Ecological processes depend on …

14 05 2014
© Cagan Sekercioglu

© Cagan Sekercioglu

I have been known to say (ok – I say it all the time) that ecologists should never equivocate when speaking to the public. Whether it’s in a media release, blog post, television presentation or newspaper article, just stick to ‘yes’ or ‘no’. In other words, don’t qualify your answer with some horrid statistical statement (i.e., in 95% of cases …) or say something like “… but it really depends on …”. People don’t understand uncertainty – to most people, ‘uncertainty’ means “I don’t know” or worse, “I made it all up”.

But that’s only in the movies.

In real ‘ecological’ life, things are vastly different. It’s never as straightforward as ‘yes’ or ‘no’, because ecology is complex. There are times that I forget this important aspect when testing a new hypothesis with what seem like unequivocal data, but then reality always hits.

Our latest paper is the epitome of this emergent complexity from what started out as a fairly simple question using some amazing data. What makes birds change their range1? We looked at this question from a slightly different angle than had been done before because we had access to climate data, life-history data and most importantly, actual range change data. It’s that latter titbit that is typically missing from studies aiming to understand what drives species toward a particular fate; whether it’s a species distribution model predicting the future habitat suitability of some species as a function of climate change, or the past dynamics of some species related to its life history pace, most often the combined dynamics are missing.

Well over two years ago, a few of us (including Miguel Araújo, Regan Early, Barry Brook, Damien Fordham, Salvador Herrando-Pérez) met in a small village in the mountains north of Madrid to hash out a few papers. It was a most pleasant way to workshop, let me tell you. We were using the second-to-none UK birds database, which gives detailed area of occupancy data for 100s of species, and most importantly, how those have changed over many decades. After bashing out what I thought was a pretty clear story, we eventually submitted a paper, although after several rounds of review and the addition of quite a few more specialist authors (Cagan Sekercioglu, Phil Cassey, Steve Delean), both the analysis and answers became increasingly complex.

© Cagan Sekercioglu

© Cagan Sekercioglu

Our first challenge was how to model the direction of range change within the same framework – some species had contracted their range between the 1970s and 1990s, while others had expanded. The second challenge was accounting for phylogenetic non-independence (a staple assumption of linear models is that the underlying data are independent – evolution makes that impossible for individual species). Our third challenge was to find a way of characterising the configuration of a species’ range (i.e., degree of fragmentation) using a metric uncorrelated to total range size. Finally, we needed to account for non-linearities in response to climatic conditions like minimum winter temperature and annual precipitation. Phew!

Well, after many iterations of the analysis and manuscript, we are pleased to report the publication of the paper in Proceedings of the Royal Society BPredictors of contraction and expansion of area of occupancy for British birds.

Let’s try to run through the major discoveries without too much focus on that niggling ‘… but it depends on …” qualification.

  1. Interestingly, many species that had contracted their range were not locally threatened, nor were species that expanded their range universally considered safe.
  2. Also, the direction of change mattered less than the magnitude of change. In other words, our predictors generally had a similar effect regardless of the direction of the change.
  3. Climate (not weather) came out as the most important predictor of range dynamics, but the relative influence of each driver depended on where the species spends most of its time: northern species changed their range more in warmer and drier regions, whereas southern species changed more in colder and wetter environments.
  4. Species with slow life history (larger body size) tended to change their range more than species with faster life histories, whereas species with greater natal dispersal capacity resisted contraction and, counterintuitively, expansion.
  5. Counter-intuitively, we found that higher geographical fragmentation of a species’ range also increased expansion probability, possibly indicating a release from a previously limiting condition.
© Cagan Sekercioglu

© Cagan Sekercioglu

Well, that’s a bit to get one’s head around. I think the generalisable statement one can make from all this is that it’s all about context, context, context. Just when you expect a species to behave in a particular way to an environmental pressure, it might just surprise you that it does exactly the opposite.

We also have a few more caveats about generalisability (don’t we all):

  1. Britain has been extensively modified for centuries – we’re probably dealing with a fairly perturbation-resistant remnant avian biota, so the results might not apply to areas having recently undergone habitat change.
  2. Britain is also not the only place these species hang out – migratory species likely respond differently than sedentary ones (indeed, most of the world’s bird species are sedentary).
  3. Britain, like much of Europe and eastern USA, has had quite a lot of forest regrowth following agricultural abandonment over the last few decades. This likely affects some species differently than others.

Hopefully all this makes sense to you. I feel we’ve probably raised more questions than we’ve answered, but isn’t that what the pursuit of knowledge is all about? To me the main take-home message is that simple species distribution models predicting range change based on some future climate probably don’t capture all the nuances of what makes a species inhabit the range it currently does.

CJA Bradshaw

1Throughout this paper I’ll use the term change in ‘range’ to mean ‘area of occupancy’. Although semantically the latter is more correct, it’s easier to understand if we just talk about range change.

 


Actions

Information

3 responses

14 05 2014
Robert Lawrence

I have trouble with the concept of hypothesis testing having anything to do with what scientists actually do. All of your post is about analysing data and examining the strength of different effects in very complicated systems. Hypothesis testing seems a far too simplistic concept. I looked for examples in your post and I could not find any. Indeed, I have been looking since 1983 when I first read “What is this thing called science?” by A.F. Chalmers. I think we need a more accurate portrayal of science for the community to understand the process.

14 05 2014
CJAB

On the contrary, Robert – the paper was full of hypotheses, but multiple working hypotheses (not the bog-standard, unrealistic Neyman-Pearson hypothesis testing framework). I didn’t exactly put them down as such in the post, and admittedly, they evolved as the statistical analysis did, but it is still very much a hypothesis-driven analysis. I suggest you read the paper.

14 05 2014
CJAB

Well – PRS-B promised the paper would be online as of 00:01 BST today, but clearly that hasn’t yet happened. Will try to get the online link active as soon as possible.

Leave a Reply

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 )

Twitter picture

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

Facebook photo

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

Google+ photo

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

Connecting to %s




Follow

Get every new post delivered to your Inbox.

Join 6,691 other followers

%d bloggers like this: