Succession is one of the first things that students learn about in ecology. Each intervening stage modifies the environment in such a way that lays the groundwork for the next stage, while making the environment less hospitable to its own offspring. Only the final stage is self-perpetuating and stable.

Frederic Clements, one of the pioneers of community ecology, saw ecological succession as an ontogenic process in which the community - a superorganism - developed into its final, mature form. The orderly progression from bare ground to mature forest is orderly, progressive…and very Victorian.

When Victorian science provides a picture of nature that is, in its very essence, Victorian, there’s good reason to re-think your models. And in the study of succession, people have done that. People working on succession tend to stress the role of chance, and see the system as cyclical - there is no stable “climax” community, there are no undisturbed forests. But old ideas die slowly. Textbooks still teach succession using a number of early studies which were based on the idea that you could substitute space for time, what ecologists call chronosequence studies.

In a paper published in the May issue of Ecology Letters, by Edward Johnson and Kiyoko Miyanishi¹ take a look at some of the classic succession studies, and came to the conclusion that “empirical evidence invalidates the chronosequence-based sequences inferred in these classic studies“. While this is not totally surprising, it's important to document.

Henry Chandler Cowles is one of the founders of the science of ecology. His work on the plant communities of the Lake Michigan sand dunes is one of the foundational studies in plant ecology, especially in the United States. Cowles was one of sixteen charter members of the Ecological Society of America, and went on to train the second generation of American ecologist including William S. Cooper.

Lake Michigan is of glacial origin. Over time, water levels receded; as you move inland from the lake shore you are looking at older and older sand dunes. The youngest dunes, along the lake shore are colonised by annual plants. The sand is loose and is able to drift around. Further inland the dunes are dominated by bunchgrasses like Ammophila. These grasses are deep rooting and able to grow rapidly if they get buried by sand. They stabilise the dunes. Older dunes support shrubs and fast-growing trees, even older ones support pines, and the oldest dunes have been colonised by oak forests.

Based on this chronosequence, Cowles was able to construct a successional sequence. The bunchgrasses stabilise the dunes and increase their organic content, allowing shrubs to colonise the area. The shrubs and cottonwood trees that colonise the stabilised dunes are able the shade out the grasses, but are later displaced by the pine trees that are able to establish. But pines themselves create too shady an environment for their own seedlings to establish, so as the pines age they are replaced not by their own offspring, but rather, by oak forests. By substituting space (distance from the lake shore) for time, it’s possible to construct this elegant, intuitively pleasing, progressive sequence of events. But as Johnson and Miyanishi point out, reality isn’t always quite so elegant.

Annual plants grow on unstablised dunes close to the lake shore, while bunchgrasses grow on stabilised dunes further inland. Since bunchgrasses help stablise the more inland dunes, it make sense that once they colonise dunes, they stabilise them. But this neglects the alternate explanation, that on unstable dunes, annual plants do better than perennials. According to the evidence presented by Johnson and Miyanishi, this explanation appears to be more probable. Similar questions exist about the other stages.

They go on to raise similar questions about a number of other classic ecological studies - hydrarch succession, which sees lakes as transient entities which are gradually filled in and converted to meadows, Cooper’s classic succession of succession at Glacier Bay, and oldfield succession in the North Carolina Piedmont. In each case, they raise substantial questions about the substitution of space for time when studying succession. They close by saying

In conclusion, we think it is time to recognize the overwhelming empirical evidence invalidating these classic chronosequence-based examples of succession and to stop using them in ecology textbooks and course curricula. It is also time to require ecological studies using chronosequences to provide strong tests of its critical assumptions rather than ignoring them or simply paying lip-service to them.

Johnson, E.A., Miyanishi, K. (2008). Testing the assumptions of chronosequences in succession. Ecology Letters, 11(5), 419-431. DOI: 10.1111/j.1461-0248.2008.01173.x