Sonogram of bat echolocation call with quantitative parameters labeled.

Automated acoustic species classification of bird & bat vocalizations

Bats and wind energy

Bat echolocation

Comparative physiology

Physiological ecology

Bird acoutic recording station in a Sierra meadow.
 
Much of the recent and current work in my lab involves monitoring and identifying bats and birds from their vocalizations. I developed the the SonoBat echolocation analysis methodology and recently extended it into an automated system for continuous monitoring and species identification. The signal processing side of this system could justifiably be called a voice recognition system for bats. Initial development began in 2003 and tested artificial neural network and other machine learning systems developed with collaborator Stuart Parsons of the University of Auckland, New Zealand. That intial project was supported by SERDP for its potential to reduce the costs of monitoring threatened and endangered bats. Collaborator Michael L. Morrison of Texas A&M lead the wildlife study planning and implementation designs. Our lab has continued investigating and developing methods to improve acoustic species recognition and monitoring.

Other ongoing work in my lab understanding and developing solutions to the problem of bat mortality at wind turbines. Working with Ed Arnett at Bat Conservation International, and in with support from the Bats and Wind Energy Cooperative, we initially investigated potential acoustic attraction of wind turbines, and continue development of potential acoustic deterrents and contributed to methodologies for monitoring wildlife at wind energy facilties (Kunz et al. 2007) .

Installation of ultrasonic bat deterrent on a wind turbine in Pennsylvania.
 
With support from the California Department of Transportation we are developing hardware and software to automatically monitor and recognize bird vocalizations. In cooperation with the California Department of Fish and Game, we have tested this methodology in parallel with standard point count surveys in montane meadows.

 
Sonogram of Great Gray owl calls.

In another collaboration CA DFG, we recorded known great gray owls in Yosemite National Park and the northern Sierra and developed methodology to identify individuals from their calls with a 95% correct classification rate. This demonstrates the potential for non-invasive population census of these rare birds using automated acoustic tools.

 
In other studies and projects, I follow my interest in physiological interpretations of organisms in their environment through survey and monitoring initiatives of bats. Some recent and ongoing projects involve a meadow restoration project in the northern Sierra, the Pacific Northwest bat grid survey project, and the Lower Owens River Restoration Project, which involves the help of student interns and REU students. I also teach a field classes and workshops on bat biology. I also pursue the acoustic analysis of bat echolocation calls (here's a BCI workshop on that), and have coded SonoBat software for that purpose.  
       
My lab research pursues the nuts and bolts of how animals work.In particular, I investigate processes that involve gas exchange such as acid-base state, the control of ventilation, and how animals use that gas, i.e. metabolism. Where possible, I like to explore these processes in their most extreme forms. Just as we can reveal critical elements of automotive engineering by racing Formula 1 cars, we can also learn much by studying physiological function near its limits. For this reason, the most aerobic of mammals, bats, have often been a subject for my research. I have explored their physiological functioning from the cold depths of torpor, to how they acclimatize and fly at high altitudes. Captured Townsend's big-eared bat.  

I enjoy figuring out how animals solve problems particular to their unique place in nature. Of course all animals must use the same basic tools that evolution has provided them, however animals have developed surprisingly different strategies with those tools, and that is what I find of interest. For example, bats have the highest metabolic capacity among mammals. Yet many bats endure winter in a state of heterothermic torpor, thereby matching their lofty metabolic capacity with an equally impressive metabolic reduction. For the 15 g bat Eptesicus fuscus, from the highest estimated metabolic cost of flight to the lowest measured metabolism at a body temperature of 5°C (Szewczak and Jackson, 1992) yields an impressive metabolic scope of 1,270! (Typical mammalian values for metabolic scope are on the order of 10.)

Because starving during winter may account for the greatest mortality among bats that hibernate, bats have surely selected strategies to operate their high capacity gas exchange system at minimal cost during torpor. This may not seem much of a problem, but, continuing with the automotive metaphor, imagine the inefficiency of powering a scooter with a Formula 1 engine. My work has found that one component by which bats surmount this dilemma is by breathing intermittently, with nonbreathing (apnea) periods of up to 2 and a half hours (Szewczak, 1998). Even with the miniscule metabolic rate of torpor, these bats only have an estimated 17 minutes of on board oxygen reserves. Therefore, they must have a means to acquire oxygen without breathing. We found that torpid bats acquire this oxygen by a passive process of diffusive convection down their airway (Szewczak and Jackson, 1992). Over the duration of a winter hibernation, this can save an estimated one to two weeks of hibernation time. Bats have a large cutaneous surface area in their wings, and this might contribute to apneic oxygen uptake. However, we tested the airway convection theory in intermittently breathing torpid pocket mice, and the results fit our predictions (Sullivan and Szewczak, 1998). Other projects I pursue include reduced metabolism in low oxygen environments and cold, high altitude acclimation in bats, other small mammals, birds, and lizards, acid-base regulation in small mammals, and nonlinear analysis of biological signals.