500,000,000 Years Later, Eating Patterns Persist
By Rob Knies
April 28, 2008 5:00 PM PT

Most of us struggle to live in the present. So many challenges, you know? And if they ever do subside, we tend to turn our thoughts to how much better things could be tomorrow. The past? It recedes rapidly from our focus. Been there, done that.

But then there’s Rich Williams of Microsoft Research Cambridge and his colleagues, able to peer with improbable accuracy into biological interactions from, oh, half a billion years ago.

It’s hard to believe, but the proof’s in the latest issue of the Public Library of Science (PLoS) publication PLoS Biology. PLoS is a nonprofit committed to making scientific and medical literature a public resource, and PLoS Biology is a peer-reviewed biology journal covering topics from molecules to ecosystems.

The current issue includes a paper entitled Compilation and Network Analyses of Cambrian Food Webs, co-written by Jennifer A. Dunne of the Santa Fe Institute, Williams, Neo D. Martinez of the Pacific Ecoinformatics and Computational Ecology Lab, Rachel A. Wood of the Grant Institute at the University of Edinburgh’s School of GeoSciences, and Douglas H. Erwin, curator of Paleozoic Invertebrates at the Smithsonian Institution’s National Museum of Natural History.

The paper—based on analyses of two sets of Cambrian-period fossils—suggests that more than 500 million years ago, the network of feeding interaction in the ecosystems—food webs—were very similar to those found in modern ecosystems.

Rich Williams
Rich Williams

“For a long time, paleontologists have dug up fossils and categorized them,” says Williams, head of the Computational Ecology and Environmental Science group at Microsoft Research Cambridge. “More recently, they’ve been thinking about the kinds of ecosystems that those fossil organisms were in, looking at fossil deposits as records of chunks of ecosystems.”

These particular chunks are remarkably well preserved sedimentary deposits, one called the Chengjiang shale, located in southwest China’s Yunnan province, the other named the Burgess shale, within Yoho National Park in the Canadian Rockies. The extraordinary richness of these sites has enabled analysis heretofore considered impossible.

“No one has ever tried to do what we did,” Williams said, “because they didn’t think they could pull together enough data, that the data wasn’t detailed enough to reconstruct a fairly detailed network of feeding interactions for very, very ancient ecosystems like these.”

Detailed indeed, as he soon learned.

“I learned a lot about paleontology in the process of doing this, and the level of detail in these particular fossil deposits is absolutely phenomenal,” he says. “They have 500-million-year-old fossils in which they can see the contents of one organism’s guts and identify another organism in that organism’s guts.

“That’s relatively rare, of course, but the fact that you can do that at all is amazing.”

The researchers began to ponder some basic questions: Has the structure of ecosystems changed as life has evolved? Have ecosystems themselves evolved? But even getting to the point where such questions could be raised required a bit of kismet.

“Jennifer Dunne, my longtime food-web colleague, was at the Santa Fe Institute,” Williams recalls, “and Doug Erwin, a very well-known paleontologist, was visiting the institute. He needed an office, Jen had space in hers, and they got to talking. Doug was interested in reconstructing food webs but didn’t really know how to go about doing it. Jen had some of those skills and brought together a group of people to help. It’s a great example of how interdisciplinary research can happen.”

That connection was forged six years ago. Shortly thereafter, Williams got involved.

“I was at the institute for a meeting about this in 2003,” he recalls, “where there was a group of about half a dozen paleontologists talking about what data was available. We decided to focus on a few particularly rich data sets, and these are two of them.

“Neo, Jennifer and I have been working together as food-web ecologists for about 10 years now. The question of how systems have changed on the evolutionary time scale came about because of Doug Erwin and Jennifer talking about it, and then Jen getting Neo and I interested. It’s a very intriguing question, but I would have had no idea you could get any data to be able to approach it.”

He also found himself intrigued by the practice of paleontology.

“The fossils have been examined for things like bite marks to infer feeding relationships,” he says. “In the more recent parts of the fossil record, the damage on leaves has been studied and used to understand things about the insects that were living on the leaves. They don’t just analyze fossilized organisms; they analyze coprolites, fossilized feces, to understand organisms and what they were consuming.

“It’s an absolutely remarkable level of detail, and I didn’t know anything about this before I started the project. It was really a great intellectual adventure to meet all these people.”

Of course, there is no road map to guide the intrepid adventurer through a fossil record shrouded in a half-billion years of antiquity. Useful, relevant conclusions require a bit of analytical speculation.

ORGANISMS ORGANIZED

“Very quickly after multicellular organisms evolved,” Williams explains, “they had assembled themselves into complex systems whose structure was remarkably similar to the structure of modern ecosystems. This has to be said with some caveats, however, because the data is more than 500 million years old, so it’s spotty.

“We’ve done a fair amount of work when assembling the data and analyzing the data to take into account the uncertainties that exist. We have tried to analyze the data and to look at the effect of including information with varying levels of uncertainty, seeing how that affects our conclusions, so as to not be too aggressive in interpreting the data. But to get anywhere, there’s a level of speculation in this work. We’re definitely pushing the data to say things it hasn’t said before.”

The PLoS Biology paper hardly consists of wild conjecture, though. The peer-review process made sure of that.

