A thumbs up for the intermediate disturbance hypothesis (guest post)

Note from Jeremy: This is a guest post from Douglas Sheil and David Burslem, authors of a recent letter to Trends in Ecology and Evolution (TREE) responding to my argument that the intermediate disturbance hypothesis (IDH) should be abandoned. They were kind enough to correspond with me before submitting their letter to TREE, leading to an amicable ongoing discussion which I’ve found valuable. I thought readers might find it valuable as well, so I invited them to write a guest post on Dynamic Ecology. I’m very pleased that they agreed, and I’m sure you will be too. They have a lot to say, touching on a range of larger issues that go well beyond the IDH. Thanks to both of them for taking the time to write this.

Just FYI: I’m probably not going to comment on this post, unless someone specifically asks me to do so (and maybe not even then). I’ve already said a lot about the IDH, so I’m just going to read and think about what Douglas, David, and the commenters have to say. If in future I have some things to say that haven’t already been said, I’ll write my own post.


Thanks to Jeremy for the invitation to provide more context for our views concerning the intermediate disturbance hypothesis (IDH).  The letters at TREE appear to be open access (here and here) so we won’t repeat them (and Jeremy already made a summary) We’re keen to hear other views too.

Initially we didn’t see a need to respond to Jeremy’s article.  We agreed with most of what he was saying, even if not the headline idea that the IDH theory should be “abandoned”.  We reckoned that the abandonment proposal was just a punchy way to grab attention.  Later, on reading Jeremy’s blog we realised he really intended us to discard the theory – and that those who continued to refer to it risked disapproval.  So we decided we should respond.

We don’t dispute the common confusion of disturbance-diversity ideas.  We had an extended commentary piece a decade earlier (also in TREE) that also attempted to highlight and address some muddles and contradictions in tropical forest studies. We have published various studies and commentaries on closely related topics that in our view would support a careful and well defined use of the IDH in the sense that Connell originally described it (see list at end of post).

We have always considered the IDH to be a theory about succession – the idea that it includes a broader set of non-successional ideas and mechanisms doesn’t match our experience.  This may reflect our own biases: we’re both terrestrial plant community ecologists who have focused on tropical forests.  But we suspect (from our biased sampling) that others who work on different systems might agree.  When we (DB) teach the IDH to students we use Wayne Sousa’s work on rocky intertidal communities (see here and here; it’s based on elegant observational and experimental work).

Connell’s key example was a study of a forest in Uganda based on tree plots established in the 1930s and 1940s.  One of us (DS) relocated and re-measured some of these plots in the 1990s and located and read a lot of the old literature about this site and study in the process.  People had recognised since the 1930s that the forest was spreading into the surrounding woodlands and thus younger forest occurred on the edges and older forest in the interior.  They realised that the tree community initially became richer and then more species poor as the succession progressed.  They also realised that if the old-growth forest was disturbed other species would establish (they were keen to do this as the better timber trees were associated with early to mid-succession, the timbers of the late succession species though durable were dense and difficult to work).  The ecological understanding of these patterns was illustrated and written up by Joe Eggeling, a forester and botanist who established a series of plots (nearly 2 ha each) to provide data.  He published his study after the second-world war (1947), and Joe Connell used this to illustrate the IDH.  Anybody interested in the history of IDH ideas may need to go back and look at those old Ugandan observations and records (they pre-date other claims we have seen, e.g. if you can find it see “Harris, C.M. (1934) Provisional Working Plan Report for the Bunyoro Forests, Uganda. Uganda Forest Department”).  These forest managers were sharp-eyed researchers and their observations remain as valuable and as worthy of recognition as anyone else’s.

As Jeremy says, the IDH was presented as a verbal theory and that lacks the kind of precision that many might want now.  But that is true of much older theory, and in any case, the most powerful part of Connell’s presentation was the illustration: the figure, with its hump-backed unimodal curve makes clear that the purpose was to link and synthesise a complex set of issues.  Evaluating such a theory requires careful attention to the assumptions, predictions and methods.

