Any prognoses on combining ability of parental forms in plant breeding (especially for self-pollinated crops) are extremely difficult, because the real potential of hybrid combinations is revealed only in later generations. Nevertheless, many examples are known when some varieties and lines demonstrate their exclusive value as parental forms for crosses. The search for such parental forms in plant genetic diversity is extremely vital.

In many cases it is very difficult or impossible to use genebank accessions selected from the collection based on phenotype for breeding purposes. This is due to a number of reasons:

• the accessions may belong to a taxon that is reproductively isolated from varieties or lines under improvement;
• their useful characters may be controlled by genes with negative pleiotropic effects or by those closely linked to undesirable genes;
• polygenic control of the desirable trait under consideration;
• genetic peculiarities of the varieties (lines) under improvement.

Considering these reasons, the well-known terms "source" and "donor" were defined as follows:

• a source is an accession that has a high mean value for a given character and belongs to a cultivated species or its relatives;
• a donor is a source that:
  1. can be crossed with varieties (lines) under improvement to produce a viable, highly fertile, hybrid progeny;
  2. has a desirable effect on the broad spectrum of hybrid combinations;
  3. possesses no negative traits that are closely correlated to important characters or reduce yields to an economically unacceptable level.

Today plant breeders wish to know more about phenotypicaly superior accessions. It is thus necessary to look for approaches that allow us to obtain complete and objective information of long-term value. The data obtained by methods of genetics meet these requirements in the best way.
In recent years a considerable amount of work has been done to study the genotypic peculiarities of accessions possessing the highest mean values for major valuable characters. Nevertheless, the determination of genetic nature of many accessions with desirable phenotypic characteristics is not practicable. A way out of this situation is to apply a systematic approach in studying the intraspecific diversity of plants.

The figure represents the main stages of proposed systematic approach. In dependence on breeding goals, accessions with high or low mean value of character are of practical interest. The best accessions based on phenotype and accessions reflecting intraspesific variability of character can be included in sub-collection on character under study – first stage. It is important to determine the degree of genetic similarity of the best accessions (second stage) and understand the genetic system of intraspecific variability for a given character (third stage). In applying standard or specially designed genetic procedures, it is necessary:

• to determine which level of character is dominant and which is recessive;
• to clarify an influence of genetic background, cytoplasm, and environments on the character;
• to get an idea of the number of genes responsible for intraspecific diversity of the character;
• to study the effects of the most important genes and their interactions;
• to identify the major genes for the character in test crosses;
• to determine a possible association of these genes with known genetic markers.

Genetic collections of accessions with identified genes are created for characters well studied genetically - forth stage. It is also important to transfer these genes through backcrossing into a common genetic background in the process of developing near-isogenic lines. The obtained information serves as a basis for revealing in collection and developing donors of useful characters - fifth stage. Developing new donors is achieved through backcrossing. Valuable alleles are transferred into the genetic background of the varieties with well-known combining ability. A breeder can purposefully use this valuable material in crosses to solve the stated tasks.

The proposed approach permits to quickly input all data accumulated in the world on different characters. All this together provides adequate information necessary for understanding the system of intraspecific variability of useful characters and organising on this basis a way of a "phenogenetic screening" the wheat diversity. By studying a limited number of specially selected accessions, the genetic potential of a species can be determined, and genes that play a particularly important role in controlling useful characters can be identified and included in breeding programmes. The system approach promotes more intensive introduction of genetics in germplasm evaluating and developing new initial breeding material. Quite promising potentialities are open with modern methods of molecular biology.