A comparison of traffic estimates of nocturnal flying animals using radar, thermal imaging, and acoustic recording

Authors

  • Kyle G. Horton,

    1. Department of Entomology and Wildlife Ecology, University of Delaware, 531 South College Avenue, Newark, Delaware 19716 USA
    Search for more papers by this author
    • Present address: Oklahoma Biological Survey, University of Oklahoma, 111 E. Chesapeake Street, Norman, Oklahoma 73019 USA. E-mail: hortonkg@ou.edu

  • W. Gregory Shriver,

    1. Department of Entomology and Wildlife Ecology, University of Delaware, 531 South College Avenue, Newark, Delaware 19716 USA
    Search for more papers by this author
  • Jeffrey J. Buler

    1. Department of Entomology and Wildlife Ecology, University of Delaware, 531 South College Avenue, Newark, Delaware 19716 USA
    Search for more papers by this author

  • Corresponding Editor: D. Brunton.

Abstract

There are several remote-sensing tools readily available for the study of nocturnally flying animals (e.g., migrating birds), each possessing unique measurement biases. We used three tools (weather surveillance radar, thermal infrared camera, and acoustic recorder) to measure temporal and spatial patterns of nocturnal traffic estimates of flying animals during the spring and fall of 2011 and 2012 in Lewes, Delaware, USA. Our objective was to compare measures among different technologies to better understand their animal detection biases. For radar and thermal imaging, the greatest observed traffic rate tended to occur at, or shortly after, evening twilight, whereas for the acoustic recorder, peak bird flight-calling activity was observed just prior to morning twilight. Comparing traffic rates during the night for all seasons, we found that mean nightly correlations between acoustics and the other two tools were weakly correlated (thermal infrared camera and acoustics, r = 0.004 ± 0.04 SE, n = 100 nights; radar and acoustics, r = 0.14 ± 0.04 SE, n = 101 nights), but highly variable on an individual nightly basis (range = −0.84 to 0.92, range = −0.73 to 0.94). The mean nightly correlations between traffic rates estimated by radar and by thermal infrared camera during the night were more strongly positively correlated (r = 0.39 ± 0.04 SE, n = 125 nights), but also were highly variable for individual nights (range = −0.76 to 0.98). Through comparison with radar data among numerous height intervals, we determined that flying animal height above the ground influenced thermal imaging positively and flight call detections negatively. Moreover, thermal imaging detections decreased with the presence of cloud cover and increased with mean ground flight speed of animals, whereas acoustic detections showed no relationship with cloud cover presence but did decrease with increased flight speed. We found sampling methods to be positively correlated when comparing mean nightly traffic rates across nights. The strength of these correlations generally increased throughout the night, peaking 2–3 hours before morning twilight. Given the convergence of measures by different tools at this time, we suggest that researchers consider sampling flight activity in the hours before morning twilight when differences due to detection biases among sampling tools appear to be minimized.

Ancillary