https://doi.org/10.3390/insects10010019

Thanks Gene

Interesting study.  I gave it a quick read, so I may have missed some details.  First impressions, the lack of statistical significance is not surprising since not more than 1-2 colonies survived the winter from all groups, including the controls.  The authors used an unbalanced experimental design, with 9 controls and seven colonies in each of the 4 treatments.  Given the small numbers of survivors the next spring, the controls should have had a slight advantage over the treatments since there were 9 rather than 7 colonies at the start of the experiment.  

All in all, too many treatment variables for the number of colonies.  Varroa is a highly uncontrolled and variable factor.   Pesticides effects by chemical and dose are also highly variable.    The resultant degrees of freedom (df) were 3 for each pesticide treatment.   Since it's a repeated measures approach, the higher df for days is expected, but it can be misleading to a reader not familiar with repeated measures.  

Spraying the foundation with wax that was or was not treated with pesticides was an interesting approach.  I suspect that the bees may have added their own wax over top of the chemical treated wax sprayed on the foundation plastic.  It's possible that the new, drawn wax could have been relatively chemical free. 

I'd have liked to see the original wax analyzed, since wax rendering does not necessarily destroy all pesticides.  Roger at the Gastonia labs, some years ago, discovered some surprising levels of pesticides in new wax foundation.

Minimally, the final comb should have been analyzed to assess the actual residual chemical dose in the wax to determine the actual dose achieved  and to see whether any partitioning from the contaminated, sprayed layer of wax, to the fresh, bee produced and drawn wax had occurred.  In other words, the bees may have 'capped' over the treated wax with fresh wax.  There's no data other than calculations that the doses were comparable to levels seen in 'field realistic' comb samples.  We've long known, that in brood nests, chemicals and spores can be layered in cells that have been used over and over.  That's good for the short term, but not so good when the bees decide to tear down and rebuild comb; consequentially potentially releasing dormant spores and chemicals that are relatively stable when sealed in wax.

All in all, they started with 30 package colonies, divided them into 7 per chemical treatment, 9 for the control. By the end of the summer, they had 5 remaining colonies in each treatment (5 of 7) and 7 of the 9 controls.  By spring, only 7 of 30 colonies survived.   They had 2 survivors in two of the chemical treatments (2/7), one in one of the chemical treatment groups (1/7), and 2 in the controls (2/7).

After overwintering, they had 28.6% survival in two chemical treatments, 14.3% in one of the chemical treatments, and 23.8% overall survival of chemically treated colonies,  There was 22.2% survival of controls (got to watch those unbalanced groups).  Based on this,  the control colony survival was lower than 2/3 of the chemical treatments.   All of this is likely an artifact of inadequate replication. 

I'd argue that there was NO trend.    Obviously, with 2 survivors in all but the chlorpyrifos treatment, which had one survivors; the trend, if any, was lowest survival when exposed to varroa and in combination  with an organophosphate pesticide, second lowest survival for controls, and slightly better survival for the other two chemical treatments.  I'd have to conclude that the controls were at a disadvantage when subjected to varroa mites.   Obviously, that would be a specious conclusion, there were too few surviving colonies to make any such conclusions, whether statistically or subjectively.

For studies of pesticides and/or mites, we aspire to have a minimum of 12 colonies per treatment, more if possible.  Adequate replication is always a challenge - and it's what drives up experimental costs.    

Power and Size and Animal Use and Care - Switching Gears.

The following comments are not meant to criticize the authors of the TX paper.  It does bring up the issue that tests to estimate the Power of a Test and Sample Size are no longer simply statistical considerations.

Those of us at Academic institutions now have to send our experimental designs and protocols through Institutional Animal Care and Use Committees.  I did a quick review of this Power and Size issue in this context.

Given public concerns about use of animals in experiments, one of the points Institutional Animal Care and Use Committees should consider, when reviewing experimental protocols, includes “justification for the number of animals used.”   In other words, many of us now have to justify numbers of test subjects before being allowed to conduct an experiment, release a survey, etc.   

I found a 2013 paper on poultry trials that includes these considerations when determining experiments and sample size (replication) ( https://academic.oup.com/ps/article/92/9/2490/1562216).  They first list the following considerations when determining sample replication:  


"...specify the following:
   
   - i   
the significance level, α;

   - ii   
the power desired, 1 − β;

   - iii   
the number of treatments, k, to be investigated;

   - iv   
the minimum size of the difference to be detected between a pair of treatment means, as measured by Δ;

   - v   
the magnitude of the uncontrolled variation expected, σ   2.


In obtaining values for these, often the results from a previous experiment you or another researcher has run will be useful."     Note, none of these listed considerations listed are novel, except that the researcher may now also need to consider justification of numbers of animals (e.g., colonies) used.   

From our own institutional oversight, we've learned that PETA and other groups have forced researchers and  institutions to pay more attention to experimental design.  It's a two-edged sword.  The obvious issue is not using more animals than necessary.    The other issue though, that may not be as evident,  is that using too few animals may result in the power of the trial becoming too low to justify the experiment.  The animal protection folks were able to show that many animal trials lacked adequate replication. 

Using more animals than necessary to achieve sufficient power and confidence in the data should be avoided.  Conversely, if the power and size of the experiment is not likely to result in being able to show a difference, then it may be even more difficult to justify the animal trial. 

With respect to poultry, the authors concluded: "Fewer replicates will mean that there is less chance of detecting a difference of this size and that only larger differences are likely to be detected. An alternative is to look to reduce the error variability, for example, by improving the experimental protocols or the selection of more uniform birds."

It's a new age for animal scientists.  The stats for assessing power and size haven't changed.  The concern about not using too many or too few animals is new.  Using too many was seldom a problem - costs go up, labor goes up, and most of us don't have unlimited resources.  The point that PETA and others have made in the courts is  that inadequately replicated animal studies can't be justified  - that's something we haven't tended to consider.

 





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