In article ,
wrote:

That's interesting and worth knowing. Just curious, Nick, how did you
carry out this experimental test and was there statistical analysis?

To answer the last question first: "no". But I am a statistician,
and the results were enough to debunk the theory that ants are
always the primary cause of aphid infestations. To extend that
statement to "usually" would mean a much wider set of data and,
to extend it to "a significant" would need a lot more data.

My information gathering has been on the net, reading a random
collection of books and papers and so on.

The experiments:

Looking at the patterns on blackfly infestations on my beans
and in a fair number of other gardens. Flying females produce a
scattered (nearly Poisson process) distribution, concentrated
towards edges. Rain wash produces colonies on adjacent plants
lower down and younger. Ant distribution would spread to
adjacent plants, with younger on the outside - that is IMPORTANT,
as flying females might lay eggs on adjacent plants, too.

I have never seen evidence of ant distribution.

Leaving blackfly colonies to grow on my beans, and observing
the correlation of their growth rate with (a) whether they were
near and ant's nest and (b) whether I saw ants on them. Also,
seeing whether there are ants on EVERY colony in other locations,
or only on some.

I have seen no evidence of correlation, though I would have
noticed only a strong one (which is what I was looking for).


I can fully believe that there is SOME effect, but all of the
information I have is that it is negligible compared with other
factors. It is CERTAINLY so in my garden, and I have reason to
believe that it it is so generally. That includes statements
to the effect by entymologists.

Nobody that I know of has EVER carried out a test of causality,
so there is no definite statistical proof whether killing ants
has any effect on aphid infestations or not. All my experiments
indicate is that the claims of a strong and near-universal link
are bogus.

Given space (which I don't have) and time, it would be interesting
to do a proper experiment. This would be worth repeating in a few
parts of the country, because there could well be a geographic
distribution.

It is indicative that even the pesticide manufacturers do not
provide anything for killing ants in commercial crops - this
indicates to me that few scientists believe that this is significant
effect.


Regards,
Nick Maclaren.