Trying to find an endangered Jemez Mountains Salamander is like looking for a needle in a haystack, except that haystack is an entire mountain range and the needle is buried underground. The salamanders spend their lives in rotten logs and rock crevices and emerge only at night to hunt insects and breed. Their home ranges are roughly the size of a large kitchen pantry: 86 square feet. Yet scientists and land managers must know where the salamanders are to protect them. That was the problem an interdisciplinary team from Los Alamos tried to solve by developing a sophisticated new tool to better understand the remaining salamanders’ habitat. “We wanted a map that considered how climate, geology, and topography influence where salamanders live,” says Andrew Bartlow, a Lab ecologist working on the project.
As with most threatened endemic species, Plethodon neomexicanus, which live only in New Mexico’s Jemez Mountains, evolved in a narrow range of environmental conditions that are now in rapid flux. The salamanders breathe through their skin and require moist soil to survive. Over the past couple decades, a warming and drying climate has parched the moist soil they need to breathe and led to five destructive fires that scorched more than 300,000 acres of Jemez forest. That included about a third of the known prime salamander habitat. In 2013, the U.S. Fish and Wildlife Service listed P. neomexicanus as an endangered species.
To focus and expand the search for the remaining salamanders, a multiagency team was assembled with researchers from the Lab’s Earth and Environmental Sciences, Bioscience, and Environmental Protection and Compliance Divisions, as well as the University of New Mexico and the New Mexico Department of Fish and Game. They turned to the machine-learning modeling algorithm MaxEnt (maximum entropy modeling), which predicts species distribution by mapping the environmental variables they require to survive. The algorithm could reveal on the landscape—down to the scale of five meters— where survey teams would have the best chance to find them. To train the algorithm to identify salamander habitat, the team used seven decades worth of data, collected from multiple local land management agencies, that looked at ten key habitat variables, such as climate, soil moisture, and precipitation. To test a hypothesis that geology and topography could be used to predict the salamanders' preferred habitat, they also added new layers to the model: geologic maps that underlaid the species’ range and LiDAR (Light Detection and Ranging) data that looked at slope and elevation. “For species that live mainly underground, it’s important to consider underground habitat,” Bartlow says. “If you ignore it, you’re ignoring a big part of their habitat requirements.”
MaxEnt’s results showed that geology and climate—especially winter maximum temperatures—were critical predictors of P. neomexicanus distribution. Half the salamander locations were identified in a certain geology type that was dominant in only 35% of the study area. Most exciting though was the algorithm’s predictions. Not only did MaxEnt correctly identify most of the known salamander locations, but it pinpointed several pockets of prime habitat that lay outside of the salamander’s federally protected area. No surveys have occurred in these newly identified locations yet, but Bartlow is optimistic about what researchers will find when they do.
“For the first time, ever,” Bartlow says, “We have maps that should lead us directly to the places where these creatures are most likely to be found, and that should help us protect the areas that could be important for the species' persistence.”