John Lawton’s ‘Patterns in ecology‘ is a minor Oikos classic (cited 73 times since its publication in 1996). As he recognized, John was writing at a time of transition in ecology. Over the previous 15 or so years ecology (especially community ecology) had become increasingly dominated by small-scale field experiments. Many ecologists increasingly felt that this approach missed the big picture. Ricklefs and Schluter’s influential edited volume, Species Diversity in Ecological Communities: Historical and Geographical Perspectives, appeared in 1993. Jim Brown’s Macroecology, calling for ecologists to refocus on large-scale, universal patterns, appeared in 1995. That same year that the National Center for Ecological Analysis and Synthesis (NCEAS) was founded, with a mission to ‘advance the state of ecological knowledge through the search for general patterns and principles in existing data’. John’s article is a nice compact argument in favor of this shift in emphasis from ‘microecology’ to macroecology. The shift took hold, big time–NCEAS in particular was a stunning success, and spawned numerous imitators in various fields. So now seems like a good time to revisit John’s argument, because I think its strengths and weaknesses are both reflected in much of the work which followed it.
The core of John’s article is a very creative and compelling analogy–from which I think John draws a conclusion that’s only half right. Here it is in John’s own words:
‘Imagine you are a fairy, a bit larger than an atom, sitting in a world made up of a mixture of gasses. You would see balls (molecules) of different sizes and colours whizzing about, occasionally colliding, and sometimes combining to yield different sized balls. Being a curious fairy, with an experimental bent, you attempt to manipulate the molecules by building fences in a very local part of this imaginary world, to keep out the big red balls which appear to be attacking the smaller blue ones. The grant application to the king of the fairies says that you wish to: ‘Understand and predict the role of large red balls in structuring the assemblage of other balls.’ Does it sound familiar?‘
What’s right about this argument is that the fairy, the ‘microecologist’, absolutely misses a lot. This gaseous world has macroscopic properties–for instance, temperature, density, and pressure–of which the fairy is not even aware. These macroscopic properties emerge from the essentially random behavior of a large number of individual interacting entities. It’s a compelling analogy for how large-scale patterns in the distribution and abundance of organisms emerge from the essentially random behavior of large numbers of interacting individual organisms.
But what’s wrong with this argument is that it suggests that the macroscopic features of the system are the only features of the system worth paying attention to. I want to undermine this, not by rejecting John’s gaseous analogy or offering a contrary analogy of my own, but by taking John’s analogy more seriously than I think he himself takes it.
In physics, the ideal gas law describes the relationship between the temperature, pressure, and volume of an idealized gas. It’s a good approximation to the behavior of real gases under a wide range of conditions, and so can be regarded as a general, macroscopic pattern in the behavior of gases. This is precisely the sort of macroscopic pattern that John wants ecologists to focus on documenting and explaining. But how do physicists explain the ideal gas law? By appeal to the microscopic properties of gases. Specifically, the ideal gas law is derivable from first principles using the kinetic theory of gases (a theory about how individual gas molecules move and collide with one another), under various simplifying assumptions (e.g., gas molecules are point masses with no volume and undergo only elastic collisions). Importantly, kinetic theory has nothing to do with trying to predict the random trajectories of particular gas molecules, as John’s fairy futilely proposes to attempt. Kinetic theory is all about scaling up: it’s about rigorously deriving the macroscopic consequences of the collective behavior of a large number of microscopic particles. You say you want to focus on the big picture that emerges when one averages away random, microscopic noise? Kinetic theory tells you precisely how to do that, and what the resulting picture looks like.
Further, the approximations used to derive the ideal gas law can be relaxed, allowing derivation of more refined laws which account for subtle microscopic effects omitted from the ideal gas law. For instance, if you incorporate the facts that molecules have a nonzero volume (which matters when the volume occupied by the gas is small), and that they’re attracted to one another, you can derive the Van der Waals equation, a more precise version of the ideal gas law for which Johannes Van der Waals received the Nobel Prize for Physics in 1910. Which just goes to show that, just because the microscopic is random doesn’t mean its details don’t matter.
So if explaining (as opposed to merely describing) large scale patterns is our goal, the example of physics suggests, not that we should ignore the microscopic, but that we should focus on the microscopic–specifically, how to scale up from the microscopic to the macroscopic. If we’re serious about taking the ideal gas law as our model of how to do ecology, then we need an ecological equivalent of kinetic theory on which to base rigorous derivations.
