A whispering bat that screams: bimodal switch of foraging guild from gleaning to aerial hawking in the desert long-eared bat

All animals require accurate sensory information in order to interact with their environment, particularly with regard to foraging. Vision is often the dominant sense; however, for animals in poorly lit environments, other senses such as olfaction or hearing become more important. Another solution is biosonar, whereby individuals emit echolocation pulses and gain sensory information from the returning echoes, as used by most members of Chiroptera and some cetaceans (Griffin, 1958; Thomas et al., 2004). Echolocating animals can control the sensory input they receive by manipulating the outgoing signal. All echolocating bats (except Rousettus) emit ultrasound calls by pushing air through the larynx and out through either the nose or the mouth (Schnitzler and Kalko, 2001; Jones and Teeling, 2006). This sound is reflected by objects in their environment and returning echoes are detected by the bats' specialised ears (Griffin, 1944).

Recording from bats in the hand, Griffin (Griffin, 1958) classified echolocating bats into two groups: ‘loud’ aerial insectivores that catch insects on the wing and ‘whispering’ bats that are frugivores, nectarivores, carnivores or glean insects off the ground/vegetation. More recently, bats have been classified into broad foraging guilds including aerial hawkers that catch flying insects on the wing, gleaners that take prey off the substrate, which they detect by either listening for prey-generated sounds (passive gleaners) or by echolocation (active gleaners), and trawlers that catch prey from water surfaces (Norberg and Rayner, 1987). The echolocation calls of the bats in each guild conform to similar properties (Fenton, 1990).

However, foraging modes are not always fixed and animals may switch strategies depending on prey availability or environmental conditions. Orca (Orcinus orca) will use a variety of hunting strategies including creating waves that knock seals from ice floes (wave washing) and hunting in open water (Pitman and Durban, 2012). Hummingbirds (Trochilidae) and honeyeaters (Meliphagidae) have diets consisting of both nectar and arthropods (Pyke, 1980). This is true also for insectivorous bats, which may choose foraging strategies from multiple guilds (Fenton, 1990). For example, Myotis evotis is equally adept at gleaning moths from the substrate and aerial-hawking flying moths, and the duration and frequency of their echolocation calls change depending on the mode of attack (Faure and Barclay, 1994). Two further species of Myotis, one traditionally labelled as an aerial hawker (M. lucifugus) and the other a gleaner (M. septentrionalis), are both capable of catching insects using both tactics (Ratcliffe and Dawson, 2003). This is also true of M. emarginatus, which change the temporal structure of their calls (e.g. pulse duration and pulse interval), and Macrophyllum macrophyllum, which will use or omit distinct echolocation phases including a terminal buzz, depending on the tactic in use (Krull et al., 1991; Schumm et al., 1991; Weinbeer et al., 2013). Hence, foraging guilds in bats are not mutually exclusive as at least some species are able to use different tactics when necessary.

Modern advances in ultrasound recording technology have allowed for more accurate measurements of source levels, a measure of the sound intensity at 10 cm from the source of a call (i.e. mouth of the bat). Aerial-hawking bats (e.g. Eptesicus spp.) emit calls in the range of 103 to 137 dB peak-equivalent sound pressure level (peSPL) (Surlykke et al., 1993; Jensen and Miller, 1999; Boonman and Jones, 2002; Holderied and von Helversen, 2003; Holderied et al., 2005; Surlykke and Kalko, 2008), while quieter passive gleaning bats (e.g. gleaning Myotis spp.) only call with a source level of between 77 and 102 dB peSPL (Faure et al., 1990; Faure et al., 1993; Miller and Treat, 1993). There is also greater flexibility in the calls produced by bats than previously understood. Eptesicus fuscus decreases call intensity during the landing approach (Koblitz et al., 2010) and the decrease in source levels during the target approach has been well documented (Boonman and Jones, 2002; Holderied et al., 2005; Nørum et al., 2012). Following the initial discovery that M. macrophyllum and Artibeus jamaicensis, both said to be whispering bats, emit calls louder than previously thought (Brinkløv et al., 2009), Brinkløv et al. (Brinkløv et al., 2010) found that source levels in M. macrophyllum increased from flying in a flight room to semi-cluttered foraging, with the loudest calls being produced during open-space foraging.

