> >I too am a little uncomfortable with the sampling method for varroa but
winter survival data indicates no varroa problem.

> > Unless we have something to compare it with, samples in the middle of
winter mean next to nothing. I assume he did this in March, though he
doesn't say.

Actually I did say, and we do have something to compare it to.  We have done
winter spot checks for years and years (although not exactly this way).
Everything is documented at http://www.internode.net/honeybee/diary/ and
easily located by a Google site search, but I suppose that is beside the
point.

The point is that our spot checks in October showed a maximum of 10 mites on
a 24+ hour Apistan drop test (from what I am able to find in my notes at
this moment).

We are merely double checking here, and watching for fluvalinate resistant
mites which, if present,  we would expect to have put a few hives -- at
least -- into the 100% infested region.  The major purpose of the checks was
to look for tracheal and to tally early winter loss.

> In my experience, this is too early to assess winter losses.

Winter is over in a few more days on March 20+/-, so it is not too early to
spot trends, and we can see what our losses from fall and most of the winter
are.  Spring loss is another matter, and spring weather is a huge factor in
further loss.  If we had seen huge winter loss, we would be expecting the
trend to continue into spring, barring unusually good weather in March and
April, but, so far, our losses seem on the good side of normal compared to
Feb checks in other years.  Having said that, though we just had our coldest
day of the year a few days ago, and a week of bitter weather.

> Without sampling the center of the cluster *and* the edge, we will never
know if sampling the edge is effective in detecting *any* levels of varroa.
Remember, his numbers were consistently zero. But that proves only that
there weren't varroa at the edge, which I wouldn't expect there would be. I
would expect them in the warm center of the cluster.

That is a very interesting hypothesis.  The fact that we found a varroa mite
on one of three samples from a neighbour's bees using the same sampling
method brings this into question -- in my mind at least.  Someone with
sufficient varroa should test this, but our levels (established by late
October Apistan drop board sampling) are so low that I could not do so.
Since our bees break cluster from time to time during winter, and we sampled
during mild weather, I should think that the bees would have been reasonably
well mixed.  How do you propose that varroa mites migrate to the centre of
the cluster?  By jumping from bee-to-bee? That is interesting to speculate,
and although it does not seem totally improbable to me, I have seen nothing
to substantiate the idea.

 > per apiary is statistically large enough (I would still opt for a sample
size around 250 -- how else can you detect levels as low as 1%?

Who cares about 1%?  Not me.

> If he is just looking for dead-outs and colonies with varroa levels of
30 -40%, he may find them this way. I would still rather sample 3 or 4 hives
(10% of 35) correctly and extrapolate the data from this.

At our recent research meeting in Calgary, it was pointed out that, in
Manitoba, a late October varrroa level of 12% was not considered fatal -- if
I understand correctly -- so if you are correct about only finding that a
few colonies have 30-40% varroa levels with the sampling we did, that would
be *exactly* what we are looking for -- and our test was then, in fact,
correctly designed.  This is assuming that a mid-February sample is
equivalent to a late-October sample, an assumption I think we can question,
but which should be reasonable close.

> Doing a correct sampling of 10% of the hives will give a pretty good idea
of the infestation rate of the apiary, in my opinion. It involves opening
fewer hives, too, although it involves killing more bees (750 to 1000).

And that is if you don't count the loss of hives that is proven to result
from such invasive action justified only by a mere hypothesis.

> Again, unless we are going to compare this to data we have collected
previously, or data someone else has compiled -- in itself, it won't mean
much unless the levels are high.

You've got it.  High levels is *precisely* what we were looking for.  We
don't care much about low levels and we are not doing academic work.  We are
doing IPM.  We are watching for levels that demand treatment soon.

> Then you better be thinking about what your treatment options are.

We are always thinking about that, and are currently planning spring Apistan
application.  However, with our present results, we are now thinking that we
may be able to consider softer options for this spring than anticipated --
subject to further testing and consultation.

allen
http://www.internode.net/honeybee/diary/