Hawaiian hoary bat (Lasiurus cinereus semotus) behavior at wind turbines on Maui

Abstract

This study examined the activity of the endemic Hawaiian hoary bat (Lasiurus cinereus semotus) at wind turbines operated by Auwahi Wind Energy, LLC, on southern Maui Island, from August to November 2018. The research was conducted to assess the potential effect of wind speed and turbine operation on bat presence and behavior and compared information obtained from both acoustic monitoring and thermal videography. During the four months of nightly surveillance at four wind turbines, we observed 384 visual (videographic) and 244 acoustic detection events involving bats. Bats were infrequently detected, averaging 0.08 events per hour for both visual and acoustic samples. Detections occurred throughout the monitoring period, but bat presence was only evident for a fraction (acoustic: 30%; visual: 44%) of the turbine-nights sampled. Bats were present throughout the night, but detections exhibited a unimodal peak centered on the first third of the night, with events largely absent in the latter half of the night and no apparent seasonal trend towards earlier or later occurrence within nights. However, a decline in the visual detection rate was noted over the four-month period (a similar assessment was not available from acoustic samples due to missing data for much of the later months). Visual bat detections were not significantly correlated over nights (i.e., temporally), but were positively associated among turbines (i.e., spatially). Visual detections were generally brief (median = 9.0 sec), infrequent (median time between events = 49.0 min), and involved single passes (57%) largely comprised of a single bat (94%). The amount of time during which bats were visually observed amounted to only 0.05% of total videographic monitoring (2.5 hours of 5,066 total hours). Although not directly comparable to the video results because of differences in the volume of airspace sampled and nature of observation, acoustic detection events were similarly brief (median = 6.0 sec), infrequent (median time between passes = 38.8 min), and also composed only 0.05% of the total period of acoustic monitoring (1.6 hours of 3,036 total hours). Most visual observations (61%) were of individuals flying at some point during the event to within about 15 m of the turbine nacelle (machinery housing atop the monopole). Erratic flight paths were the most prevalent flight type with bats often repeatedly approaching and circling the nacelle. Terminal-phase (“feeding buzz”) calls were only noted in 3% of all acoustic events. Bats were most frequently detected visually at relatively low wind speeds (median = 3.4 m/sec); however, 10% of events occurred at wind speeds over 8.5 m/sec. Nightly bat detection rates for the four-month period of monitoring were negatively correlated with total daily precipitation. Generalized linear mixed model analysis confirmed that detection rates were negatively associated with wind speed and precipitation and indicated a positive relation with intermittent wind speed and its consequent effect on turbine blade rotation (i.e., frequent intervals of starting and stopping). The co-occurrence of bat detection obtained from videographic and acoustic monitoring methods was generally low, and in instances when individuals were visually observed, bats were detected acoustically during only 12% (within a 10-minute window), 22% (within a 2-hour window), and 56% (at some point during the entire night) of such events. Most visual detections (65% within a 2-hour window) lacking an acoustic detection involved bats observed flying within about 15 m of the turbine nacelle on which acoustic detector microphones were situated.