A question we didn’t explore in our recent TREE comment, due to word limits and our own uncertainties, is when it is appropriate to “abandon” a theory and what that implies.  It’s worth thinking about.  All major ecological theories are misunderstood, misapplied and mis-explained from time to time.  If the IDH were discarded on this basis, and we want a consistent approach, then presumably many other theories would have to go too (Darwin and Wallace’s theory of evolution comes to mind).  And what do we refer to after that when we see a rise and fall in species richness over a successional series?  What are we actually suggesting?  Are theories with valid elements and applications ever “abandoned”?  It seems simpler to highlight the problems, suggest fixes and refinements, and encourage precise approaches.  Imprecise theories tend to fall out of use when challenged with empirical data, not because they are abandoned –but because they are refined and evolve, they don’t become extinct.

Alternatively, if the IDH, as its name implies, is genuinely a hypothesis and not a theory (we suspect it is both … but let’s leave that aside here), then it is subject to testing that would allow it to be falsified, or (more likely) the scope of its relevance to be sharpened based on the ecological contexts in which it is supported. By extension, if the IDH is genuinely a hypothesis then it cannot be abandoned – it is simply falsified sufficiently often that the weight of evidence against it becomes overwhelming. But this subject runs into deep semantic waters (what are falsification, rejection, revision etc. and should we necessarily expect consensus on when such choices are justified, and is such consensus even desirable?) – it’d be interesting to hear what others think on that.  Do any philosophers of science read this blog?

The realisation that disturbance can lead to increased species richness is crucially important in a world where such species richness is often used to guide conservation choices.  For example, believers in the IDH should not be surprised that plots in logged over forests are frequently richer in species than unlogged forests, and this perspective allows us to recognise that 1) this is a transient state and 2) the extra species are often (but not always) “weedy” species of lower conservation significance than the species they replace — see e.g. here.  In different contexts, grazing is often used to maintain a high diversity of forbs in grassland communities that would otherwise become dominated by competitively superior grasses, and management is applied because the additional species are valued for conservation. These widespread practices acknowledge that disturbance is a natural property of plant communities and that local scale, short-term stability is neither desirable nor something that conservation managers should seek to impose. As an aside, the prevalence of practices that manipulate disturbance regimes (grazing, fire etc) to enhance the diversity of terrestrial plant communities provides anecdotal support for the idea that conservation managers think they achieve this impact (we recognise that understanding such effects requires much more than Connell’s IDH).

One issue of key importance is to identify the kinds of data that might represent a valid test of the IDH (or of its underlying assumptions and requirements). Spatial scale is fundamental because the IDH suggests that sampling should match the spatial scale of the disturbance regime. Successional change also has an implied temporal dimension, and appropriate scales may not be intuitive or easily studied. Pollen core records capturing community composition over long time sequences and containing evidence of disturbance (such as charcoal) may be a source of data for testing large-scale IDH predictions if data of appropriate species-level resolution and sufficient spatial replication can be found – are there any useful studies out there?

Anyway, that’s a few thoughts to get things going.  You know what Jeremy thinks, and now what we think too, but what do you think? We look forward to hearing your thoughts,

-Douglas and David

Related papers from Douglas and David

Sheil 1999 J. Veg. Sci.

Sheil 2001 Plant Ecol.

Sheil 1999 Plant Ecol.

Sheil et al. 1999 Science (technical comment, with reply by Cannon et al.)

Bongers et al. 2009 Ecol. Lett.

Slik et al. 2008 Conserv. Biol.

Burslem and Whitmore 1999 J. Veg. Sci.

13 thoughts on “A thumbs up for the intermediate disturbance hypothesis (guest post)

  1. Your comment that highlights succession as key reflects my own thinking (see here https://dynamicecology.wordpress.com/2012/12/17/the-zombie-idh-achilles-tortoise/comment-page-1/#comment-9056 , but as I wrote there, I have no idea if this is/was representative for the view in the wider community.

    Succession or not, the IDH is a curious case of “ecological theory”. Imo the problem with the IDH is that it can nearly always be stretched and adjusted until it fits, which is also the reason why it’s practically impossible to reject it. One reason is that

    1) If we keep on INCREASING disturbances in any given system, there will likely be a point where (some) organisms can’t cope, so diversity goes down, check!