This point of view isn’t original to me (see, e.g., Brian Maurer’s Untangling Ecological Complexity: The Macroscopic Perspective and Lotka’s classic Elements of Mathematical Biology). And I think it’s consistent with the recent history of macroecological research. The macroecological advances that have been most influential, and that have spurred the most productive work, have been advances in microscale theories from which rigorous predictions of macroscale patterns have been derived. Think of the use of detailed metabolic optimization models to predict the 3/4-power scaling of metabolic rate and body size (West et al. 1997). Or think of neutral theory, especially its use to predict the form of the species-abundance distribution. We now know, on the basis of very general arguments, what kinds of microscale assumptions about per-capita birth and death rates can reproduce the commonly-observed forms of the species abundance distribution. It turns out that the class of empirically-viable microscale models is very large–almost any stochastic model in which species are equally fit on average will work. That includes models with and without frequency dependence. This ‘many-to-one’ mapping from microscale processes to macroscale pattern is itself a satisfying explanation for why species-abundance distributions take the shapes they do–there’s no strong reason for them to take on other shapes. Work on John’s favorite macroecological pattern, linear local-regional richness relationships, followed a similar path. The hope of John and others that linear local-regional richness relationships could be used to infer something about the strength of local species interactions (i.e. that they’re weak) were dashed by microscale models, and empirical work directly linking microscale experiments and macroscale patterns. This work, some of it published in Oikos, showed that linear local-regional richness relationships arise from many different combinations of microscale processes.
John’s article concludes with a call for a productive pluralism in our research approaches, because those various approaches complement one another. I heartily agree, and I’m cautiously optimistic that such pluralism is the way of the future. The pendulum swing towards ‘pattern-oriented’ research that began in the mid-90s seems not, as far as I can tell, to have swung so far as to devalue small-scale experiments (in part because NCEAS has supported syntheses of experimental as well as observational data, and also supported theoretical work). With any luck, in years hence calls for future research in ecology won’t need to attack current approaches, and will instead call for ‘more of the same’!
Pingback: Are some general ecological concepts TOO general?: revisiting an Oikos classic « Oikos Blog
Pingback: It’s the end of NCEAS as we know it (and I feel fine) « Oikos Blog
Great, thought provoking, post. A couple of quick thoughts.
1. It seems like you’re basically equating macroecology with pattern and microecology with process, hence the emphasis on the process aspects of WBE and Neutral theory as being “micro” (and the explicit statement in the preceding paragraph). I think this is a common perspective, but one that I (and probably most folks who would describe themselves as macroecologists) disagree with. There is nothing in the definitions or approaches of macroecology that avoids/excludes processes (including very local scale individual level interactions). Since the two ecological models you mention (and the physical ones as well) are focused on describing emergent statistical patterns I think of them as being macroecological models. The fact that they include processes operating at the individual and/or species level doesn’t change this, just like including biological processes in a climate model doesn’t change the fact that it’s a climate model.
2. I think the fact that we still think and talk in dichotomies like this is a sign that we are a long way from truly accomplishing the pluralism that Lawton envisioned. Papers like Paine (2010) and Ricklefs (2008) make this clear. Folks tend to feel like they’re competing with other fields, and historically there are clearly cases where fields were discriminated against (macroecology for one), but we really need to get past this and focus on supporting and rewarding the best science regardless of (largely imaginary) disciplinary and methodological boundaries.
3. If we’re going to talk about dichotomies and pendulums, then I’d love to see an analysis of the proportions of “macro” vs. “micro” or “synthetic” vs. “new data” in the general ecology journals (and funded grants). There’s no doubt that macro/synthetic are trending up, but I suspect they’ve got a long way to go to catch up. Plus, the paper could be titled “The macroecology of ecology”; you know, since it would just be pattern based and all :).
Again, great piece. This is exactly the sort of thing that we should be discussing in the blogosphere. The Oikos blog may be the new kid on the block, but right now it’s putting the rest of us to shame in terms of quality, thought provoking posts.
Glad you liked the post Ethan. I was hoping that I’d draw some comments from proper macroecologists. In response to your thoughts:
1. If you say that macroecology today is process oriented, and is all about linking processes and their emergent statistical consequences in rigorous ways, then I think that’s great. And if you say the difference between macroecologists and microecologists is merely that they’re focused on different consequences of the same processes, I’m happy to agree. What I have a problem with is a focus on pattern (at any scale) to the exclusion of underlying processes, or a focus on pattern (at any scale) accompanied by mere hand waving about the underlying processes, or (worst) using observations of pattern (at any scale) to make bad inferences about the underlying processes (e.g., ‘a neutral model can reproduce observed species abundance distributions, therefore the underlying population dynamics must be neutral’, or ‘the local-regional richness relationship is linear, therefore local species interactions must be weak’).
Having said that, I do wonder if there isn’t still a useful distinction to be drawn between ‘pattern first’ and ‘process first’ people. By that I mean, is your research question ‘What are the processes that generated this observed pattern?’ or ‘What are the consequences of this particular process or processes?’ That is, do you start with the pattern and try to work out the processes, or do you start with the processes and try to work out the pattern? I’m very much a process-first guy, but not because I think pattern-first people are somehow doing it wrong. Might be interesting to think about the advantages and disadvantages of pattern-first vs. process-first research.