Otonycteris hemprichii Peters 1859 is a desert passive gleaner that feeds primarily on non-aerial arthropods in dry, sparsely vegetated environments (Arlettaz et al., 1995; Gharaibeh and Qumsiyeh, 1995; Fenton et al., 1999; Korine and Pinshow, 2004; Holderied et al., 2011; Hackett et al., 2013). As with most gleaning species, it has long ears, low wing loading and a low aspect ratio (Gharaibeh and Qumsiyeh, 1995). Gleaning O. hemprichii fly close to the ground (40–100 cm) and land for 2–5 s to catch prey, which are then consumed while the bat adopts a slow, gliding and widely circling flight 3–7 m above the ground (Arlettaz et al., 1995; Holderied et al., 2011); thus these bats forage in a background-cluttered environment with echoes coming from a non-uniform ground surface. Holderied et al. (Holderied et al., 2011) recorded O. hemprichii gleaning tethered scorpions from the ground using a microphone array and recorded frequency modulated (FM) multi-harmonic echolocation calls typical of a whispering gleaning bat at 75.25±6.9 dB peSPL. However, anecdotal evidence suggests that, on occasion, O. hemprichii calls louder and adopts a different flight pattern than is typically reported (T.D.H. and C.K., personal observations).

Here we test the hypothesis that O. hemprichii can switch between foraging modes, namely passive gleaning and aerial hawking. We therefore predict that aerial-hawking individuals will change their flight behaviour with regard to flight height and flight speed. We also expect that aerial-hawking individuals will alter their call behaviour, namely call duration, pulse interval and, in particular, source level, which we expect to be significantly higher than during gleaning attacks and in the range of typical aerial hawkers.

In total, we recorded 849 individual echolocation calls in 60 flight paths from four different sites (mean: 15.38±11.25 calls per flight path; range: 4–76): 20 paths from Zefira (natural pool) and 40 paths from foraging sites (one from Nahal Bitaron, seven from the Moshav Hatzeva pool area and 32 from the Moshav Hatzeva date palm garden). We did not record or observe any drinking attempts by O. hemprichii at Zefira, but cannot fully exclude this as the behavioural context, so we did not pool Zefira with the other sites and will refer to it as the drinking site.

Two typical flight paths from foraging sites in Moshav Hatzeva are presented in Fig. 1. The bats recorded here did not fly in the typical search pattern as described in Holderied et al. (Holderied et al., 2011), having a different flight height (F_2,120=17.577, P<0.001; Table 1) and flight speed (F_2,120=13.893, P<0.001; Table 1). Bats at both the drinking site and the foraging sites flew higher (Tukey's HSD; drinking–gleaning: P<0.001, foraging–gleaning: P=0.01, drinking–foraging: P=0.001; Table 1) and slower (Tukey's HSD; drinking–gleaning: P=0.004, foraging–gleaning: P<0.001, drinking–foraging: P=0.34; Table 1) than during gleaning at Park Sapir (Holderied et al., 2011).

