This isn't a debatable question.  Just as several plants in a pot are  not 
true replicates, subsampling bees from one colony does not qualify as a true 
 replicate.  This is statistically non-defensible.  You're sampling  
individuals from closely related sub-families - all same mother, different  
fathers.


>  2.  >     She also made the questionable decision to  take all the bees 
for
> all “replicates” from only one colony, instead  replicating the 
experiments
> with bees from multiple colonies, which  may differ in health and
> susceptibility to the toxin.

You  answered your own question. Bees from multiple colonies may differ  in
health and add further factors to an already complicated study.  These> 
studies are complicated which is why many researchers were never  interested
in doing.
No, any researcher doing this doesn't understand statistics.  Its a  
fallacy introduced by some of the private testing companies looking to save  money.
 
Atkins and Johansen knew this - their protocols called for sampling at  
LEAST three colonies.  
 
Here's the problem - any colony may range from susceptible to average to  
resistant to say a chemical and/or pathogen.  If you BASE your study on one  
and only one, let's say its a pesticide label study  - you run the risk of  
basing the entire decision on a colony of unknown status - does it represent 
the  majority of bee colonies?  Is it a particularly sensitive colony?  Is  
it so robust that nothing is likely to be seen.  You can't answer these  
questions if you only use bees from one colony.

             ***********************************************
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