>Steve Modena writes:
>One of the primary obstacles to bioengineering miracle plants is that
>aside from a few "major" genes, most of any plants traits are controlled
>by multiple genes. My impression is that current progress with QTL
>research via RFLP's is that multi-gene factors are not well behaved. This
>may be surprising only to non=plant breeders. The possible combinatorials
>involved with even 10 genes and a few alleles is astronomical.
As an RFLP/QTL mapper (in Populus) I would say that QTLs identfied
so far in crop plants behave quite well. The analytical methods
are such that only well-behaved QTLs will be detected, in fact.
This is a limitation of the work, but being able to account for
a very large proportion of phenotypic variance using a handful
of molecular markers is, I think, an important advance. Andy
Paterson has shown that many QTLs are conserved across tomato
species and across environments; others are identified only
in a subset of environments. I don't expect breeders to be surprised
by this, but after all, we mappers are just getting started.
>The idea behind bioengineering to the identify factors that are important
>and alleles that are superior. Yet somehow, this seems to still overlook
>the combinatorals behind making those identifications. In other words,
>why should analysis by RFLP gels be more efficient that traditional
>methods employed by breeders/geneticists? Both suffer the same problem:
>inability to grow and screen a sufficient number of plants to identify
>the _most_ superior genetypes via phenotypes.
The ability to screen large numbers of individuals is precisely
the benefit of genotyping (as opposed to phenotyping). Working
on trees, I see many potential benefits for early indirect
selection for traits of high value that are expressed only
in mature trees. If I can show at the seedling stage that a
seedling will have low specific gravity wood as an adult, and
all I'm interested in testing are those clones with high
specific gravity, I could either reduce the number of clones
in my very large and expensive clone test, or keep the test the
same size and test many more promising (as opposed to random)
I don't mean to imply that we are close to this goal, but we
are taking the first steps by trying to understand the mendelian
genetics of quantitative traits. I bet Ron Sederoff in your
Forestry Department can tell you some of the neat things they
are doing along these lines in loblolly pine.
Biochem and Forest Resources
University of Washington