I've been lurking on the side-lines. Much of this discussion can be
simplified:
1) For anyone 'musing' about pesticides and how to study, you need to
start with a proper background concerning variables, interactions,how
traditional assays have been conducted, as well as the historical and yearly
publications comparing pesticide toxicities, residual effects, field weathering,
etc. which no longer occurs since Universities have gotten out of the
pesticide testing for label registration and extension services no longer have
easy access to the data needed to provide the comparative data updates.
Go buy a copy of Carl Johansen and Dan Mayer's Pollinator protection: a
bee & pesticide handbook, 1990.
You can get it from Wicwas. Read all 200+ pages, then we can return to
the discussion/musings.
2) Laboratory studies are fine, but one must always remember, small
numbers of bees in a lab or not the same as the superorganism that a whole colony
represents.
I spent 20 years developing honey bees as pollution sentinels. In this
discipline, we distinguish between indicator species and monitor species.
We also talk about keystone species. An indicator species is characterized
by organisms such as some lichens that are extremely sensitive to some air
pollutants and virtually vanish as soon as an area is impacted. Also, many
aquatic organisms are used as indicator species - my favorite, the daphnia
immobolizatioon assay - in other words, the daphnia die.
A true monitoring species has to persist for enough time to be able to
measure a change. Hence, we used whole colonies of bees as pollution monitors
- using the bees to collect chemicals of interest, or using bee counters
that profiled forager bee flight activity and flagged any event that
decreased the number of bees returning. For a healthy colony, we expect daily
return rates to run about 94-96%. Obviously, some bees die each day of old
age, get splattered on windshields, become food for birds, etc. With
counters, we could detect events ranging from a slowly declining return rate (over
days/weeks), a drop to a lower value - returns in 80% rates often
indicated a low level toxic chemical in the environment, or a catastrophic event -
drops below 50%, such as a spray incident.
Honey bees represent one of the few species that is a keystone species
(because they are part of a bigger picture, the pollination syndrome), an
indicator (individual bees due to small body size may receive a high dose and
quickly indicate a response, whether behavioral, physiological, death, or
other assessment end point), and a monitor (the colony as a whole is
remarkably resilient and adaptive).
Lab trials are by definition usually limited to bees in a cage, small
numbers, rarely whole colonies. As such, one is likely to see indicator
responses. But, in the context of whole colonies, outdoors, the indicator
function become a subset of the system as a whole or may even disappear due to
compensatory mechanisms or simpy dilution (few bees out of tens of thousands).
3) For years, I've been surprised that many of my colleagues from the
social sciences argue vehemently that the social sciences are true science.
Given that many of our initial statistical programs became widely available
as a consequence of social studies, I've never had a problem considering the
social sciences a science.
But, the recent discussion underscores a really important issue, one that
I've seen increase dramatically with on-line publications and journals
managed by journalists, not experienced scientists. Here, I'm showing my age.
I'm used to a system of editorial oversight and peer-review, where a
seasoned scientist or group of scientists act as the editors. They imposed a
level of screening for objectivity, knowledge of the field, and appropriate
design. And, I realize the down side was that some of these editors had an
agenda or their own bias, but MOST were experienced. They knew the
difference between a well designed study, the basics of statistical analysis, and
most importantly the difference between Inductive and Deductive Reasoning.
A paper like the Harvard Imidaclprid study would not have made it past the
editors desk, much less be sent to review, and then get published. How
many fails did that represent?
Yet, I increasingly see authors and editors who do NOT understand this
difference between Inductive and Deductive Logic, which to a scientist is
critical - its a cornerstone of the scientific process. And, I'm amazed to see
so many reviewers who don't know how to properly set up and analyze
data from a statistical perspective, and don't recognize serious flaws when
they occur. My guess - us old timers had to manually punch in numbers, do
the math, when we learned our stats.
I started with rotary calculators and punch cards. But now, anyone can
grab an Excel spreadsheet or a canned stats program, crank in data, and get
an answer out the other end. Whether the appropriate data transformations,
replication, adequate degrees of freedom, proper stats were applied is any
one's guess.
My hat is off to Peter Borst and Randy - I don't have the energy or time to
enter into every discussion, but once in a while, something gets my
attention.
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
Guidelines for posting to BEE-L can be found at:
http://honeybeeworld.com/bee-l/guidelines.htm
|