Hi Lennard,

In your email you spoke about the limits of models as predictors, based on lousy data input.

Do you think models are useful as tools for designing experiments?  I was trying to think that way in my earlier clumsy effort.

So what I am getting at is, if you put all the *major* factors you know about into a model (perhaps lumping some of them together, such as "chemical agents" for the sake of simplicity) and then think about overall implications using that math, could you come up with a priority list of experiments?

For instance; let's say that the most parsimonious model reveals that the only way to reduce effect with one (lumped) variable is to change the environment (this was my monoculture comment).  Then one could set up field trials with colonies in a manner (I hope) that varies only by the way in which nearby agriculture is conducted.  All other parameters would be the "same" insofar as possible.

All the pollutants, chemicals, pests would still be there.  But the effect might be reduced.

Then, going back to the model, one could start tp refine it to separate out the elements of the "chemical agents" etc to re-assess which parameter to tweak, and so on, peeling off layers of complexity one at a time, without ignoring all the others.

Doable??? Crazy???

What I see as problematic after "lurking" here for many months is that people tend to focus on one thing at a time, when in fact the issues are all interconnected.  So there will never be a solution, IMO, until the approach is altered.

Christina

             ***********************************************
The BEE-L mailing list is powered by L-Soft's renowned
LISTSERV(R) list management software.  For more information, go to:
http://www.lsoft.com/LISTSERV-powered.html

Guidelines for posting to BEE-L can be found at:
http://honeybeeworld.com/bee-l/guidelines.htm