 |
 |
|
|
||
The article below also touches on a subject about which I have great concern. This article discusses modeling. Some groups and managers keep trying to use models as if they are a perfect science. What really needs to be understood is that a modeling is more art than science. A modeler uses data and mathematical formulas to create a picture of a specific species or event. What a modeler portrays with the data reflects his or her interpretation and assumptions about the data available. They also extrapolate for data that is missing, making more assumptions along the way. The problem is not that modelers do this. Models have a tremendous value in sharing information and attempting to interpret what is happening. The problem occurs when we accept models as absolute truth and use that “truth” to make decisions. It is rare that a report lists all the assumptions and extrapolations that lead to the conclusions, leaving the readers with little guidance in interpreting the reports.
- Tom Fote
New York Times
September 21, 2004
Second Thoughts for a Designer of Software That Aids Conservation
By JON CHRISTENSEN
(from Jersey Coast Anglers Association November 2004 Newsletter)
Australia's recent decision to ban fishing on a third of the Great Barrier Reef represented an important endorsement for an obscure computer program that has become the brains behind many conservation plans around the world.
But last month, one of the inventors of the program raised questions about its usefulness in protecting the environment, suggesting that simpler rules used by conservationists for years may be more effective than elaborate plans concocted with computer software.
The program, Marxan, was developed in 1998 as a Ph.D. thesis by an Australian graduate student, Ian R. Ball, working with Dr. Hugh P. Possingham, a professor of mathematical ecology and director of the Ecology Center at the University of Queensland. The program, which is free, takes data about species and their habitats and spits out an optimal design for the most efficient network of reserves to preserve biodiversity in a region.
Since the software was developed, conservation groups and academic biologists worldwide have been using it and various offshoots to create plans to guide land purchases and government policies for protecting endangered species.
But in a paper in Ecology Letters and at a meeting of the Society for Conservation Biology at Columbia University, Dr. Possingham warned that, according to a recent study he conducted, unless such plans were put in place within a year after they were created, it was better to follow the simple rules of thumb that conservationists have long used, before the advent of complex computer-based conservation designs.
Many conservationists have criticized these conclusions, arguing that computer-based plans have many uses, in fund-raising, for example, and in providing goals for governments to meet.
Dr. Ball, who is now a researcher for the Australian government in Tasmania, said he welcomed dialogue on his program's effectiveness. "Any young scientist is very happy to see their work being used," he said in a telephone interview. "This one has had an effect because it was the right thesis at the right time."
Indeed, conservation planning is booming these days; the Great Barrier Reef is just one example of how planners are thinking big. In the United States, every state must have a conservation plan by 2005 to qualify for millions of dollars in federal financing for wildlife programs.
At Columbia, Dr. Possingham praised Dr. Ball's work as the "thesis that changed the largest area in the world," referring to the reef, which is now the world's largest protected marine reserve. But the plan for the reef was put in place all at once, he said, while most such plans, designed for the long haul, are carried out piecemeal over many years. They also generally do not take into account changes as some properties in an area become protected and others become developed.
"All these tools assume the world never changes, and that's a disturbing thing," Dr. Possingham said.
He wondered what happened to the value of a plan as environmental changes occurred. Because this was impossible to test in the real world, his study was a kind of thought experiment on the brain of Marxan.
He teamed up with colleagues from the United States - Dr. Sandy Andelman at the National Center for Ecological Analysis and Synthesis, an ecological computer center at the University of California, Santa Barbara, and Dr. Eli Meir of the Massachusetts Institute of Technology - to assemble data and run the experiment.
The scientists used real data about species and habitat from a conservation plan that Dr. Andelman directed for the Nature Conservancy along the upper reaches of the Columbia River plateau in the Pacific Northwest. They compared the results of following a Marxan-generated optimal plan over 10 years to the results of following simpler rules, which were to buy, and protect, property if it harbored the richest biodiversity available at the time, if it encompassed irreplaceable habitat, or if it was home to some species that conservationists wanted to protect.
The computer program assumed that only a few properties would become available for protection each year and that some were developed or degraded and thus lost for conservation. Under the optimal planning scheme, a property was purchased only if it was identified as essential in the plan.
Under the other schemes, properties were purchased if they met the simple rules.
In nearly every case, the simple rules performed better than the optimal plan. Even the simplest opportunistic decision-making - buying any available property that sheltered some species of concern - performed just about as well as the optimal plan.
Dr. Andelman said she was surprised at how badly the plan performed and how quickly it became useless compared with the simple rules. "People always complain about plans being put on the shelf," Dr. Andelman said, adding that her plan for the Columbia River Plateau ended up that way. "They probably belong on the shelf because their practical value is so short-lived. That was the surprising thing. Given the rate at which the world is changing, unless you can implement the plans in a year or so, they're outdated."
But other planners and conservationists defended computer modeling.
Dr. Peter Kareiva, a lead scientist for the Nature Conservancy's Pacific region, said most of the time and money spent on plans goes into gathering the species and habitat data needed for programs like Marxan; that data, he said, is needed to make good rule of thumb decisions too.
In California, he added, the conservancy is moving toward faster, cheaper plans that can be updated annually. But, he said, "You need those data no matter what."
Dr. Eric Dinerstein, chief scientist for World Wildlife Fund, said conservation plans were a good fund-raising tool, pointing out that they had persuaded governments to expand protection for tropical forests in Brazil and the Congo Basin, and helped get the World Bank's Global Environment Facility to allocate $11 million this year, as part of a $61 million, five-year plan, to protect the Cape region of South Africa. His fund, he said, spends $100,000 to $150,000 on ecoregional plans.
"That's not a bad return," Dr. Dinerstein said. "If you think big and have a comprehensive plan and have a map in front of decision makers and donors, you have a much better chance of generating the funds you need."
Dr. Mark L. Shaffer, an environmental program officer for the Doris Duke Charitable Foundation, agreed. The foundation is planning to spend up to $60 million over the next five years on land acquisitions and other conservation efforts around the country. But, he said, plans are more than fund-raising aids.
"If you can show people in Florida a map of what it will take, it changes how people think about what is possible," he said. "It reveals this problem in a more useful way than anything else."
Dr. Shaffer said Florida was the first state to develop a comprehensive plan for biodiversity in 1994, and that has been a model for conservation plans developed since then. Dr. Hilary M. Swain serves on a panel appointed by the governor to review state land acquisitions. In Florida, she said, the real estate market is moving so fast that the plan must be updated every six months.
A plan also provides public accountability, said Dr. Swain, who is director of the Archbold Biological Station in Lake Placid, Fla., one of the state's oldest nonprofit ecological reserves. "It provides a scorecard or check sheet for how we're doing," she said.
Dr. Possingham believes that there is a mathematical solution to this problem.
"If you can't formulate a problem mathematically, nobody knows what it is," he said.
"We need some tools for a dynamic world," he added. "We can provide some better rules of thumb." He is currently trying to incorporate information about real estate markets in his formulas.
He added that his latest paper was not an attack on plans. "It's an attack on plans that don't change."