Ecologists should quit making things hard for themselves and focus more on model systems

A model system is one that has features which make it a particularly good system in which to address whatever question the investigator has posed. That can be for a couple of reasons. One is that the system has features that make it easy to collect needed data. For instance, that’s why small organisms with short generation times are a good model system for long-term studies of population dynamics. Another is that there’s lots of relevant background knowledge, on which you can build in order to ask questions that couldn’t be asked without that background knowledge. For instance, Drosophila melanogaster is a model system for modern-day genetics and genomics not just because they’re easily and quickly reared in large numbers, but because Thomas Hunt Morgan used them to study inheritance over a century ago. Morgan’s work was built on by others, who couldn’t have done their work had Morgan not done his. And others built on that work to do work that they couldn’t otherwise have done, and so on. Perhaps the most powerful model systems are those that are good for asking many different sorts of questions, so that the answers to those questions can be linked together.

Of course, if you’re going to work in a model system, you’re obliged to first decide what question you want to ask, and then choose the model system that lets you answer that question. Or, if you insist on choosing your model system first, you’re obliged to restrict yourself to asking only those questions for which your chosen system is a good model system. For instance, protist microcosms aren’t a good model system for most questions to do with individual behavior, because protists don’t have much in the way of behavior and because it’s difficult to mark and track individuals.

That question-first, system-second approach is how I operate. It’s how workers in other fields operate. Developmental biology, genetics, and other fields of sub-organismal biology all are famous for their laser-like focus on a relatively small number of model systems. But it’s mostly not how ecologists operate. And before you say, “Well, ecologists can’t do that because ecology is all about variability and uniqueness,” you should recognize that our sister discipline of evolutionary ecology, which is just as concerned with variability as we are, increasingly focuses on “field model organisms” like threespine stickleback and anoles.

So ecologists could focus on model systems–but we mostly choose not to. Instead, what ecologists seem to do far too often for my taste is pick their system, then pick a question that in principle it would be interesting to address in that system, then cast about for some way to address that question in that system–often a partial, limited, flawed, or highly-indirect way. After all, if you choose your system first, how likely is it that that system will just so happen to be a good model system for whatever question you subsequently chose? And if pressed on this by some contrarian like me, ecologists too often respond by saying that their approach isn’t perfect but they’re doing the best they can within the constraints imposed by their system.

Which, as I’ve pointed out before, is like tying your own shoelaces together and then saying that it’s only feasible to walk slowly.

Look, there certainly are good reasons to choose one’s study system first. For instance, if there’s some rare species or threatened ecosystem that policymakers have decided ought to be conserved, then we have no choice but to go study that system as best we can. Or maybe it’s a question that’s only relevant in that system, so if we want to study that question we have no choice but to work in that system (although in that case, one might ask why you insist on trying to answer that question, rather than a more tractable question that could be asked in some other system) Or maybe it’s a question that’s only ever been asked in one sort of system, and for comparative purposes it’s really important for us to have some information from a contrasting system, even if that contrasting system makes it very difficult to get an answer. But if you’re doing system-first research for whatever reason, I think it is incumbent upon you to be up front about that. Don’t write or speak as if you’re doing question-first research, and then defend the quality of the answers you get by saying they’re the best answers you could get in your system.

Now before you say it, yes, focusing on model systems has a drawback: systems which make it easy or feasible to ask certain questions may provide a non-random sample of the answers to those questions. For instance, bacteria are a model system for experimental evolution, in large part because of their extremely short generation times. But bacteria also are asexual and have massive population sizes, and experiments on them tend to begin with no standing genetic variation. All of which might cause bacterial evolution experiments to provide results that aren’t representative of what would happen in sexual organisms with smaller population sizes and standing genetic variation. I suspect this argument is why so many ecologists are so reluctant to focus on model systems. “But that model system isn’t at all like my system! In my system, things might be different!” In response, I’d make three points.

First, in your system things might not be different. Just because a system has features that make it a good one for addressing some specific question doesn’t necessarily mean that the answers it provides will be biased. For instance, Thomas Hunt Morgan chose Drosophila melanogaster in part because it’s easy to rear lots of them–but why should that make Drosophila melanogaster any more or less likely than any other organism to exhibit Mendelian inheritance (a topic Morgan set out to study)?

Second, the putatively-unrepresentative features of model systems often can be manipulated, and so turned into testable hypotheses. For instance, you say protist microcosms are an unrepresentative model system in which to study population dynamics because they’re closed to immigration and emigration? Fine, open them up and test how immigration and emigration matter.

