of ear length and Group II by a variance equal to the untreated control (see Fig. 1). In the following M3 the progenies of these two groups of 1000 lines each differed visibly, as well as in the various measurements performed on the mature plants after harvesting. The variability of Group I was always significantly higher. Moreover, 7.9% of the 1000 lines segregated for new drastic mutants in Group I, but only 3.8% in Group II. |
| <-- Back | | | --> Next |