As per why not mark bees and let them forage - that's what Neukirch did to determine the average lifespan of forager bees.  Winston did a variety of trials with marked bees.
We tried doing this and found it to be extremely problematic.  Foremost was that depending on the source colony, many blocked the return of any marked bees, presumably due to the odor of the marker.  We tried a variety of paints - acrylic, oil, shellac (worked best, but still not good), and even glue on tags.  When one is trying to assess mortality, having bees prevented from entering their home hive, by being blocked or even balled, throws a big false mortality factor into the trial.  Then there's the problem of seeing the returning bees.  Unfortunately, they don't always land right-side up, and finding them inside the hive is really labor intensive.  Bar codes, and cameras, all have limitations in terms of getting accurate return counts.  RFID technologies are a bit better, but require grabbing bees to stick on the tags (which in itself is stressful and prone to injury to the bees), the RFIDs that a bee can carry add a load that affects foraging behavior, and the readers are expensive.  We did some pioneering work on RFIDs on bees and concluded that they altered bee flight, making foraging more difficult.  Finally, when trying to assess the toxicity of a pesticide, one has to control as many other variables as possible.  As Jose pointed out, and as the discussion of one simple factor, whether to supply water or not to the bees, even in an incubator, there are lots of things to take into consideration.  Marked bees from full-sized colonies, free to fly for several days under changing weather, plant forage, and vulnerable to birds, vehicle windshields, spiders, etc. all add uncontrolled factors that contribute to bee mortality.
The alternative is to use whole colonies, probably the gold standard for assessing what may happen, but now we are talking about harm to tens of thousands of bees, and the colonies themselves.
By comparison, a few cups of 30 bees taken from a colony is far less likely to cause harm.  And, if the cage assays are properly conducted, and the bees live normal lifespans, they're not necessarily stressed, other than they can't fly, but that's the case for months in any northern colony such as those in my part of Montana.  Regardless, there will obviously be stress involved with the bees exposed to the pesticide, but that's unavoidable.  
Could one mark or tag bees to conduct pesticide registration tests - sure.  But, given the plethora of additional external factors that would impinge on the control and exposed bees, the number of bees needed to yield statistically significant results would go up dramatically compared to the number of bees needed for cage trials.

From an individual bee's perspective, being stuck in a cage, and especially being dosed with a pesticide would suck.  From the colony perspective, a few hundred bees taken for testing out of tens of thousands of bees is a bit like us shedding skin cells.

Jerry

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