Third, and most importantly: Would you rather have a good answer to the question of interest (even an answer that might not generalize to all study systems), or no good answer? Isn’t some good answer, in some system, better than no good answer at all? Because that’s the choice you’re faced with when choosing between model systems and non-model systems to address any question of interest. Given the choice between a good answer from a model system, and a mediocre or crappy answer from a non-model system, I know which one I’d choose.*

I want to conclude on a positive note, by emphasizing the tremendous opportunities that ecologists are missing by not focusing more on model systems. Remember, one reason why model systems are model systems is because of cumulative accumulation of knowledge. The more you know about a system, the more, and more quantitative, and more sophisticated, and more difficult questions you can ask. Especially in model systems that are good models for more than one sort of question. Lately on this blog we’ve been talking a lot (and will be talking further) about the challenges of “scaling up” and “scaling down” in ecology. Well, the best model systems are ones in which it’s easy to work at multiple scales, thereby making it possible to link from one scale to another.

So what are, or should be, the big model systems in ecology? Systems in which we can easily and cheaply conduct measurements and manipulations on individuals, populations, communities, and ecosystems? Systems we can work with both in the field, and in the lab? Systems with relatively fast dynamics? Involving an organism about which we already know a lot, about everything from its genome sequence, to its individual physiology and life history, to its behavior, to its population dynamics, to its interactions with other organisms, to its biogeography, to its role in energy flow and material cycling? What do you think, class? Any ideas?

Meg: [raises hand] Oooh, I know, I know!

Me: [ignores her because she’s always first to raise her hand] C’mon, someone else. What’s the first system that comes to mind when you try to think of the ideal model system for ecology?

Meg: [half-stands up, raises hand so high she’s at risk of pulling a muscle]

Me: [sighs] Ok, Meg?

Meg: DAPHNIA!!!!!111!1!!!

Me: Yes. There’s no need to shout, but yes.

Yes, as my dynamic colleague Meg can no doubt attest, Daphnia, and the ponds they live in, are perhaps the ultimate model system for ecology, for all the reasons listed above. I do think some ecologists (including here at Calgary) are starting to treat Daphnia the way geneticists and evolutionary biologists treat Drosophila melanogaster. Which is awesome–they should keep doing that! They’re setting a great example for the rest of us.

Of course, ecology can’t live by one model system alone. Here are some other possibilities:

How about freshwater algae? (Oh, sorry Chris, didn’t see your hand up, were you about to suggest them?) As Chris can attest, we know an awful lot about their physiology, growth, reproduction, resource use, population dynamics, and biogeography, and about interspecific variation in all those things. We probably have, or could obtain without too much difficulty, genome sequences for many of them. They have fast generation times. They’re easy to grow and manipulate in the lab. They’re even of interest for biofuel production, so there’s an applied angle too. And while we can’t easily manipulate them directly in nature, we certainly can and do manipulate lots of biotic and abiotic factors that affect them. And of course, they’re consumed by Daphnia, making the two of them together a model predator-prey system.

Other ideas for key model systems on which ecologists ought to focus more? There’s the aforementioned evolutionary ecology models, threespine stickleback and anoles, of course. How about Arabidopsis? We don’t know tons about its field ecology yet–but that should be easy to rectify. Maybe aphids and their host plants and parasitoids? Maybe bacteria and phage? Bacteria and phage are great for asking questions at the interface of ecology and evolution, and already a model system in genetics and evolution. They would mostly be a lab-based model system for now, but I’m starting to see cool work coming out on their field ecology. Note that I would not suggest “protist microcosms”. They’re fabulous–but for too limited a range of questions to serve as a model system for a large body of ecologists, or for a large chunk of ecology (or maybe I’m just saying that in order to keep y’all from crowding into my system and competing with me).

So what do you think? Aren’t you a little bit sick of working your butt off to get partial, limited, flawed answers to our biggest questions? Wouldn’t you rather work your butt off and get a really complete, definitive answer for once? C’mon, take the plunge–start working in a model system! I’ll bet Meg will even send you a few Daphnia cultures to get you started. 😉

*Well, in an ideal world with no opportunity costs, I’d rather have both answers. I’d rather have as much relevant information as possible from as many systems as possible, including both good information from model systems and much-less-good-but-hopefully-better-than-nothing information from non-model systems. But of course, opportunity costs are ever-present.