Redesign of breeding programmes. Most applications
of molecular genetics to breeding programmes have
attempted to incorporate molecular data into the
existing programmes. The effective use of molecular
data might, however, require a complete redesign of
breeding programmes. For example, in plants, the
optimal design for MAS is to allocate test resources to
a single, large population, such that the probability of
detecting QTL is high, whereas for phenotypic selection,
the optimum is to have smaller populations in
several locations to control for environmental variation51.
In addition, population structures and statistical
methods that allow the combination and use of
QTL information across lines are needed. Other
changes that are required for plant breeding programmes
are reviewed by Ribaut and Hoisington52.
Similarly, in animals, strategies are required that integrate
the collection and analysis of phenotypic data
for QTL detection with the use of this information for
MAS (for example, REF. 37).
Furthermore, breeding strategies must be developed
that take better advantage of the unique features
of molecular data. For example, to capitalize on the
ability to select on molecular score at an early age, several
rapid rounds of selection exclusively on molecular
score could be conducted. The speed of selection is
then mainly limited by the reproductive cycle. Such
programmes have been proposed for plants by
Hospital et al.47, by incorporating one or two generations
of off-season selection on molecular score alone,
and have been shown (by simulation) to increase
genetic gain greatly. In animals, such strategies are
effective only if combined with technologies that break
the normal reproductive cycle. For example, in several
livestock species, the technology exists to recover
oocytes from the female before puberty, as early as
from the unborn fetus. When combined with in vitro
fertilization and embryo transfer, this reduces generation
intervals to several months, compared with at least
3 years with regular reproduction in cattle53. Haley and
Visscher54 suggested that the time required for one
generation could be further reduced if meiosis could
be conducted in vitro. Such technology, combined with
nuclear transfer, would allow a breeding programme to
be conducted in the laboratory, without creating animals.
Although some of this work is at an early stage, it
is clear that the benefits of MAS will be much greater
when molecular technology is integrated with reproductive
technologies.