Model systems don’t work at teaching universities


Many research strategies, developed inside large research institutions, don’t work well in small teaching-centered institutions.

One of these strategies, I suggest, is the use of a biological model system.

What do I mean by model systems? Any system, used in multiple laboratories, that has been optimized for a broad variety of lab investigations. I’m thinking of zebrafish, Drosophila, mice, C. elegans, E. coli, Arabidopsis, and honey bees. For the ant people, this includes Temnothorax.

It might seem counterintuitive that model systems can prevent you from getting research done in a teaching-centered institution. After all, these model systems were developed, in part, to make research easier and remove methodological barriers experienced by those working in a broad variety of organisms.

Even though model organisms are easy to work with, have lots of standardized methods, and a massive bank of potential collaborators, the costs and risks of working with a model system far outweigh the benefits and the opportunities.

My initial thought on this topic came from observation. I don’t see much research on model systems getting done within teaching institutions. I also have seen a variety of people, who work on model systems in teaching institutions, experience substantial challenges that have kept them from getting work done. Here’s a list of problems that I’ve seen crop up for those using model systems in teaching institutions.

  • The low-hanging fruit has been picked. In model systems, the things which are a combination of easy, obvious and trivial have already been done. That only leaves things which are either difficult, requiring particularly deep insight, or trivial.
  • Model systems have intense competition. In model systems, many labs compete rather than collaborate. A small lab in a teaching institution isn’t built to deal with this kind of competition. This doesn’t mean that you should avoid competition by working on minutia, but it also means that you should choose an avenue that has high pressure competition. As any ecologist will tell you, competition is an interaction which has a negative effect on both parties. (An economist will tell you the same thing, though it’s better for the consumer).
  • You might get scooped. The probability that someone is doing exactly what you are working on is higher in a model system.
  • Collaborating is difficult. The labs that run model systems typically have lots of routine methods happening like clockwork, run by technicians and students who have become specialized in the model system. If you’re working on this model system, you are not likely to be able to contribute anything of substance to a big lab, which could do the same thing much more quickly than you could offer. Collaboration is easier if you have a specific resource or skills that others want or need, and that’s harder from a teaching institution if you work in a model system.
  • You may become isolated. In the community of people working in model systems, if you have a small undergraduate lab you’re more likely to be overlooked by your peers. I’m not sure how or why this is the case, but it’s what I’ve observed on several occasions. Everyone I know who has worked on a model system wasn’t well integrated into the community of frenemies that works on the system. It’s hard to have a scientific impact if you are unrecognized by your peers.
  • The bar for evidence is very high. Peer review is more likely to be rough and demanding. People with model systems tend to be territorial, and especially if you’re not perceived as a competitor that can bite back with teeth, then at least one reviewer is likely to go the extra mile to find something wrong with what you’ve done.
  • The minimum publishable unit is greater. If you’re working in a model system, odds are that the big labs working in the field are able to heavily replicate their experiments, and sandwich multiple experiments together into a paper. This establishes a standard for a large volume of data in a paper. I’m not arguing that you should seek to publish as soon as you have enough data for any kind of publication. However, if you have a genuine, interesting, and cool finding and want to get it out for everybody to read, it’s a bummer to have others telling you that you that it’s ready for publication because you don’t have a mountain of data.
  • Maintaining colonies takes continuous effort. Mustering any kind of consistent student work in the lab is a success. You don’t want your success in long-term student research to be squandered on the maintenance of lab colonies of a model system. You want to work on a system in which students can dive in and collect actual data, rather than spending all of their time keeping critters fed, clean, reproductive and healthy. Also, you want to be free to disappear for long periods of time, and not chained to the lab to maintain your organisms.

So why would one ever work with a model organism? One great reason would be to provide lots of valuable hands-on laboratory experiences to students. But in my observation, this kind of approach is a poor recipe for completing original research in a teaching institution.

In my opinion, if we are giving students research training, that means we’re giving them research training. Which means that they’re doing research, and research is something that ends up in new findings that are shared with the scientific community. If you want to give a genuine research experience to students, then you need to regularly publish the work that you are doing. From my vantage point, it’s very difficult to do this while working with a model system in a teaching institution.

I have a couple other alternative explanations for the (perceived) phenomenon that researchers working on model systems in teaching institution have a harder time publishing their work. The first alternative is that I’m just wrong, and that my experiences or perceptions aren’t representative or accurate.

The second alternative is that less research is completed because of the use of strategies from R1 labs that don’t translate to a teaching institution. Most people at teaching institutions, who work on model systems, are continuing with their dissertation system, I think. The research approaches from a dissertation lab probably don’t translate well to a teaching institution. I haven’t spent much time inside the labs of teaching institution professors using model systems, so I don’t know how they run things on a day-to-day basis, and am not too sure about this.

Do you think that it’s hard to get research done on a model system at a teaching institution? Did you switch into one, or out of one, to get research done? Thinking about it?

12 thoughts on “Model systems don’t work at teaching universities

  1. I agree – but think a strategy is to mimic a model organism program where you can test know theories/results in a new format. No need to reinvent every wheel – just keep them rolling.

