Horseshoe (Rhinolphidae) and Old World leaf-nosed (Hipposideridae) bats are high duty cycle (HDC) echolocators sharing a suite of adaptations including long duration signals relative to their signal periods, peak energy concentrated in a narrow spectral band dominated by a constant frequency (CF) component, ‘auditory fovea’ (over-representation and sharp tuning of neurons responsible for frequencies at or around the CF) and ability to compensate for Doppler shifts in echoes. HDC bats separate signals from returning echoes in the frequency domain. Rhinolophids are more specialised neurobiologically than hipposiderids, producing longer duration signals at higher duty cycles, and have narrowly tuned auditory fovea and almost full Doppler shift compensation. Here, I examine whether these differences have produced ecological divergence between the families by testing predictions of differences in prey perception, prey capture behaviour, foraging habitat and diet. I found no discernible differences in these variables between the two families. Rhinolophids and hipposiderids both forage close to vegetation, capture prey by aerial hawking and gleaning from surfaces, and consume mostly flying insects with spiders and terrestrial, flightless arthropods taken occasionally. The data presented here show that the two families are similar in foraging ecology despite differences in echolocation and audition.
Introduction
Echolocation is a sensory system involving the tight coupling of signal production and echo reception that is used by species in four orders of mammals and two orders of birds. Echolocating animals orient by collecting information from the difference between signal pulses and returning echoes to form an acoustic image of their environment (Fenton et al. 2012). While all echolocators use the system for orientation, echolocation is also used for prey detection by bats and odontocete whales (Fenton et al. 2012).
Among bats (Chiroptera) several distinct approaches to echolocation have evolved, with each approach differing in the structure of the echolocation calls and in how the calls are separated from echoes. Each approach has resulted in the evolution of a unique set of auditory adaptations that enable signals to be received and processed by the brain (Neuweiler 1990; Fenton et al. 2012). Most echolocating bats avoid forward masking, the process by which louder outgoing signals mask or reduce the sensitivity of the animal to the weaker returning echoes, by separating pulse and echo in the time domain (Fenton et al. 1995). An alternative strategy is found in high duty cycle (HDC) echolocators that separate pulse and echo in the frequency domain (Schuller 1974). The duty cycle (DC) is the percentage of time that a bat is producing sound (Fawcett et al. 2015). Duty cycles of HDC bats range between 25 and 70%, whereas those of other echolocating bats typically range between 5 and 20% (Fenton et al. 1995).
Fenton et al. (2012) state that HDC echolocators share a unique combination of four adaptations. First, they emit echolocation signals with long durations relative to their signal periods. Second, the peak energy of each signal is concentrated in a narrow spectral band dominated by a constant frequency (CF) component. These signals typically consist of two or more harmonics each composed of a long constant frequency component terminated by a brief frequency modulation (FM) (Neuweiler 2003). The most intense harmonic is the second harmonic (CF2). The resting frequency of CF2 is very stable and shows little deviation from the mean value. Third, they possess an ‘auditory fovea’, a spatial over-representation and sharp tuning of the neurons responsible for frequencies at or around the second harmonic of the CF call component, both on the basilar membrane within the cochlear of the inner ear and in the neurons of the centres of the ascending auditory pathway of the brain. Narrow tuning is achieved via several pronounced anatomical adaptations within the fovea, including structural specialisation of the basilar and tectorial membrane of the inner ear (Neuweiler 2003). Last, they show Doppler shift compensation (DSC), an adaptation to overcome Doppler shifts, the change in sound frequency associated with the movement of the sound source (nostril of the bat) relative to the receiver (ears of the bat). Typically, Doppler shifts result in the frequency of returning echoes being higher than the narrow frequency range to which the auditory fovea is tuned, therefore, DSC involves lowering the frequency of the next outgoing signal to compensate for the Doppler-shifted increase in frequency of the echo from the previous signal resulting from the bat’s flight (Neuweiler 2000, 2003). DSC is a precise behavioural mechanism that operates at exceptional speed (Grinnell 1989).
HDC echolocation, as defined above, is known from ∼200 species of echolocating bats. All except one species are in three families in the suborder Yinpterochiroptera: Rhinolophidae (horseshoe bats, 103 species in a single genus), Hipposideridae (leaf-nosed bats, 88 species in seven genera) and Rhinonycteridae (trident-nosed bats, nine species in four genera) (Mammal Diversity Database as at May 2020: American Society of Mammalogists 2020). The only other known HDC echolocator, Pteronotus parnellii (Mormoopidae) is within the suborder Yangochiroptera.