    2) If we DECREASE disturbances, we homogenize the environment, and as illustrated by the logging example, if we go down to small enough scales, we will probably find a scale where increasing environmental heterogeneity will increase diversity, check!

    Saying IDH refers only to the disturbance scale might seem a solution, but in practice it will often be impossible to define the disturbance scale (e.g. fire or pathogens typically have no fixed scale). Moreover, I’d be very interested whether disturbance increases diversity also at larger scales, both for ecological (do disturbances create stabilization?) and practical (conservation) reasons.

    Apart from the scale issue, it is unclear to me whether the IDH is a hypothesis about adaptation or stabilization. What I mean by that is that, in an environment with a given average disturbance regime, species will naturally adopt to this regime, so in any given region, one would expect diversity to peak at the most common (regionally different) disturbance conditions. On the other hand, if the IDH makes a statement about disturbances enabling succession to stabilize diversity, there should be a universal disturbance regime at which diversity peaks, assuming the process of succession works similar across regions. It seems to me the IDH has often been applied in the former, but interpreted in the latter sense. In reality, both effects might be confounded.

    So, I’m not sure whether it should be abandoned, but I think for the IDH to be useful, it must make more quantitative statements about how intermediate is defined, how the relationship scales, and ideally also about the mechanisms that are supposed to create the IDH pattern.

  2. Thanks Florian

    We seem to agree on most of this. Yes it can “nearly always be stretched and adjusted”. In tropical forest science we still remain uncertain whether succession might progress to lower diversity in the long-term (with low disturbance) … as we dont know how long we have to wait to observe that.

    It is striking to me that some ecologsts can consider it a truism (as you imply) while others feel it is wrong or has been falsified. I think that the key to reconciling these is to better clarify the contexts where the IDH should or should not operate (David and I tried to do that in the 2003 TREE paper) and to try and disentangle the multiple processes (as e.g. Shea, Chesson and now Jeremy have tried).

    I’m interested in your ” adaptation or stabilization” thoughts but I’m not sure quite how to view your suggestion that “to stabilize diversity, there should be a universal disturbance regime at which diversity peaks, assuming the process of succession works similar across regions”. Could you expand a little?

    Best wishes

    • Hi Douglas,

      regarding ”adaptation or stabilization”: what I mean is that if the mechanism behind the IDH is globally the same (e.g. succession), and disturbances are comparable in type, then the disturbance level that creates maximum stabilization through succession should be the same everywhere (potentially we would have to correct for organism size and life span and such alike). I haven’t systematically looked at this question, but if we take tropical trees for example, I’m not sure whether there is much evidence for a universal disturbance level where diversity peaks, it seems to me that there are regions with relatively low disturbance levels that nevertheless have very high diversity.

      Regarding “truisms” and such: I would say Jeremy’s paper was asking to abandon certain explanations / mechanisms that he equated with the IDH. I guess he would still agree that it’s interesting to look at the question whether disturbances / fluctuations are potential mechanisms for generating / maintaining diversity, and if there are such mechanisms, they will create something like an IDH peak, whether or not we call it IDH in the end (Jeremy, can you still hold back 😉 )?

      • Yes, there are mechanisms by which disturbance and environmental fluctuations can maintain diversity. Sometimes models of those mechanisms predict peaks in diversity at intermediate disturbance, sometimes not. See my TREE paper and various posts here for review.

      • Thanks Florian

        Interesting. I dont think we are close to having the data (even the methods) to address a question like that. I suspect that disturbance regimes are too heterogeneous in time and space to allow an easy contrast (maybe a series of islands or the like). Can give it more thought though …

        You say “it seems to me that there are regions with relatively low disturbance levels that nevertheless have very high diversity.” — yes that does seem to be the case. But there are two big queries related to that: 1) Are we able to judge that such regions are truly low disturbance (as new information comes to light we see droughts, fires etc. and major vegetation shifts have been commonplace, at low frequencym across the tropics)?. 2) if we look closely we often find localised areas of low diversity old-growth forests … can we explain why? Not yet.
        Of course it is also relatively clear that whatever the role of distrubance in these systems — it is not the only factor influencing/maintaing richness/coexistence.

        best wishes

  3. Pingback: Friday links: liberal arts ecologists, potato beetle bombs, good modelers “lie, cheat, and steal”, and more | Dynamic Ecology

  4. Pingback: It’s the Friday link dump! | BioDiverse Perspectives

  5. Two comments spring to mind when reading this post. Mind you, I’m not deeply into the IDH litterature, and these comments are coming from an ecologist with general familiarity of the IDH. I might state some things seen as obvious.

