Reply of Buzz Holling to a message from Marten Scheffer:

I did like your models (Scheffer and van Nes) that show that lumps of species emerge naturally from competition, augmented with predation (see post on self-organization of ecosystem lumpiness). I like also that the simulated lumpy distributions you demonstrated are very evolutionarily robust. That is what has been shown in so much of the multi species community data and experimental manipulations that have been published by various authors.

But let me offer some other empirical evidence that suggest something else may ALSO be operating. Then I’ll end with some words on the philosophy of science!

First, in one of my tests in my “lumpy paper” I asked if herbivores and carnivores had different lump patterns among the mammals in each of the communities of the boreal prairies and boreal forests (Holling, C.S. 1992. Cross-scale morphology, geometry and dynamics of ecosystems. Ecological Monographs. 62(4):447-502). In my mind, this was a crude, crude test of the role of competition and predation in forming body mass lump patterns. Herbivores and carnivores turned out to have the same lump patterns, suggesting that something more than competition must be causing the similarity.

Second, I asked if mammals compared to birds in the same communities (boreal forests and boreal prairies) had different or similar lump patterns. It turned out that the patterns differed when compared by untransformed masses, but were very similar when the bird masses were expressed as Mass to the third power and mammal masses as untransformed mass. Birds, it was argued, therefore must see their world in three dimensions, and mammals in one dimension. Some behavioral data seems to conform to that conclusion. But note the data and tests only applied to two systems, and clearly more ecosystems need to be compared. Adding bats, mammals that fly, offers a set of additional tests that would be very valuable. Data, however, are tough to get. Such tests need data on masses of birds, non flying mammals and bats from the same ecosystem communities.

Third , one of my colleagues, Jan Sendzimir, asked if body mass lump patterns among birds or mammals in different ecological communities around the world were more similar if they came from the same type of community. In essence, for example, are lump patterns in boreal forest communities in Europe and North America more similar to each other than to patterns among totally different communities? The answer was that they were more similar to each other, even when comparing different continents. That suggests that patterns on landscapes over a wide range of scales is central in defining lumpy patterns.

Fourth, Craig Allen has discovered a remarkably consistent pattern that the masses of endangered and invasive species of birds and mammals, exist disproportionately on the edge of the body mass lumps. He has shown that for at least five ecosystems around the world. It is as if crisis and opportunity are shared there, where, I suppose, the resources available at those scale breaks were poor. As scales increase, points are reached where the available resources suddenly change. (Think of viewing a high resolution satellite image of a landscape at different powers of ten from meters to tens of thousands of kms). Initially you see leaves and branches , and suddenly a scale is reached where you see trees and vegetation types- and on and on. What you see, I thought, is what you can , potentially eat or , better, utilize).

Finally, I am delighted with your emphasis on self organization. I have long felt that self organization , integrated with Darwinian evolution, has huge explanatory power, and the combination answers many of the existing puzzles of evolution.

But I also felt that the self organizing patterns were ones that involved not just interactions among biological species of similar trophic relations (like your competitors) but also interactions across trophic categories , and even interactions among organisms and abiotic features. Maybe, I thought, that is why anoxid, pre Cambrian organisms evolved the ability to use the dreadful poison of oxygen to explosively release a new burst of evolution. I imagined oxygen breathing species evolving in water where diffusion of oxygen was slow enough that local concentrations could be built up by organisms with facultative oxygen metabolism. Suddenly anoxic metabolism there becomes a penalty and oxygen metabolism a huge benefit. From those local concentrations, a world with oxygen could bloom.

There are two very distinct ways to suggest and test such ideas. One is via models, where very specific causes are designed and tested as a deductive excercise. That is what your paper does, in a significant expansion of that tradition. The other comes from a larger scale of testing of whole assemblies of potentially interacting variables . The conceptual ideas might come from earlier models. That was what my lump paper attempted as a test of a set of conclusions comparing earlier simulation models of ecosystems. Each of about a dozen models /studies indicated a small number of variables (less than five) at distinctly different scales explained those systems changes in variables. Hence I suggested, real data covering broad scales should show a small number of lumpy clusters. Body mass was chosen as a test because that is the data that are available. The data indeed showed the lumpy structure.

Now what is needed is slightly expanded deductive models with up to five variables/processes, much like those that Steve Carpenter and Buz Brock have structured for economic-ecological mini systems (see papers in Ecology and Society on lakes and fishing). And much more testing is needed with large scale assemblies of variables as I attempted in my lump paper. That is what Craig Allen has been doing, with collaborators, for a number of systems- biological, ecological , economic and social,. A book is about to be published, I believe. He might have some revealing comments that we all could learn from.

C.S.Holling