The features of the echolocation system of HDC bats are clearly interconnected and appear to have evolved in response to the same selective pressures, thus representing a complex adaptation (sensu Walter 2003). The auditory fovea could not function effectively without the ability to compensate for Doppler shifts. In turn, these auditory adaptations enable bats to call at high duty cycles because they can cope with temporal overlap between signals and echoes. Although HDC echolocators share the four characters outlined above, there are differences in these characters among the families of HDC echolocators. The functioning of the HDC echolocation system has been examined in most detail in the Rhinolophidae where a close link has been established between the use of HDC echolocation and foraging within dense vegetation, which has high levels of acoustic clutter (i.e. echoes from background objects that interfere with the perception of echoes from the target). This work indicates that the rhinolophid echolocation system is adapted for coping with clutter while enabling horseshoe bats to detect and capture flying insects (e.g. Neuweiler et al. 1987). Because of this understanding the rhinolophid approach to echolocation is described as a flutter detection and clutter rejection system (Schnitzler and Denzinger 2011).
Multiple studies suggest that rhinolophids are the most specialised HDC echolocators in terms of neurobiological characters, with the hipposiderids and rhinonycterids the least specialised and Pteronotus parnellii in between (Fenton et al. 2012, and references therein). Jacobs et al. (2007) expanded this perspective and questioned whether the neurobiological differences meant that rhinolophid and hipposiderid bats had diverged in foraging ecology and, therefore, should not be regarded as belonging to the same foraging guild. The purpose of this review is to examine the proposition of Jacobs et al. (2007) that the differences in neurobiological adaptations among the two families has resulted in differences in foraging ecology, i.e. to address whether Rhinolophidae and Hipposideridae belong to the same foraging guild. I have excluded Rhinonycteridae from this analysis. Formerly included within the Hipposideridae, Rhinonycteridae was recently separated into its own family (Foley et al. 2015) and little information is available for most species. The Rhinolophidae and Hipposideridae both possess an ornate noseleaf and have broad mobile ears (Hall 1989a, 1989b). The two families broadly overlap in wing morphology, having wing designs suited for slow and manoeuvrable flight (Norberg and Rayner 1987).
The review first provides a summary of HDC echolocation in rhinolophid bats focussing on explanations of how neurobiological adaptations influence important aspects of foraging, including prey perception, prey capture behaviour, habitat and diet. Next, I summarise differences in echolocation signals and neurobiological structures between rhinolophids and hipposiderids and make specific, testable predictions of how these differences could influence foraging ecology. I then test the predictions by summarising published information on prey perception, prey capture behaviour, habitat and diet of the two families.
The HDC echolocation system in rhinolophid bats
Rhinolophid bats frequently forage around vegetation (e.g. Neuweiler et al. 1987; Jones and Rayner 1989; Pavey 1998a). While foraging around vegetation, rhinolophid bats are able to reject acoustic clutter and detect the wingbeats of fluttering insects (Schnitzler et al. 1985; Neuweiler et al. 1987; Neuweiler 1990). The pure tone echoes from the calls of these bats are highly noise resistant, being able to maintain their structure despite the movement of foliage (Neuweiler 1989, 1990). Rübsamen et al. (1988) proposed that the requirement for clutter rejection led to bats, which already had pure tone signals, developing narrow auditory foveae. A pure tone signal is more resistant than any other type of signal provided the receiver is tuned to the frequency of the signal. The fovea provides such auditory tuning. The use of such a narrow receiving filter requires a long duration signal. Therefore, a byproduct of the evolution of a narrow auditory fovea was the use of long duration pure tone signals. Thus the narrowly tuned auditory filter can be seen as an evolutionary adaptation for clutter resistance (Neuweiler 1989).