    First, in the pattern of higher followed by lower diversity during succession (e.g. at forest plot edges) the peak can be explained as a mixture of two different communities (the distict communities are simplification, but nevertheless). This should be specially true looking at tree species (as the original Ugandan data), where the intermediate “community” is retaining species that established in earlier successional stages (but cannot establish in more closed forests) but has also started to include old-forest species. This might match your original description of IDH as “..considered the IDH to be a theory about succession”, but I fail to see why disturbance should be seen as the direct cause of the pattern.

    Second, although similar to the first point, in the examples you mention, Magurran’s (and others) ideas that sampled communities are a mixture of “local” and “vagrant” species seem relevant (also that the abundance distributions differ between these groups). In intermediate successional stages (especially in border habitats between e.g. clearcut and old growth forest) the pool of vagrant species is probably larger than in the clearcut and old-growth respectively. Most likely, the relative abundances of vagrant species are also larger in the intermediate habitat (in relation to “true” successional species) due to mass-effects, as compared to the old-growth or clearcut. This “mechanism” should contribute to a higher species richness in the intermediate/successional habitat, but this is not due to the intermediate habitat supporting more species per se (defined as “local” species adapted to this “disturbed” habitat).

    In both my points above “disturbance” is not in any way the cause of the peaked diversity, at least not in a direct sense. To me, it therefore seems contraproductive to attach these patterns to a theory labelled “intermediate disturbance”. The name “intermediate disturbance” implies that diversity should peak when we in a continuous way dial up disturbance (e.g. by increasing grazing pressure – igoring the aspects of homogeniety of disturbance in both time and space). It also seems that landscapes with patchy disturbances are used as support for the IDH, when this (to me) is simply due to heterogeneous landscapes having more species than homogeneous landscapes (everything else equal). I realize some of these issues boils down to semantics (i.e. what is seen as “disturbance”), but the labelling can definately cause unnecessary misunderstandings.

    • Hi Tobias

      thanks for the views. I agree with much of what you are saying. These semantic issues can be viewed in multipe ways quite reasonabally — and that does little to clarify the core ideas. Connell commented on the spatial aspects a little but focused on the intensity, frequency and time-since aspects of the disturbance. There are a lot of good simulation studies that explore spatial patterns — it can be viewed as heterogenity of course, but also requires a dynamic interpretation (to be maintained).

      What I dont grasp in your reaction is how we would have any early successional species in these larger landscapes without a process/mechanism that we call “disturbance”. It does seem key if you consider these as “fugitive species” that will be eliminated by superior competitors if disturbance does not intervene.

      Take a look at our older longer TREE article for a fuller examination of these ideas – see labs.bio.unc.edu/Peet/courses/bio255_2005f/papers/SheilBurslem2003.pdf (not sure that pastes correctly – see third para’ of the blog for a working version).

      best wishes

      • Thank you for your comment Tobias. Fairly similar views to yours were expressed to me verbally by my former postdoc advisor Tim Whitmore in the mid 1990s, and this is what underpinned his view that the IDH was effectively a ‘truism’ derived from persistence of species of successional status in a heterogeneous landscape. However it’s hard to envisage any model of succession that doesn’t require disturbance to driver life history variation, therefore it seems reasonable (to me) to retain disturbance as a label for the idea.
        Best Wishes


  6. Pingback: Modelling cyclic populations: thoughts on the workshop | Dynamic Ecology

  7. Pingback: Are there Buddy Holly ideas? (or, has any line of ecological or evolutionary research ever been prematurely abandoned?) | Dynamic Ecology

  8. Pingback: Favorite good (or bad?) examples of “operationalizing” vague ecological concepts? | Dynamic Ecology

Leave a Comment

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s

This site uses Akismet to reduce spam. Learn how your comment data is processed.