Clemson conservation and computational scientists team up – for nature’s sake
CLEMSON – Like a jigsaw puzzle with missing pieces, the natural world’s swaths of once-unbroken habitat have been fragmented and disconnected – much to the detriment of many animals and plants.
Biological diversity – the evolutionary and ecological process that gives rise to new species and assemblages – is paying a heavy price. One of fragmentation’s most insidious side effects is that it shuts off the ability of animals and plants to intermingle throughout their previous ranges, thereby reducing the flow of genes that is crucial to the health of most species.
“When these populations become isolated, they become too small to carry out their normal processes,” said Rob Baldwin, associate professor of Conservation Biology and GIS at Clemson University. “As a result, populations diminish and become susceptible to lots of negative things, and over time they can disappear.”
Baldwin and postdoctoral scholar Paul Leonard have long studied this issue. One of their more important avenues of research has been redesigning and enhancing state-of-the-art habitat connectivity maps that can be paired with satellite imagery to display the potential corridors used by animal populations to move between both large and small areas. The research has homed in on everything from highly mobile mammals like black bears to slow-moving reptiles like box turtles.
“We did not invent habitat connectivity modeling at Clemson, but we have greatly improved it,” said Baldwin, whose conservation career has spanned decades throughout the U.S. and Canada “And it is applicable in a wide variety of ways.”
One of the most significant improvements occurred around Christmas break last year. Edward Duffy, a computational scientist in the Department of Computing and Information Technology at Clemson, settled down in a quiet office and rewrote software that significantly sped up the computational limits of the mapping process. By then, Duffy and Leonard had long since developed a close collaborative relationship. As a result, Duffy designed the new and improved software to be compatible with the Palmetto cluster, the university’s primary high-performance computing resource that is the fourth fastest academic computer in the country.
Without the use of the Palmetto cluster, the maps envisioned by Clemson researchers would have been virtually impossible to generate. Off-the-shelf software run on a single desktop computer would have taken more than 2½ years to compute the millions of linear Algebraic calculations necessary to construct a large-scale, high-resolution map. But thanks to the collaboration between connectivity researchers and CCIT, what once would have taken years now takes less than two days.
“My job is to reach out to researchers and help them make more effective use of the technological tools at their disposal,” said Duffy, whose trailblazing software could be viewed as a Christmas present for the conservation industry. “One of my roles is to help people use the Palmetto cluster effectively. For the map project, the original software was poorly conceived in terms of speed and memory usage. So I rewrote the code from scratch and reduced individual calculations for this project from 25 minutes to three seconds. And I reduced the amount of memory consumed from about 90 gigabytes to fewer than 20.”
For his dissertation, Leonard used the Palmetto cluster to combine billions of bytes of data – including fine-grain satellite photographs and on-the-ground research – to produce geospatial maps of habitat connectivity modeling in South Carolina and the Southeastern United States. In terms of scale, resolution and detail, Leonard’s maps are the best in the business. They include habitat ranges and corridors, as well as areas of natural and artificial “resistance” to movement, such as mountains, rivers, cities and highways.
As an example, South Carolina conservationists have long known that black bears sometimes wander from the mountains and interbreed with bears that originated closer to S.C.’s coastal areas. However, the middle portion of the state has high levels of urbanization and major highways. The new maps display – in exquisite detail – the potential corridors the bears might be using to make these journeys, which they are able to complete despite all the obstacles.
“These maps help both individual landowners and NGO’s (non-governmental organizations) redefine their conservation priorities,” said Leonard, whose postdoc research applies to a 15-state region. “And then we’ve got landscape conservation cooperatives for the federal government that are trying to develop plans that span multiple states. The maps provide novel information that can help all these sources come up with smart and effective conservation plans. They aren’t the panacea, but I think of them as another arrow in the quiver.”
Because of Leonard’s work, users of the maps are now able to scale in and scale out of areas as relatively small as an individual forest or as large as the entire Continental United States.
“We can see patterns influencing local animals like box turtles, as well as wide-ranging animals like black bears,” Baldwin said. “We can reconnect fragments around cities in South Carolina. We can see the corridors between the Clemson Experimental Forest and Table Rock, or the Ace Basin and Georgetown. We can scale out and reconnect the Southeastern Coastal Plain with the Blue Ridge Mountains.”
Before the release of Leonard’s dissertation, fine-grained analyses of this caliber were unattainable. But the groundbreaking collaboration between Clemson’s conservation and computational scientists has literally unveiled a whole new world.
“Clemson is pioneering the marriage of high-throughput, high-performance computing with conservation biology,” Baldwin said. “And it has been made possible by the forward-thinking of CCIT, skilled postdoctoral researchers, and students who are competitive at the national level.”