All recorded calls were of a higher source level than expected for a bat typically considered to be whispering with an overall mean of 118.8±7.7 dB peSPL (range: 93.5–140.5 dB peSPL). The mean frontal on-axis calls were 124.2±9.0 dB peSPL (range: 106.6–142.9 dB peSPL; 86 calls from 32 individuals) and the mean lateral (perpendicular to on-axis) calls were 119.8±6.4 dB peSPL (range: 109.7–132.3 dB peSPL; 68 calls from 24 individuals). The overall echolocation calls differed between the drinking site, foraging sites and while gleaning in pulse duration (F_2,130=33.064, P<0.001; Table 1), pulse interval (F_2,120=10.37, P<0.001; Table 1) and source levels (F_2,120=635.39, P<0.001; Table 1). There was no difference in pulse duration (Tukey's HSD; P=0.29; Table 1) or source levels (Tukey's HSD; P=0.29; Table 1) between the drinking and foraging sites, but calls were longer (Tukey's HSD; drinking–gleaning: P<0.001, foraging–gleaning: P<0.001; Table 1) and louder (Tukey's HSD; drinking–gleaning: P<0.001, foraging–gleaning: P<0.001; Table 1) than calls recorded while gleaning. Individuals showed no difference in the pulse interval between gleaning and at the foraging site (Tukey's HSD; P=0.96; Table 1), but at the drinking site the pulse interval was greater (Tukey's HSD; drinking–gleaning: P<0.001, drinking–foraging: P<0.001; Table 1). One of the flight trajectories contained an apparent abandoned aerial attack on an insect at a height 3.5 m from the ground (Fig. 1B). The corresponding echolocation behaviour, consisting of short-duration FM sweeps, resembled that of the approach phase of an aerial attack, with the pulse interval in this sequence decreasing to 18 ms (Fig. 2).

We collected five faecal pellets from five individual females, caught on three nights at the Moshav Hatzeva sites. Four of the pellets consisted of only one prey type – 100% by volume Diptera, Lepidoptera, Isopoda and Coleoptera each – and one pellet contained 75% Hymenoptera and 25% Lepidoptera.

Our results indicate that O. hemprichii can operate as an aerial hawker, as supported by a late approach or potential abandoned feeding buzz at a height 3.5 m from the ground, consistent differences in flight (speed and height) and echolocation (duration and source levels) behaviour between bats recorded here and those recorded while gleaning, and dietary differences from previous studies (Arlettaz et al., 1995; Fenton et al., 1999; Daniel, 2005). A decrease in call interval to 18 ms is slightly longer than typical pulse intervals during a terminal buzz of an aerial-hawking bat, but similar to late-approach and pre-buzz call intervals (Britton and Jones, 1999). Other bats have been shown to use multiple foraging tactics (Faure and Barclay, 1994; Ratcliffe and Dawson, 2003; Levin et al., 2009), and it is likely that most bats will catch insects on the wing when possible (Fenton, 1990). In fact, even primarily frugivorous and nectarivorous bats have been shown to catch insects on the wing opportunistically (Howell, 1974). Thus it is in line with previous research on bats' foraging adaptability that O. hemprichii is able to aerial hawk. The presence of insect wings in the majority of the faecal pellets indicates that O. hemprichii are feeding at least partly on winged insects, although this does not rule out the possibility that these were gleaned from substrates rather than taken in flight. Previous studies on this species' diet have found a higher prevalence of non-flying arthropods (e.g. Coleoptera, Solifugidae, Scorpionidae and Araneae), and only occasionally Diptera, Hymenoptera, Lepidoptera and Orthoptera (Arlettaz et al., 1995; Fenton et al., 1999; Daniel, 2005).