2. Post-NCEAS, I don’t think macroecologists have to worry about discrimination, not just because NCEAS encouraged people to think about large-scale patterns but also because NCEAS encouraged collaboration, and encouraged synthesis of different lines of evidence from different subdisciplines. Which isn’t to say there won’t always be some tensions. Not everybody does science the same way, and everyone naturally thinks that their way of doing science is awesome and deserving of more support than it receives.
3. The proportion of review papers in the literature is certainly going up. Unfortunately, it’s going up at least in part for reasons that have nothing to do with increasing prevalence of ‘synthetic’ science, but instead have to do with journals chasing higher impact factors.
I’m flattered and gratified that you’re liking the posts so much. Just trying to do my bit; some of my fellow editors should be adding their voices soon. Right now, I’m basically just doing a brain dump of all the ideas I’ve ever had since about 1995. Once my head is empty–which could be any day now!–I’ll need to think of some new ideas. I’m having a lot of fun with this right now, but the sustainability, at least in terms of my pace of posting, is an open question.
Jeremy – a couple of follow up thoughts.
1a. Yes, that’s how the best macroecology works, and even when we fall short of it that what everyone’s shooting for. I think if we’re going to talk about fields as a whole we need to focus on the best the field has to offer (though if you’d prefer I am more than capable of going on at some length about the rather substantial short comings of the average experimental ecology paper when it comes of furthering general scientific knowledge).
1b. Yep, there are weaknesses in the causal inference of observational studies, but we’ve succeeded in understanding things like gravity, black holes, and astrophysics in general with observational data. It’s really all about figuring out how to look at it.
1c. I think the whole induction vs. deduction is overplayed. Science doesn’t work if we only start from first principles and ignore observations, and it also doesn’t work if we only catalog pattern and don’t seriously work on process. Sometimes it might take a lot of one before we can get to the other, but I certainly hope we all agree that they are both critical parts of doing science.
2. I think you’d be surprised, but my point was there are lots of folks out there right now who don’t like, or at least doubt the efficacy of, entire fields of ecology, and we’d be a lot better off if we could get beyond that. Broader training anyone?
3. I totally agree. I feel like I even read something about that somewhere (http://library.queensu.ca/ojs/index.php/IEE/article/view/2386) :).
1a. I’d actually be more interested to hear you go on at length about the substantial shortcomings of the average macroecology paper. Similarly, I suspect I’d have much more interesting things to say about the shortcomings of the average microcosm paper than the shortcomings of the average paper in a subfield far outside my own. I spent the fall in Ottawa working with Rees Kassen, who’s an experimental evolution guy, and it was really interesting to hear his ‘insider’s’ take on the strengths and weaknesses of various microbial experimental evolution papers. By no means were his thoughts all about picky technical details….
1b. Sure. Although when macroecologists raise those examples I always find myself wishing they’d keep following that line of thought. What is it about, say, astrophysics, that’s made it a successful science able to make reliable causal inferences without experimental data? Can macroecology emulate (or to what extent can macroecology emulate) the astrophysics model? I don’t mean this as a rhetorical question, but as an honest question to which I don’t know the answer (not knowing enough about the nitty-gritty details of how astrophysics is done). Similarly, I wish John Lawton had kept following his own line of thought about the ideal gas law.
2. I’m all in favor of broader training. I feel like my liberal arts background, with little in the way of specialized ecology or stats training until I hit grad school, is reflected in my approach to science (and this blog!). (Insert your own joke about how I’ve just admitted to being a dilettante here). Having said that, I do think it’s possible for entire fields or subfields of science to go off the rails, or at least have significant blind spots or shortcomings that are only evident to outsiders. I think broader training would not only help us appreciate the strengths of different subfields and approaches, but would also help us better recognize those (hopefully rare) cases when an entire approach or subfield really does need to be killed off. Broader training doesn’t necessarily mean we’ll all agree with each other–but what it would (hopefully) mean is that we’d disagree with each other in more productive ways.
I’m not surprised that there are individuals who just hate macroecology. Every subfield has some haters. I’d only be surprised if their numbers were substantial (and no, don’t ask me how many ‘substantial’ is) and not declining. Interestingly, I haven’t had a review from a microcosm hater in years. I thought they’d become extremely rare, but after conversations with some of my microcosmologist colleagues I think it’s possible I’ve merely been getting lucky with my reviewers.
3. Yes, I’ve seen that.
Pingback: Ignorance is bliss (sometimes) « Oikos Blog
Pingback: Species pools and the fallacy of composition (UPDATED) « Oikos Blog
Pingback: Over at Sociobi… « Enquist Laboratory news blog
Pingback: Drilling down vs. scaling up « Oikos Blog
Pingback: Is macroecology like astronomy? « Oikos Blog
Pingback: Are feedback loops ubiquitous in ecology and evolution? | Dynamic Ecology
Pingback: ESA 2012: Wednesday highlights | Dynamic Ecology
Pingback: Scaling up is hard to do: a response | Dynamic Ecology
Pingback: Book review: Community Ecology by Gary Mittelbach, and Community Ecology by Peter Morin | Dynamic Ecology