© Bat Conservation and Management, Inc., Janet Tyburec, and John Chenger, 1998-2012. Absolutely no rights of distribution by sale or other transfer of ownership or by rental, lease or lending, preparation of derivative works, or reproduction, in whole or in part, is granted. No text, graphics or photos may be downloaded and used on another Internet site, without express permission of BCM, Inc. For information on obtaining photo useage and rights, please see our contact page. BCM, Inc. reserves the rights to actively protect against infringement.
To illustrate he variability in different detector’s and their responsiveness and the types recordings each produces and what this might mean to confident species ID classifications, I regularly deploy multiple detectors in the same location on the same night to calibrate units, measure the relative effectiveness of available settings, and responsiveness.
These results from a 5-hour monitoring period illustrate some of the variable results from different models. Each recorded different numbers of “bat passes,” with the D240x returning far fewer calls for the time-expansion limitations already discussed.
More interesting however, were the numbers of calls that were of high enough quality to render disambiguous ID classifications. The AR125, and Pettersson units all performed similarly. The SM2 however returned statistically significantly fewer calls of acceptable quality using the SMX-US microphone. Testing like this is essential to developing metrics when attempting to compare different units used for inventory and monitoring.
These three sonographs depict the same bat pass recorded with two different detectors:
Upper left is a D500x. Upper right is an SM2-384 w/SMX-US microphone.
Though they look similar enough, subtle differences in the microphone quality and responsiveness affects the call quality, rendering the left-hand recording more likely to be confidently classified to species.
Understanding the microphone limitations of the SM2, SonoBat includes a “compensator” designed to adjust for known discrepancies between the two microphone’s responsiveness.
Even-so, when the compensated call is compared to the D500, it still lacks certain clarity of detail such as power centers and quality that could render spurious ID results.
Finally, another mechanical concern: An example of power fluctuations. Two different D240x units were used to record a hand-released bat. The top one was functioning fine, the bottom one had a dying battery.
Finally, occasionally, the best-seeming locations and times to deploy bat detectors render poor quality calls.Bats, especially out west, are attracted to remote water sources, usually stock tanks or ephemeral pools. Unfortunately, the still water surface will often reflect echoes of the bats calls back to the detector, causing distortion. Often the distortion is severe and affects analysts abilities to discern important information about call shape and/or minimum frequency. (top screenshot)
Another problem that will impact call quality is when bat calls have to travel through air of differing densities, as is often caused by air masses of different temperatures.
This factor is often most pronounced right after dusk, right when bat activity often is at its peak as different surface substrates (roads, fields, water sources) cool off at different rates.
As the high-frequency bat calls and their returning echoes pass thru air masses with different densities, the sound waves travel at slightly different rates creating destructive interference and occasionally will cancel each other out, creating “null” spaces in the waveforms that affect the signal clarity. (bottom screenshot) This call was recorded around 7:45pm.
By 9pm, temperatures had equalized and recording conditions improved, rendering higher quality calls.
However, this is not to say that all passively recorded call sequences will be of insufficient quality for analysis and/or identification.
The sequence in the top screenshot was collected passively and it contains eight high-quality call pulses, including harmonics, indicating that the bat was echolocating within acceptable range of the detector microphone and within acceptable direction to the microphone for the majority of the call.
This is a high-enough quality call sequence from which to render a robust classification decision.
Yet, as evidenced by the next recording, made on the same night, in the same location, other call sequences during most passive recording periods will not be of sufficient quality for analysis, due to bats echolocating out of range of the microphone, or in a direction along a tangent to the cone of reception of the microphone (bottom screenshot).
The take home message is that not all call sequences recorded during a passive monitoring event will be of suitable quality to render a classification.
One limitation is the very real difference in call quality between actively collected calls and passive recordings. The top screenshot is a call collected with a TE D240x bat detector deployed in an active manner: it was held by a researcher who observed the bat flying, at dusk, as it was back-lit by the twilight sky. The researcher was able to follow the bat as it flew above the microphone and trigger this recording just as the bat was leaving the area.
The bottom screenshot recording was collected with a TE D240x bat detector deployed in a passive manner at the exact same location. It was positioned in a stationary manner, pointed up towards a likely cone of reception for intercepting commuting bats. The detector was rigged to record incoming calls to an MP3 player for later download and analysis.
It is easy to see the difference in call quality between the two recordings. The top recording contains twice as many call pulses, and more of the upper frequencies of each pulse are represented. This allows more potential for comparing the unknown pulses and the sequence in general with known bats from our library in order to render a decision.
The passively collected calls only contains a single call pulse emitted by the bat as it was flying over the microphone which is suitably strong enough for analysis.
Basing ID on a single call pulse provides a decision that is not nearly as robust as choosing several pulses from an entire sequence.
Once we’ve adjusted our efforts for known factors affecting bat activity during a survey period, then it is time to maximize the potential to collect high-quality calls that are more likely to yield confident classification decisions.
Back to Top
Recording Call Quality