The source levels that we report here are considerably louder than those previously reported for O. hemprichii during gleaning (Holderied et al., 2011). This species adjusts its source level by more than 43 dB, corresponding to an increase in sound pressure by a factor of 141. We used the same tracking equipment and identical calculation methods as used to determine gleaning O. hemprichii source levels (Holderied et al., 2011), so this is not an artefact of experimental differences. This adaptability is considerably greater than the 11 dB reported for M. macrophyllum moving from open to cluttered environments (Brinkløv et al., 2010) as well as the 30 dB difference recorded during target approach in Myotis daubentonii (Nørum et al., 2012). The typical on-axis source levels of aerial-hawking bats are between 120 and 130 dB (Holderied and von Helversen, 2003); O. hemprichii in the recordings presented here are calling at almost the same level. Interestingly, there is no apparent change in call design that might serve to increase detection range even further, i.e. away from this species' typical short-duration, broadband, FM sweep towards a more narrowband, lower-frequency call (Jones and Holderied, 2007). Using the same selection criterion as Holderied et al. (Holderied et al., 2011), we included all calls, and not just on-axis calls. On-axis recordings were 5 dB louder, so rejecting off-axis calls would have artificially inflated the source level difference between our measurements and those from gleaning individuals in Holderied et al. (Holderied et al., 2011). Actually, as O. hemprichii looks down with ears pointing forward while searching for terrestrial prey (M.H., personal observation), lateral rather than on-axis calls may provide a better comparison, and even the lateral calls were only 1 dB different from the mean of all calls as used here. In conjunction with concurrent changes in flight and echolocation behaviour, we conclude that O. hemprichii is indeed able to switch between the two foraging modes of passive gleaning and aerial hawking. This is the first study to our knowledge showing a link between source level and foraging mode in the same species without a concurrent change in habitat clutter, and O. hemprichii shows the largest known adaptability in the sound pressure level of its call.

There are trade-offs in morphology when specialising in one foraging mode. Gleaning bats usually have long ears for the detection of faint sound cues and low wing loading to allow them to carry heavy loads. But these features increase drag and are linked to slower flight (Norberg and Rayner, 1987). Hence a long-eared bat such as O. hemprichii that specialises in gleaning is less adapted to aerial hawking, which is associated with higher flight speeds and agility (Gardiner et al., 2008; Gardiner et al., 2011). Interestingly, O. hemprichii recorded here during aerial hawking adopted a slower flight speed than during gleaning. Aerial-hawking bats of the genus Molossus have greater metabolic costs associated with more manoeuvrable flight in confined spaces, presumably because they are flying slower than the optimal speed predicted by their physiology and increased centripetal acceleration in curves (Voigt and Holderied, 2012). The slower aerial-hawking flight of O. hemprichii might thus allow for abrupt initiation of tight turns to chase more manoeuvrable flying prey.

It is thought that most bats will opportunistically switch to aerial hawking from their normal foraging mode if they detect flying prey; for instance, Myotis daubentonii switches between trawling and aerial hawking depending on prey availability (Todd and Waters, 2007). However, it does not appear that this is what we recorded; all flight paths consisted of loud calls that are atypical of usual gleaning behaviour and we did not observe any classic O. hemprichii low-flying search tactics (Holderied et al., 2011). Thus, the individuals in our study were most likely exclusively aerial hawking.

One possible reason for a bat adapted to one foraging mode adopting another is changes in environmental conditions or prey availability. The time of recording was toward the end of an exceptionally hot and dry summer with multiple record-breaking days, so it is likely that usual prey items were scarcer. Desert scorpions (e.g. Scorpio maurus palmatus), for instance, become more inactive during dry conditions (Shachak and Brand, 1983). It is therefore possible that O. hemprichii enters different habitats with artificial irrigation or attacks different prey and then adopts a different foraging mode in times of scarcity of its typical prey. Indeed, O. hemprichii in the region are typically found in natural habitats and rarely in artificial habitats such as the date palms and village gardens where we recorded (Hackett et al., 2013). Additionally, on moonlit nights, adult common yellow scorpions (Buthus occitanus israelis) are less active, and those that are active tend to ambush from under vegetation (Skutelsky, 1996). The moon was in its first quarter and setting between 22:00 and 00:15 h during this study, so it is unlikely that this was affecting the scorpions' behaviour. However, it is possible that O. hemprichii also switches to alternative prey and thus a different foraging mode on moonlit nights, when ambush predators such as desert scorpions are less active. Both S. maurus palmatus and B. occitanus israelis are common in the study area and, when presented, are readily taken as prey by O. hemprichii (Holderied et al., 2011).

