North American Fruit Explorers mailing list at ibiblio

["Joe Hecksel" <jhecksel@voyager.net>]

> My reply:
> 
> Just one caution, and this may "excite" a few people.  Before I 
planted a whole lot of blight resistant pears I would trial them in my 
area.  

I had personal conversations with a fruit tree breeder.  They 
characterized a selections as blight resistant if the infection self-
limited at the junction between the current seasons growth and older 
wood.  They justified this definition in two ways.  
-Economic impact was limited as the tree did not die.  Commercial 
growers care about economics.
-There are interactions between what is genetically available and the 
statistics involved in combining them with other economically important 
traits.

Quick statistics lesson for those who suffer from insomnia.  The 
distribution of traits within a population are often characterized with 
a bell shaped curve (no relationship with the pear breeder).  
Approximately 2/3 of the population will fall within +/- one standard 
deviation of the average.  The variation within that 2/3 of the 
population might not show enough difference to be able to reliably 
measure. 

The breeder normally cares about only one side of the 
distribution....more resistance to cold, fireblight, salt, etc.  So 
they quickly eliminate about 85% of the population as genetic sources 
of those traits.  Then there is the issue of how well a characteristic 
passes from generation-to-generation.  Some characteristics depend on 
multiple genes or are reinforced by "general vigor".  Those 
characteristics tend to have poor inheritability.  So the hapless fruit 
breeder is stuck with using an even smaller slice of the population as 
a genetic source for those traits.  

It often happens that the source of resistance is very deficient 
in "quality" attributes like smooth skin, good skin color, good taste, 
good texture, and so on.  So the breeder will engage on a program of 
trying to breed quality and resistance into selections.

So what is the big deal?  Why is it so tough to breed a selection that 
has +3 standard deviations for 5 independent characteristics?  Because 
the first characteristic will weed out all but 3,000 of the first 
1,000,000 seedlings.  Because the characteristics are independent 
culling for the second characteristic will remove all but 9 of the 
3,000 that remain from the first cull.  The third characteristic will 
eliminate the nine.  

To have a reasonable chance of getting one survivor for +3 standard 
deviations of five independent characteristics you need to start with 
200,000,000,000,000.  That is a 2 followed by 14 zeros. 

The work arounds are to use multiple generations to enrich the gene 
pool and to settle for less than +3 standard deviations for some 
characteristics like fire blight resistance...at least for early 
releases from the breeding program.

-- 
Joe Hecksel
Personal Webpage http://my.voyager.net/~jhecksel