  2. I’d not thought about this before, but I think you’re right Terry. I’d add another reason: space. Teaching-focused institutions simply don’t have as much lab space available to house the large labs and associated technical support required for serious model system research.

  3. If you’re trying to compete head-to-head, model organism research in a teaching situation is probably not a good idea.

    But if you are not working on a model organism, how are you going to make your research relevant? Model organisms have the advantage of having a distinct community of fellow researchers. Being part of a research community helps, as you indicated. And if you think that you’re isolated for working on a model organism because you’re in a teaching institution, I still suspect you’re going to risk even more isolation if your work is on something obscure.

    As I’ve mentioned before (, my approach is to have a mix: some stuff related to model organisms, some that is not. I think it bears repeating: “Spend too much time chasing citations, and you do end up simply one of the pack, anonymous. Be too convinced that all knowledge is valuable, even in the face of evidence that nobody cares, and you end up as an oblivious loner.”

  4. Great points. It is really possible to work on something that isn’t a model organism, but isn’t obscure. It definitely helps to work on a question, or a taxon, that has a strong community of researchers. In my case, people who work on ants are all generally friendly with one another, and are friendly and helpful to one another. And the same can be said for the community of tropical biologists, in general. So I have a couple strong communities. But, I’m not working on any particular species – or any particular narrow question – that puts me in the same niche as bigger labs. And, in the rare case of some unintentional (comically so at times) niche overlap, I’ve had some pretty negative consequences.

  5. My own reaction is that there are various sorts of model systems. I think your concerns apply to many of them, but not to many others:

    I think some of the issues you raise aren’t so much issues of model system as issues of popular systems. “Model” and “popular” often go hand-in-hand, but not always. My own system, protist microcosms, is a model system, meaning that it’s a very convenient and tractable system in which to ask certain sorts of questions. But it’s not an especially popular system. So I don’t have too much difficulty carving out my own niche without going into self-imposed obscurity.

    Similarly, I think some of the other issues you raise apply to some model systems but not others. Protist stock cultures are easy to maintain, for instance. And because protist microcosms are super-cheap, super-easy to learn (no difficult techniques or fancy equipment required) and typical experiments can be done in 1-3 months, I’d think they’d be well-suited to a research program at a teaching institution. Indeed, for many years my own research program ran almost entirely on summer undergraduates and undergrad honors and independent study students.

    I think some of the other concerns you raise aren’t necessarily specific to model systems. For instance, the bar for evidence is always getting raised in all of science–as well it should.

  6. I’d have to disagree with your thesis, with a few qualifications. If we limit the discussion to using one of the major model systems in a “standard” way to answer the same questions that researcher at R1 institutions address, I agree that it won’t work at a teaching institution. But that is really just the larger issue that because we lack armies (or at least a small battalion) of grad students and postdocs, researchers at teaching institutions simply can’t keep the same pace. However, if your question is relevant and original, using appropriate model systems offer several advantages, including abundant published background material to guide your research, a community of readers, colleagues, and consultants with intimate knowledge of your study system (often from a different angle, which is great), and research infrastructure ranging from varied genetic strains, available antibodies, and established rearing methods, to whole genomes and transcriptomes.
    I am actually mostly a macroecologist, so my model system is mostly OPD (other people’s data), but I have recently started working with colleagues to use Manduca caterpillars as a model system for studying biological scaling (they grow ~10,000-fold in 3 weeks!). But many of the biologists in my department, and most of the really productive ones, use fairly common model organisms, including E. coli, Xenopus, Arabadopsis, Physcomitrella, as well as Manduca. I think that “it ain’t what you use, but the way that you use it.”

  7. I see your point. I think. (Is running these stock cultures something that you could do if you were teaching three or four classes at the same time, and do research with the cultures as well?) Regardless – this isn’t what I was referring to as a ‘classic model system.’ Having a model to represent something bigger or more broad is what we’re all into – just avoiding the same exact system as others could be better, I think.

    As for the bar for evidence – of course a high bar is important. Sometimes, though, a very compelling result – from a well designed experiment, in which all replicates are great representatives of the general statistical population and are randomly selected, and the probability value is incredibly low, and statistical power in the experiment is high – could be dismissed by reviewers because there weren’t enough samples. I think this would happen more in a model system, in which some labs can power through nonreplicable ‘significant results’ with low effect sizes by using larger sample sizes.

  8. Re: keeping stock cultures when I’m teaching 3-4 classes, yes, I still could. My stock cultures only get touched once/month or so, when you subculture them. Otherwise the protists fend for themselves.

  9. Huh! This goes against my observed trend, which is obviously a limited sample. I’d be curious to learn how the folks your department use model systems – whether they are doing something different than the large R1 labs do, or doing the same kinds of things on a smaller scale. I think at least some of the things that I’ve identified as challenges are real – though perhaps not all of them – and I wonder how they work.

  10. I think part of it is that they have found particular niches that are not already co-opted by R1 researchers. Some of them collaborate with R1’s to a greater or lesser extent as well (which I would like to hear you comment on in another post, if that interests you).

Comments are closed.