The evidence that we present here indicates that not only does O. hemprichii switch foraging modes to aerial hawking, but when doing so they drastically alter both their call and flight behaviour. The malleability in source levels is the most extensive change seen in any bat species, indicating that perhaps bats are more flexible in their echolocation behaviour than previously understood.

We collected data in August 2010 at four sites in the Arava and Judean Deserts, Israel: one natural desert area with no water (Nahal Bitaron, 30°49′4N, 35°16′58E); two village gardens in Moshav Hatzeva, one of which was located next to a swimming pool (30°46′08N, 35°16′44E) and the other in an orchard of date palms (30°46′11N, 35°16′39E); and one natural pool where we observed other species of bats drinking (Zefira, 30°20′00N, 35°17′00E). All sites contained background clutter with bats flying less than 5 m from vertical structures, but sites ranged in openness from Nahal Bitaron, a relatively open site in a 200 m wide wadi (dry river bed) with 10 m between the tree and the cliff edge, to the date palm orchard in Moshav Hatzeva with 8 m between trees. Nahal Bitaron and Moshav Hatzeva are located ~20 km from Park Sapir, where Holderied et al. (Holderied et al., 2011) recorded source levels of gleaning O. hemprichii.

We used an array of 2×4 microphones (Knowles, BT1759) to track flight movements of O. hemprichii as described in Holderied and von Helversen (Holderied and von Helversen, 2003). We constructed flight paths based on the time of arrival differences of echolocation calls between these microphones using the custom-written software BatSonar (University of Erlangen). From these flight paths, we determined flight height and flight speed. To measure source levels, call duration and pulse interval (from start to start of two consecutive calls), we additionally recorded echolocation calls with a calibrated microphone (CO-100K, Sanken Microphone Co. Ltd, Tokyo, Japan, with Quadmic amplifier, RME, Audio AG, Haimhausen, Germany). Holderied et al. (Holderied et al., 2011) measured flight and echolocation behaviour of O. hemprichii gleaning tethered scorpions using the same tracking apparatus; see their paper for further details.

We calculated source levels from the fundamental frequency in dB peSPL using the sonar equation (Møhl, 1988). We determined the bats' distance to the calibrated microphone from the flight paths and, using measured temperature and humidity (Skywatch GEOS No.11, JDC Electronic SA, Yverdon-les-Bains, Switzerland), we compensated for spreading loss following Bazley (Bazley, 1976). We analysed all flight paths with more than four calls and employed three different selection criteria: (1) all calls including those that were not on-axis (these were used for the statistical comparisons); (2) only those calls on-axis, i.e. where the microphone was ±20 deg in azimuth and elevation relative to flight direction vector; and (3) only calls in a ±20 deg range of lateral directions, i.e. where the microphone was ±20 deg in elevation and between 70 and 110 deg or between −70 and −110 deg in azimuth relative to the flight direction vector.

On nights when we were not recording flight paths, we collected faecal droppings from five bats caught in mist nets at sites within Moshav Hatzeva. We inspected all pellets visually (Leica EZ5 OEM microscope, Leica Microsystems, Wetzlar, Germany), recorded any identifiable insect remains and report contents as percentage of arthropod type (often order) by volume.

Bat captures were conducted under license no. 34615 given to C.K. by the Israel Nature and Park Authority.

To remove pseudoreplication caused by different numbers of calls in individual flight paths, we calculated a mean value for each flight path for all parameters. All data were normally distributed and homoscedastic, allowing the use of parametric statistics. We used an ANOVA with Tukey post hoc tests to determine differences in flight height, flight speed, call duration, pulse interval and source levels between the drinking and foraging sites compared with gleaning data published in Holderied et al. (Holderied et al., 2011). All statistical tests were computed and graphs were created using R2.7.1 (R Foundation for Statistical Computing).

Lauren Holt assisted with data collection. This is publication no. 842 of the Mitrani Department of Desert Ecology.