- 13 Oct 2015
Over the course of two nights in September, SFI’s Vice President for Science Jennifer Dunne presented two talks on the broad topic of “The Web of Life” to the public of Santa Fe. Her talks, titled “The Hidden Order of Complex Ecosystems” and “The Ecological Human,” were part of the annual Stainslaw Ulam Memorial Lecture Series, which honors the memory of late theoretical mathematician Stanislaw Ulam.
The following is a summary of her lectures and their dominant themes by Connor O'Neil. For a more thorough treatment, you can view the lectures in their entirety here.
Part 1: The Hidden Order of Complex Ecosystems
If you’re not already aware, the concept of a food chain - insect eats plant, bird eats insect, wolf eats bird – is very limited. The food chain is, at best, a reminder that energy is transferred among species, but ends about there. In recent decades, the complexity of feeding interactions became more appropriately described as a food web, but most depictions were still simplified cartoons.
In the early 2000s, Dunne and her colleagues at the PEaCE Lab (Pacific Ecoinformatics and Computational Ecology) changed this with the development of an analysis and visualization tool: Network3D. The software allows detailed analysis of feeding relationship data, as well as 3D graphic capabilities that really justify the death of food chains.
The most important function of Network3D, however, is as a platform for bringing network theory to ecological data, for example in the Little Rock Lake Food Web graphic shown below.
A quick look at the Little Rock Lake Food Web suggests that a network perspective is a natural fit for ecology. The convenience of representing organisms (or groups of organisms) as nodes, and their interactions as links, has only become apparent with the luxury of large datasets and the computational power to analyze them. Now that there is growing interaction between the two fields – ecology and network theory, hidden properties of food webs are coming to light.
One such property of recent interest is the balance between generalists (those that eat lots of species, like raccoons) and specialists (those that will eat a few species, like koalas) in an ecosystem. Dunne and colleagues investigated this topic using a tool familiar to network theory – degree distribution. The degree distribution of a network is a measure of how many links each node has, or in our case, how many feeding interactions each species has. It’s a great tool for determining the number of generalists or specialists, but is not of much use unless there’s sufficient food web data.
Gathering this data is the dirty work, the kind that Dunne is “incredibly grateful that someone else does.” For perspective, one food web for the Northeast U.S. Marine Shelf was based on extracting and examining the stomach contents of over 300,000 organisms.
Once the data is compiled and loaded into the Network3D platform, one can talk meaningfully about the analysis. In this case, analysis meant determining the degree distribution from a variety of food webs using their respective datasets. When the distributions of distinct food webs are compared, the result is striking. Once the number of nodes and links are corrected for, a universal pattern pops out, which shows that most species in any food web are relative specialists, while a few species are extremely general in their diets.
Image at left: In the top center figure, the plot’s horizontal axis represents the generality of consumers; the vertical axis is the cumulative distribution. Points far to the right consume many species, but their lower value on the vertical axis means there are fewer of these species. The degree distributions for 13 different food webs are overlaid.
Given how different the species and environment are between a desert ecosystem, a rainforest, and a lake, you wouldn’t expect such order to emerge, but it does. Besides comparing the ecosystems listed on the left, Dunne’s team took the extra step to determine how universal this hidden order is by looking at food webs that include parasites, a huge and often ignored part of the diversity of an ecosystem. Like the webs without parasites, this web follows a similar distribution of specialists and generalists.
But to really settle the point, Dunne included data from the ancient Messel Shale and Burgess Shale fossil assemblages, which also fit comfortably on the above distribution. How were feeding interactions documented based on fossils that date back tends to hundreds of millions of years ago? You’ll just have to watch the video.
Part 2: The Ecological Human
Despite the title of the lecture, Dunne began the second night by detailing some depressing instances of humans acting in ecologically insensible ways.
The first case study put forth focused on human impact on marine life in the northern Adriatic Sea. By constructing food webs that date back thousands of years, Dunne’s colleagues were able to observe changes in species abundances in the Adriatic over deep human time. The interactive images below tell a story of decreasing species diversity and a depleted food web over the last 200 years in particular. But did you expect anything else?
The second example of humankind’s impact follows a piece of the food web of Ancient Egypt from ~12,000 years ago to today. In order to reconstruct networks of mammal carnivores and the herbivores they eat through time, SFI Omidyar Fellow Justin Yeakel and his colleagues conducted a unique survey of ancient Egyptian art to determine the representation of mammals through the ages, and supplemented that data with archeological and paleobiological data. For a thorough discussion of the results you can read the paper, but here is a quick summary: of the 38 larger-bodied mammals around 12,000 years ago, only 8 remain today.
The critical concern in both studies is the decreased stability of the food webs. What happens if we change the population size of one organism, or if we lose an entire species or set of species from the food web? In robust ecosystems, the change is ‘absorbed’ and the species interactions return to how they were. In less robust ecosystems, any change can result in unstable dynamics and further species extinctions. Both the current Adriatic Sea food web and the Egyptian mammal network have lost critical redundancies and resilience, which make them extremely vulnerable.
As a counter-example to these depressing cases, Dunne highlighted an extensive study of Sanak Island, Alaska, where the first high-resolution food webs that explicitly include human as predators have been compiled.
Sanak Island, one of the eastern-most Aleutian Islands, had been inhabited by humans for at least 6,000 years until recently, but has had no recorded long-term local extinctions. This is in spite of the fact that analyses of the marine food web there show that these human hunter-gatherers were playing special, potentially destabilizing roles in the ecosystem.
Then why didn’t the humans mess up the marine ecosystem that they invaded thousands of years ago?
One obvious answer is that they didn’t use industrialized technology like super trawler boats to indiscriminately strip mine fish along the coast. The subtler lesson can be discovered in their feeding behavior: the Aleut, although super-generalists, engaged in prey-switching. Whenever their preferred source of food decreased in quantity or became harder to find, they had no problem switching to another prey that was more abundant or easier to capture. This behavior is an ecologically “normal” behavior shared by generalist species, and it allows prey species to sequentially recover from the impacts of predation. In contrast, the modern industrial fishing of species like Bluefin tuna for the luxury seafood market shows the opposite behavior. As Bluefin tuna become scarcer, their economic value goes up, and they are fished more intensively—driving them towards extinction and potentially destabilizing their food webs.
Dunne’s research forces us to consider what behavior is ecologically responsible and irresponsible. The answers may not always be obvious, but as our impact on Earth becomes more intense, new approaches and limits need to be developed to help us become sensible Earthlings.
Dunne finished her Ulam lectures with a shout out to one such approach that she has a hand in, the Moorea IDEA project. The Moorea “Island Digital Ecosystem Avatar” will be a detailed, multi-dimensional digital model of the island of Moorea. The idea of “IDEA” is to create a powerful computational tool that can be used to explore, in a rigorous quantitative way, the sustainability of alternate future outcomes given different decision-making and environmental scenarios. Through the development and increased accessibility of resources like Moorea IDEA and Network3D, along with the improved data to feed into these and other tools, we are poised for a major transformation in our ecological understanding.
