Annemarie Surlykke & Elisabeth K. V. Kalko

“ Thus, our results suggest that bats within a local assemblage compensate for frequency dependent attenuation by adjusting the emitted intensity to achieve comparable detection distances for prey across species. We conclude that for bats with similar hunting habits, prey detection range represents a unifying constraint on the emitted intensity largely independent of call shape, body size, and close phylogenetic relationships. ”

“ Overall, our study underlines the importance of intensity measures in the field as source level plays a crucial and so far largely underestimated role in bat echolocation. If we want to further understand which ecological and evolutionary factors shape echolocation signal design, an even larger variety of call parameters need to be considered, including sound duration and pulse interval, which may create call-echo overlap or other masking effects. ”

“ If maximum prey detection ranges are rather similar among sympatric species, because bats compensate for increased atmospheric attenuation of high frequencies by increasing output intensity, then frequency may contribute less to fine-grained niche differentiation than previously assumed. Comparison of maximum detection ranges suggests that detection probabilities of small insects are indeed rather similar across bats. ”

“ Flight speed, background clutter, distance from ground and other species specific differences most likely also play a role in shaping echolocation call characteristics as does phylogenetic affiliation, but based on our results we conclude that sonar range is an important evolutionary constraint adapting call intensity to achieve approximately similar detection ranges for prey. This may be primarily linked to the fact that the study species all share similar habitats, where they either hunt flying insects in open or edge space or take them from the water surface. ”

“ A worst case scenario would be a 1 m error at 10 m which would change the source level estimate for N. leporinus by at most 1.5 dB. Unless all distance errors were systematically biased to one side the effect on the average would be much less. Thus, we are confident that our results based on the array technique are reliable. Due to the post-triggering and long file duration, the acoustic method was much more efficient for data collection than the photographic method. Only recordings were the bats were in front of and approximately at the height of the array were included in the analyses. ”

“ The source level for this bat was rather low, but its low frequency resulted in the longest detection distance estimates within the edge space aerial foragers, both for small and large moths.
Also, the range at which hearing prey can detect bats relies heavily on both intensity and frequency (Tab. 2) . Even though prey insects usually have much higher hearing thresholds than bats, the prey's detection range is much longer than the bat's, because the prey detects the outgoing echolocation call, while the bat detects the weak reflected prey echo [25] . ”

“ Abstract Echolocating bats have successfully exploited a broad range of habitats and prey. Much research has demonstrated how time-frequency structure of echolocation calls of different species is adapted to acoustic constraints of habitats and foraging behaviors. ”

“ There was a slight indication of continued increase of source level with distance for the two Noctilio sp. at long distances (Fig. 2) . This is probably due to the sound quality criterion for selecting recordings for analysis. The further away the bat is the louder the calls had to be to reach criterion. Source level estimates for individual bats (Fig. 3) did not show this.
Source level was probably overestimated by about +6 dB for the two Noctilio due to positive interference from signal reflections on water surfaces [24] . ”

“ Estimated detection ranges still vary much less than emitted intensities, but some bats achieve almost twice as long maximum detection distances as other, mainly due to lower emitted frequency. Thus, if call frequency was to play an important role in resource partitioning, we would expect dietary differences in the amount of large insects taken by the bats and less so in the amount of small insects. ”

“ We recorded all bats in their natural habitat in the field using a multi-microphone array coupled with photographic methods to assess the bats' position in space to estimate emitted call intensities. All species emitted intense search signals. Output intensity was reduced when closing in on background by 4–7 dB per halving of distance. ”

“ The large number of high quality recordings permitted a robust data selection. The results confirmed our initial hypothesis about the specific (acoustic) conditions met when hunting over smooth water surfaces as both bats emitted very intense cries with average source levels of 136 and 137 dB SPL and maximum estimated levels above 140 dB SPL (Fig. 2, Tab. 2) . For comparison, the level at a loud rock concert is 115–120 dB and for humans the threshold of pain is around 120 dB. Thus, the levels emitted by the Noctilio sp. are extremely loud ”

“ Therefore table 1 also gives “Avg Max SL”, which is calculated as the mean of the maximum source level from each flight path. Since maximum source level is emitted at longer distances, we took “Avg Max SL” as the best representation of the emitted source level in search phase. The average maximum source level ranged from 121 to 137 dB SPL at long ranges (5–10 m) in the open well away from background clutter (Tab. 1) . At longer distances positioning of the bat became less reliable. ”

“ If anything, it is more likely that the estimates are too conservative (i.e. are underestimates) , since bat echolocation signals are highly directional with maximum sound intensity only in the acoustic axis [27] , [28] . This was the main reason for including only calls from bats flying towards the microphones, but even so bats may quickly turn their heads thus leading to off-axis recordings and hence under-estimation of source level. Data for Eptesicus fuscus [27] , [28] indicate that at about 20 deg. ”

“ In some cases, the cross-correlation process gave delays that were obviously wrong. Signals with a very shallow sweep were quite sensitive to relative differences in Doppler shifts due to difference in the bat's relative velocity with respect to the three microphones. ”