A plant breeder's job is to develop varieties that are truly "new and improved." I hope this article will give you an idea of how this is done.
The first question the breeder should probably ask is "What problems are farmers and ranchers facing that a new variety can help solve?" In other words, the breeder should identify a trait that needs improving.
As an example, let's say the breeder decides to develop a variety with improved forage quality. The breeder will then decide which species to work with, and then obtain germplasm of that species. Germplasm is a fancy word which means 'genetic material'. Think of it as a 'herd of plants'; out of this 'herd', the breeder will choose a few plants as parents for the next generation.
Commercially available varieties, seed collected from native stands, and strains developed by other plant breeders are all examples of germplasm. Usually 1000 plants or more of the germplasm are planted in the field. In cross pollinated species like alfalfa, tall fescue, big bluestem, etc., each of these 1000 plants will be genetically different. After establishment, information on the plants is gathered so that the breeder can decide which plants to keep.
In our example, forage samples will be taken in summer and analyzed for forage quality in the lab. Out of the 1000 plants, typically 50 plants or so will be selected. These plants are then dug up and re-planted in what's called an 'isolation'. This is a small field that's isolated from other plants of the same species to prevent outside pollination. Seed is produced from the 50 plants, completing one generation of selection. Often, a breeder will repeat this process and select for several generations.
The next phase involves testing. Small plots are planted in the field to compare the selected strains to varieties currently being used by farmers and ranchers. If a new strain does better than these standard varieties, then more testing is done, usually for more years or at more locations. If the selected strain continues to do well, then more seed is produced. The best test of a selected strain is to compare it to a standard variety in a grazing trial. This tells how the strain will do in a "real world" grazing situation, and how animals perform when they graze the new strain.
The goal is for grazing animals to perform better on the new strain than on the standard varieties. If this happens, then usually the selected strain is given a name, and released as a new variety. Seed of the new variety is produced, and then sold to farmers, ranchers, and landowners.
This is generally how forage varieties are developed, but different methods will be used depending on the species, the trait, and the breeder. The entire process, which includes trait identification, selection, testing, seed production, and marketing, can take eight to 15 years. So maybe that new variety isn't exactly 'new', but it is improved.