Space is a final frontier in animal ecology. I first became interested in 3-D models of animal movements back when I was conducting research on the endangered dugong, a marine mammal and close cousin of the manatee. It quickly became apparent just how much these animals made use of all three dimensions when moving through their aquatic habitat. Dugongs have a digestive strategy similar to that of horses and rhinos, so they have to feed often to get sufficient nutrition, alternating breaths at the surface with frequent dives down to the seafloor to graze the seagrass pastures like a big sea cow. The tracking data that I collected from these tagged dugongs consisted of the 2-D latitude and longitude coordinates that give your position on the Earth’s surface, just like a handheld GPS or the navigation system in your car.
I also collected data on the dugong’s 3-D vertical coordinates, or how deep the animal was diving through the water over time. Most modern telemetry devices used to track wild animals provide these 3-D location data. But when it came time to analyze my dugong tracking dataset, I found that no one had yet devised a technique that satisfactorily combines the three dimensions of animal movements in a home-range model.
Animals typically don’t wander randomly across the landscape. They restrict their movements to areas where they can successfully find food and mates or care for offspring in what we call a home range. For example, my own home range consists largely of the space around my house and my office at the San Diego Zoo Institute for Conservation Research, with forays into the field to conduct research and expeditions to the supermarket, beach, pub, etc. Ecologists have been collecting data on animal movements for several decades, and new biotelemetry tracking devices are getting smaller and more accurate, enabling a wider range of species to be safely tracked for longer periods. Consequently, we are getting swamped with huge amounts of high-quality data on animal movements.
Instead of simply plotting all of these animal location data on a map as a confusing mess of thousands of GPS dots, we can use a home-range estimator that summarizes the areas that the animal visited during the tracking period. A home-range estimator uses clever algorithms to crunch all of the location data and create a mathematical picture of where the animal has been and which areas within its habitat it uses more than others. This is very useful information for scientists and conservation managers trying to better understand and conserve endangered species. Unfortunately, up until now, home-range estimators and their representations of animal movements have all been two-dimensional: flat and unrealistic.
I joined San Diego Zoo Global to conduct spatial ecology research that enhances the conservation management of the California condors we are reintroducing to their former range in Baja, Mexico (see post “Carrion” Research to the Next Level). And, like the dugong, these birds certainly move in 3-D. Condors can soar for hundreds of miles across the land looking for food as well as thousands of feet high up into the clouds. I needed an accurate 3-D home-range estimator that would make the most of the incredible 3-D movement data I was getting from these GPS-tracked condors.
To meet this quantitative challenge, I teamed up with the formidable cerebral firepower of Dr. Jeff Tracey (a computational ecologist at the U.S. Geological Survey Western Ecological Research Center) and Dr. Jun Zhu (a statistician at the University of Wisconsin-Madison). Together, we developed the first fully integrated technique to successfully crunch big tracking datasets and produce 3-D home ranges for animals.
The 3-D technique was honed using GPS tracking data collected from three different endangered species that occupy three different 3-D habitats: condor data for avian species, dugong data for aquatic species, and giant panda data for species that traverse steep terrain (collected by our collaborators at the Chinese Academy of Sciences). A 3-D home-range model enables us to visualize and investigate the movements of tracked animals in a way that is really intuitive and informative. For example, we can take our condor GPS tracking dataset and run it through our custom 3-D program with digital elevation data of their mountainous habitat to create dynamic 3-D visualizations and models of condor home ranges projected over the terrain. These 3-D models can show us where the condors spend their time, how condors are socially interacting, and where remote cliffside nests are sited.
Unfortunately, we soon discovered that enormous computing resources were needed to run the prototype 3-D home-range estimator, and it took most of a week to produce a single output for even a fairly small tracking dataset! Time to bring in the big guns. With the generous assistance of Dr. Robert Sinkovits and his team of experts in high-performance programming and computerized visualization at the San Diego Supercomputer Center, we were able to optimize our mathematical algorithms and computer code to achieve a remarkable 1,000x speed up of the processing. Now 3-D home ranges can be generated for large datasets in under an hour!
Our San Diego Zoo Global, U.S. Geological Survey, and San Diego Supercomputer Center research collaboration will soon make this 3-D home-range estimator available for any wildlife biologist to download and use with their own animal tracking data. As ecology enters the modern era of Big Data, we will continue to refine and improve the accuracy and power of our analytical toolbox.
James Sheppard, Ph.D., is a senior researcher for the San Diego Zoo Institute for Conservation Research. Read his previous post, Sharing Spatial Ecology. His paper was published in the journal PLOS ONE.