I currently have a few papers in press or recently published, so I thought I would briefly mention them and see if I can convince you to invest the time on clicking on a download link and having a look to see what they are about.
Data, data everywhere…
We are producing more data about the Earth. More satellites in space looking down and charting the changes in the planet. More high altitude balloons. Aircraft based sensors. Flux towers. And people in the field. Struggling and sweating their way through rain forests collecting bugs and measuring the thickness of tree trunks.
Unfortunatey all this new data doesn’t immediately translate into new knowledge. In fact at times it seems as if we are drowning in data. Last year I was very fortunate to work with a group of very clever folk with the aim of trying to develop a new method for using some of this data to make better models of vegetation coverage. We were particularly interested in Dynamic Global Vegetation Models (DGVM). Essentially, we proposed some new ways that statistical tools (Bayesian statistic in particular if you’re interested in that sort of thing) that could be used to significantly improve model performance. How would this work? Well, at the moment many DGVM models are assessed in terms of how well they fit actual patterns of vegetation. If there is any difference between the two then we conclude there must be something wrong with the model as we are pretty sure the real world data we have is pretty good. So we go back to the model, fiddle with a few parameters and have another go. That was a bit better but still not perfect. Back to the model and adjust some more parameters. That’s better. What we have proposed is that this iterative ‘change something see what happens’ process can be made more efficient and principled using Bayesian statistics.
If you are a Bayesian (and I tend to think of myself as whilst not really being in the Bayesian gang, then at least wearing roughly the same clothes and sharing the same taste in music) then this will make perfect sense. If your jib is cut more to the style of frequentist interpretations, then already your teeth may be grinding in irritation without even reading the paper. And if you don’t know anything about statistics then most of this paragraph will mean absolutely nothing.
However, I think we do quite a good job of explaining the main issues in non-technical ways and whilst not exactly a ripping yarn would be interested to anyone who would like to know how we will produce models and predictions for how life on Earth, in particular forests and other vegetation, will respond to climate change in the future.
A version of the paper is available for free download here: hartig-etal-2012 Many thanks to Florian Hartig.
Birds do it. Bees do it. In particular beavers do it
Beavers are the classical example of an ecosystem engineer. They build dams and in doing so create lakes and so new habitats for aquatic creatures. We humans, like beavers, engineer our environments so that they are better suited to our needs. Our engineering is so ubiquitous that we tend not to notice it. Our houses, roads, parks, ports. There isn’t much of an industrialised country that hasn’t been altered by humans in some way.
Ecosystem engineering is a very important part of what life does. Not only because humans do it but in fact because so many other species alter their environments. I worked with a group of researchers (some of us worked on the this paper and the paper above) to try to capture these ecosystem engineering effects (we ended up calling this effect enviornmental modulation and it covers quite a wide range of procsses) in order to make better predictions for which species will grow in which particular habitat or environmental niche. Currently, most of these particular types of models assume that organisms will passively respond to changes in climate and conditions. Adding in environmental modulation can produce very different outcomes. In terms of climate change this could be a good thing (e.g. the dieback of Amazonian rain forest isn’t quite as severe as was previously expected) or a bad thing (fire promoting species may lead to more rapid deforestation). Truth of the matter is, we don’t really know how important these effects are because no one has done the global scale modelling. Yet. Hopefully.
A version of the paper is available for free download here: linder-etal-2012. Many thanks to Peter Linder.
Timing is everything
If you get hot (and perhaps bothered) you may sweat. If you get cold (and perhaps distant) you may shiver. Sweating and shivering are physiological processes that help your body maintain a core temperature. The sweat that evaporates on your skin carries away large amounts of heat. Shivering is produced by rapid contractions of your muscles that generates heat. Homeostasis (homoios = ‘like’ + stasis) is the term applied to these processes that maintain certain aspects of organisms to within relatively narrow bounds.
Iain Weaver and I have been thinking about homeostasis not in individuals organisms but ecosystems and potentially planets. Some of you may know of this work in terms of Gaia Theory and the mathematical Daisyworld model. What we were particularly interested in were the delays that operate between the different processes and how that could help or hinder homeostasis. Typically these delays aren’t incorporated into theories about such things because, well, they make the maths much harder and where possible we want to avoid certain assumptions and complications in order to make models as simple as possible. But these delays are important because pretty much all processes operating in the real world will be separeated in space and time from other processes. It takes time for temperature to build up or water to flow or ocean acidity to increase. Fortunately Iain had some very clever ideas about how to deal with delays in these models in ways that would still let us produce some clear analysis.
A version of the paper is available for free download here: weaver-dyke-2012