You yourself answered......it should be as small as possible... But at the

same time it should control maximum variation for the trait.

In my opion it should as near as near to associated marker but should also in LD (linkage diseqlibirum) range of that spieces for example in wheat it ranges 2 to 10cM). Also that QTL shuold higher phenotypic variance value i.e PVE. Another option is fine mapping of QTLs to identified candidate genes in that region.

I think here the concept of major and minor QTLs should come.

Major QTLs are generally long and have gives phenotypic variance more than 10%. The chances of number of effective genes for the trait are more. Therefore they are lengthy.

e.g. QTL qSBR11-1 has 26 defense related genes for sheath blight resistance. It is nearly 2Mb long, but proven highly effective.

Minor QTLs are small and shows phenotypic variance less than 10% and also genes present in small QTLs are less. We cannot expect higher resistance/tolerance (in case of disease or stress) from these QTLs and they will not be observed every season or year.

In short, the length of QTL does not matter. It is the phenotypic variation for the trait governed by the QTL which matters.

Hi, one possibility is to use QTL in (weighted) genomic prediction.

I don't know that much done in the plant species but in livestock species, some below references might be useful. Of course, it is a long way to go, but it can be feasible when accuracy of mapping of QTL and data collection are enough.

1. Zang et al Accuracy of whole-genome prediction using a genetic architecture-enhanced variance-covariance matrix

2. Improving accuracy of genomic prediction by genetic architecture based priors in a Bayesian model

3. Predicting bull fertility using genomic data and biological information

4. Genomics to systems biology in animal and veterinary sciences: progress, lessons and opportunities