Do 'macrosystems ecologists' know about macroecology?

Paper by Levy et al. in Frontiers in Ecology and the Environment announces emergence of a new ecological discipline called macrosystems ecology (MSE). The authors define MSE like this:
MSE studies explore how broad-scale variation in fine-scale characteristics – such as organismal behavior and fitness, nutrient transformations, and water-use efficiency – relate to broad-scale spatial and temporal processes and patterns such as climate change, landscape alteration, and topography. Because MSE research questions are defined at fine-to-broad spatial and temporal scales, the data used to examine such questions must also be at such scales.

They also give "In a nutshell" summary of what they mean by macrosystems ecology:

  • Macrosystems ecology uses new approaches and applies existing methods in novel ways to study ecological processes interacting within and across scales
  • These approaches often include multiple scales, diverse data objects, data-intensive methods, cross-scale interactions, and hierarchical relationships
  • These studies require large volumes and diverse types of data from many sources, encouraging ecologists to build field and laboratory methods, database objects, and the data infrastructure capable of the joint analysis of multiple large data streams
  • Scientists use powerful statistical methods, such as Bayesian hierarchical models, machine learning, and simulations, to find and explain important patterns in complex, multi-scale datasets

In the whole paper there is not a single reference to macroecology. Yet it is macroecology that has been dealing precisely with the cross-scale ecological processes that the authors mention. Spatial scale and spatial scaling relationships have been the central topic of macroecology ever since it was established 20 years ago. See Rosenzweig's and Brown's classics of the field, and Storch et al.'s recent treatment of scale in ecology.

But macroecology went further from simple scaling and pattern recognition. West, Brown & Enquist's Metabolic theory of ecology, Hubbell's Unified neutral theory of biodiversity or Harte's Maximum entropy theory of ecology are examples of theories that describe processes and/or make predictions across scales and across levels of life organisation (some of the predictions suck, I admit). Plus, there are hundreds of macroecological studies that haven't brought such stand-alone theories, yet their contribution has been substantial. There are even macroecological journals, some of them being among the most influential in the fields of ecology and physical geography. How could Levy et al. miss this?

Maybe it is the stated character of the data, i.e. the diverse data objects, complex and multi-scale datasets and data-intensive methods', that makes MSE distinct from macroecology. Well, I am afraid that this is just another manifestation of the 'big data' buzzword frenzy that is all around right now. The problem is that, without good theory, any 'big data research' will reduce to data mining, exploratory analysis and pattern description. Nothing wrong with that, but we had plenty of it 15-5 years ago in macroecology, and the outcome is sobering: we now have plenty of data and heaps of patterns, but we often don't know what do they mean and how they came to be.

One of the consequences is that the old-school macroecologists have also turned to Bayesian inference and hierarchical models (similarly to Levy et al.), and to methods such as path analysis, structural equation models and so on. But, regardless to the amazing size of the data, any such analysis is only as good as the model or theory that is behind it. People do wonders even with relatively small data when they have a good theory. Macroecology still needs more good theory. In the case of MSE, Levy et al. do not point out any notable theory (I suspect that there is none yet).

A note: hierarchical models really do not work very well as a data exploratory tool -- those who tried know.

To conclude: The Levy et al.'s paper is an interesting probe into desires of a bunch of academics that got high on the 'big data' dope (it is all around) and now they would like to make everything big and hierarchical and cross-scale and integrated. I can just wave to them from the distance and shout: "Hey guys, why don't you come over and join our party!? We are getting high on macroecology here, it is a bit messy but there are some valuable lessons for you, it will save you some time. And while you are here, maybe we can help each other with the theory because we also lack it!"


2 thoughts on “Do 'macrosystems ecologists' know about macroecology?

  1. I was just contacted by Ofir Levy, and he pointed me to a twin article by Heffernan et al. in the same issue of Frontiers:

    I embarrassingly missed it (even though it is mentioned in the Levy et al. paper). So I get it now: Levy et al. deal with the methodical stuff, and Heffernan et al. deal with the conceptual underpinnings. Moreover, macroecology is actually briefly mentioned in the second paper.

    Yet I am still dissatisfied with the definitions presented in the second paper. It is still not very clear how the new discipline differs from what we already have in macroecology, and it still does not go beyond the notion that "we study everything that is hierarchical and multi-scale and complex".

  2. I skimmed the papers a few days ago. Seemed to me that in their view MSE is supposed to connect macro to processes at smaller scales ... in that respect, it would be somewhat different from mainstream macroecology, where many have voiced the view that they don't find it worthwhile to include micro processes for explaining patterns at the macro scale (but may well be that the field has moved on already).

    The paper left me mostly wondering about the difference between a macrosystem and an ecosystem, and accordingly between MSE and ecosystem science / models.

Leave a Reply

Your email address will not be published. Required fields are marked *

You may use these HTML tags and attributes: <a href="" title=""> <abbr title=""> <acronym title=""> <b> <blockquote cite=""> <cite> <code> <del datetime=""> <em> <i> <q cite=""> <strike> <strong>