Highly detailed models seldom work well. In a classic paper by Weigert he
showed that the best models have an intermediate level of complexity beyond
which performance gets worse. The interactions between closely related
species are so difficult to quantify that some degree of aggregation is
essential.
Galtman didn't specify what the purpose of the model was and what he
expected it to do, but for a long time people have modelled predator-prey
systems successfully. We did this by avoiding the use of parameters that we
couldn't measure. Sure the models don't answer every question, but so long
as they answer the questions we posed that was enough.
That doesn't mean that every question can be answered by modelling. Many
years ago we had a joint workshop with the fishing industry and they wanted
us to predict biomasses of cod and haddock. I pointed out that we had
already developed a model that predicted the total landings of cod+haddock,
but the two stocks varied unpredictably with inverse correlation so that we
could not break the total down into the two components. They were
dissatisfied with that, but that was the best we could do.
Bill Silvert
----- Original Message -----
From: "joseph gathman" <[email protected]>
To: <[email protected]>
Sent: Thursday, January 22, 2009 9:12 PM
Subject: [ECOLOG-L] Subject: Ecology Certainty Uncertainty
Illusion/Delusion, was Re: Ecological Modelling
That means, I suspect, that ecology is, in the view of literal believers,
> "doomed" to be "applied."
I don't know about that, but I DO think that what may well be doomed is
our ability to accurately model complex systems. In my own limited
attempt at modelling, I tried to model a simple predator-prey system, but
it quickly led me into a rat's nest of uncertainties. It would have taken
years to collect reliable data from field samples and experiments in order
to make the model realistic. And that was with an artificially simplified
model system. I can't imagine what it would have taken to build a "real"
model.
Even the climate modellers acknowledge that over the last decade, all
they've managed to do is confirm what their models CAN'T do (and likely
never will). Somebody should tell the IPCC that bit about models not
being predictive tools.