“It’s been reviewed by people in the paleontology community to look at the quality of the data we’re using,” Williams says. “It’s been reviewed by people in the food-web community to look at the kinds of food-web analyses we’ve done. It’s both aggressive in its use of the data and, also, I think, quite rigorous in the methods we’ve used.”

The implications of the research are staggering.

Graphic showing ancient food-web interaction
In this depiction of the food web of the Burgess Shale from the Middle Cambrian, spheres represent species or groups of species, and the links between them show feeding relationships. The drawing shows a top predator, Anomalocaris, chasing one of its likely prey species, the trilobite Olenoides, with arrows indicating their positions in the food web. Many aspects of the structure of this ancient ecological network are similar to the architecture of modern food webs.

“It says to me that here are basic underlying principles structuring ecosystems, structuring food webs,” Williams says. “I think those basic structuring principles are probably rooted in evolution, thermodynamics and basic chemistry and the relative balance of carbon, nitrogen and phosphorous in these systems and how that balance shifts as things get consumed.

“It’s really very basic thermodynamics. Organisms need energy to live, and when things eat each other, one of the main things they’re doing is passing the energy around. To stay alive and to maintain their biological structure, they’re dissipating energy, burning energy to maintain the order of the molecules in the organisms.”

The more things change …

“I think these basic structuring principles,” he says, “were in place back then like they are now. They are in place in deserts and in lakes and in terrestrial forests. They were in place near the equator and in place near the poles. It’s why we can have a relatively successful model of food-web structure.”

Still, he and his colleagues understand that the acclaim with which their findings are greeted is unlikely to be unanimous.

“It builds on a message from some ecologists that there are basic principles underlying the regularities observed across ecosystems,” Williams notes. “Of course, there’s also a huge amount of variability on top of that. Some of us focus on the underlying regularities, others focus on the system-specific details. We can sense conflict between those two ways of thinking, which is unfortunate, because I see them as complementary.

“If you understand the underlying patterns better, it allows you to hone in on the system-specific things more quickly and more efficiently. That’s the kind of interplay I would like to see going on in the science.”

How were the researchers able to take ancient fossils and generalize about food-web structure over a period of inconceivable immensity? It’s all in the data.

“The natural thing,” Williams chuckles, “would be to say: ‘OK, we’ve got a couple of ancient food webs. Let’s compare them to modern food webs and see whether they look similar.’ The problem is that there isn’t a lot of modern food-web data. It’s difficult to collect, and there are not a lot of high-quality data sets. So we mediated the comparison by using a model that Neo and I developed and published in Nature in 2000 that does a pretty good job of predicting the structure of modern food webs.

“The model allows you to generate huge numbers of data sets and do a rigorous statistical comparison between the model and its data. We know where that model does and doesn’t work with modern data. We compared that model to the ancient food webs to see if they deviated from the model in the same way that modern food webs do.”

There were data challenges, both strategic and tactical.

“Dealing with the critiques of the interpretations of the fossils and then dealing with the inherent uncertainty in the data,” Williams cites as some of the challenges, “and trying to develop techniques for analyzing the data that took uncertainty into account. I think it’s quite likely that we’ll migrate some of the techniques that we’ve developed for analyzing these ancient food webs into better ways of analyzing the more modern food webs.”

In the process, Williams gained new respect for those who gather ecosystem data for analysis.

‘ABSOLUTELY INCREDIBLE’

“The amount of work, whether by paleontologists or modern ecologists, to collect the data that allow theorists like me to do what we do—it’s absolutely incredible,” he laughs admiringly. “It’s a phenomenally labor-intensive undertaking. I’m always grateful for them for getting that data so I can—as I often say—use it and cram it into my little theoretical boxes.”

That Williams, Dunne, and Martinez all spent time at the Santa Fe Institute has helped shape their research.

“The institute is one of the pioneers in the study of complex systems,” says Williams, who attended a complex-systems summer school at the institute in 1997. He returned to Santa Fe as a visiting scholar in 2001, Dunne was by then a postdoctoral fellow at the institute, and Williams and Martinez served as her postdoc supervisors.

“My collaborators, Neo and Jen, are the reason why I’m an ecologist today. I was interested in ecology and looking for people to work with when I met Jen Dunn and she introduced me to Neo. Even though I had absolutely no training in biology, they helped me get started in ecology.”

The systems approach proved critical.

“Ecosystems,” Williams says, “are paradigmatic, complex systems. You have a very large number of interacting components, each component has relatively complex behavior, and we are interested in understanding the properties and predicting the behaviors of these systems.”

As he prepares to embark on his next project, Williams will be watching the reaction to the PLoS Biology paper.

“I think there will be people who think it’s really interesting,” he says, “and there will be people who dislike it because of the novelty of our approach. It’ll be interesting to see what kind of reception it receives in both the paleontological and food-web communities. Hopefully, it’ll stimulate more work on the really difficult part of this endeavor: pulling together the necessary data.”

Just the latest step in a fascinating exploration of the structure and the functioning of ecosystems.

“Recently, there’s been a tremendous dialogue between the theory and the data,” Williams concludes.”In the 10 years that I’ve been working in this area, there’s been a huge amount of renewed interest in the study and analysis of food webs and a huge amount of work that’s gone into collecting that data. It has become a much more vibrant research area.

“I feel like we, as a community, are making progress, but we have a huge distance to go. The systems we’re looking at are enormously complex, but it’s very exciting to be at a place where there’s so much intellectual activity.”