The Trouble with Models

Today I’m going to talk about something that has proven to be both a blessing and a curse to fisheries management.  I’m going to talk about modeling. 

 
Tasteful humor.  From Sports Illustrated.
Unfortunately we’re not talking about the merits of the 2010 Swimsuit Calender (and allow me to be the millionth or so ecology nerd to make this kind of joke).  The models we’re talking about look a lot more like this:
 
From mdsg.umd.edu
Though a network modeler might see a thing of beauty in the latter image, for most of us at first glance it looks like a set of stereo instructions gone horribly wrong.  That said, the idea behind ecological modeling is actually a very simple one; you’re basically just tracing the flow of energy through an ecosystem.  Along the way it makes stops at each ecological component (also known as bacteria, plants, and animals, but can also include things like detritus) and at each stop some of it is burned off, some of it recycles back to lower levels of the food web (usually this means the organism dies and ends up in the detritus) and some of it moves along to the next stop.  The actual units of energy measured can vary from model to model, but the basic concept is generally the same, at least when applied to ecology.
The problems start to arise with the sheer amount of information that has to go into these things.  As you can probably imagine, any kind of accurate food web model has to have detailed data on feeding habits, metabolism, and life history for every single species within that system.  Often this is impossible given the amount of data published or collected, the limitations of the current technology, and the maintenance of any semblance of sanity for the researchers involved.  This requires certain abstractions to be built in and certain (sometimes huge) assumptions to be made.  Because of this, as one of my professors once said, “all models are wrong, but they’re useful.”
Probably the biggest assumption that has to be made is that the model is dealing with a discrete, closed system.  All the energy flowing in eventually equals all the energy flowing out.  This is obviously impossible in reality because sooner or later all ecosystems connect to at least one other ecosystem.  Some models try to get around this by having a certain amount of the energy being tracked bleed off into “export,” but are often vague about where they get this number.  
Another pitfall of the “closed system” approach comes from things like apex predators (you knew sharks were coming into this eventually).  These animals have high enough energy needs that they usually feed out of several smaller systems.  To include something like a sand tiger shark you’d need to either have a big enough ecosystem to encompass its whole range or figure out which fraction of its energy needs comes from eating out of your study area.  Highly migratory species present this same kind of problem, and it’s why in a lot of marine models the food web ends at predators like striped bass and bluefish, species that are high up on the food chain but not necessarily at the top.
Where things get interesting is when models are used to determine policy.  Ask any fisherman and they’ll give you an earful about the mysterious voodoo math that affects their ability to work.  They’ve got a valid complaint, and most fisheries managers are well-aware that they are making decisions that directly affect peoples’ lives based on a system that is “wrong, but useful.”  The arguments resulting from controversy over models can drive both fishermen and scientists completely nuts.  Unfortunately, the sheer size and complexity of the marine environment makes this sort of thing necessary.  The alternative is to continue managing one species a time, something that has arguably only made the fisheries situation worse.

One thing is certain, as long as there are models there will be a need for good, accurate, basic science to go into them.  Modeling has the potential to be a great tool for solving problems that until recently were simply on too large a scale to tackle, but it needs to be fueled by the best available data.  Only by having all that basic biological information on as many species as possible can models start to overcome their limitations.  Those of you afraid that field work is slowly being replaced by computers can rest easy; there is plenty of net-hauling and fish-counting to do.

So that’s my take on a rather dense and complex topic, and I fully acknowledge that it’s the opinion of a mere Master’s student just starting to get his field work off the ground.  Am I on the money?  Am I completely misunderstanding everything?  Was this whole discussion just an excuse to put a picture of a hot girl on the blog?

0 comments