The Doppler-shifted long pure tone signals of rhinolophids allow the detection of glints from fluttering insects, thus being effective in the detection of fluttering targets but not targets in other situations. Rhinolophids are able to use both positive and negative DSC and therefore are able to deal with increases and decreases in the frequency of returning echoes (Metzner et al. 2002). The echoes carry distinct ‘acoustical glints’ from fluttering targets such as flying insects. The glints are brief frequency and amplitude modulations superimposed on the CF component of the echo by reflections from the moving wings of an insect. The glints are generated by the rhythmic motion of the insect’s wings relative to the direction of sound propagation from the bat (Schnitzler 1987; Kober and Schnitzler 1990). Amplitude glints are produced by changes in the reflective area of the fluttering insect; when the insect’s wings are perpendicular to the direction of sound propagation it presents a larger reflective surface compared with the body of the insect than when the wings are horizontal. Frequency glints are spectral broadenings in the CF component of the echo produced by the movement of the insect’s wings towards or away from the bat. Frequency glints provide information on the direction of travel of the insect and are superimposed on the overall CF echo from the insect’s body (Fenton et al. 2012, and references therein). In comparison to the high level of detail obtained from an insect moving its wings, the Doppler-shifted echoes do not carry glints from insects that do not move their wings.
The information on wingbeats of insects available to rhinolophid bats from amplitude and frequency glints in returning Doppler-shifted echoes raises the possibility that the bats can recognise particular types of insects (Schnitzler 1987; von der Emde and Schnitzler 1990; Kober and Schnitzler 1990). Rhinolophid bats in the laboratory are able to classify insects on the basis of their wingbeat frequency and, consequently, may be able to actively select prey in the wild (von der Emde and Menne 1989). A higher duty cycle increases the probability that a glint is contained in an echo coming from fluttering prey and only bats with CF durations >40 ms will frequently receive multiple acoustic glints over several wingbeat cycles of a fluttering insect in a single echo (von der Emde and Schnitzler 1986; Fenton et al. 2012).
The ability of rhinolophids to detect fluttering targets is not confined to areas with high levels of acoustic clutter. However, because flutter detection is advantageous in a highly cluttered environment, rhinolophid bats are at a competitive advantage in areas of dense vegetation. Insects flying around vegetation at night are a rich, underexploited food resource, particularly in the Old World where a small proportion of bat species forage in cluttered settings (Fenton et al. 1995). Therefore, rhinolophid species are expected to favour such foraging habitat.
Comparison of HDC echolocation in rhinolophids and hipposiderids
Here I examine differences in echolocation signals and neurobiological structures between rhinolophids and hipposiderids. I summarise information on signal duration and duty cycle, Doppler shift compensation and the degree of tuning of the auditory fovea.
Signal duration and duty cycle
Data on signal duration and duty cycle were collated for those studies that recorded resting frequency of hand-held bats, to avoid Doppler-shift effects (e.g. Jacobs and Bastian 2018), or search phase signals of flying bats (i.e. feeding buzzes were excluded). Data on signal duration are available for a sample of 25 rhinolophid and 10 hipposiderid species (Fig. 1). No overlap in signal duration occurs between the two families. Rhinolophid species have longer duration calls (range of 20.8 to 53.5 ms) than hipposiderids (range of 5.4 to 12.0 ms) (Fig. 1).
Fig. 1.
Signal duration of rhinolophid (filled circles, n = 25 species) and hipposiderid (open triangles, n = 10 species) bats plotted against maximum frequency of the CF component of calls.
Data on duty cycle are available for 24 rhinolophid and 6 hipposiderid species (Fig. 2). The data summarised in the figure disagree with a previous summary (see table 1 of Jones 1999), which showed that the duty cycles of the two families did not overlap and variation within families was not great. After Jones (1999), a wider sample of rhinolophid species has been assessed. Evidence has emerged of overlap in duty cycles across the two families and of intraspecific variation in signal frequency across geographic ranges and depending on the task being undertaken and the habitat (e.g. Pavey et al. 2001a). Intrafamily variation in duty cycles can be noticeable. For example, hand-held individuals of R. macrotis (forearm length, 41.8 mm) and R. lepidus (forearm length, 41.5 mm) that roosted in the same cave had duty cycles of 39.23% and 54.70%, respectively, when measured using identical methods (Shi et al. 2009).
Fig. 2.
Duty cycle of rhinolophid (filled circles, n = 24 species) and hipposiderid (open triangles, n = 6 species) bats plotted against maximum frequency of the CF component of calls.
In general, duty cycles of rhinolophids are greater than those of hipposiderids. Typical duty cycles of rhinolophid bats in flight are over 50%, whereas those of hipposiderid species are 35% or less (Fig. 2). Fullard et al. (2008) recorded the highest duty cycle for a hipposiderid bat so far. In their study, H. ater recorded flying inside a roost had a duty cycle of 54.2% with an interpulse interval of 4.6 ms and a signal duration of 5.0 ms. In the same study, R. megaphyllus had a duty cycle of 51.7% with an interpulse interval of 55.4 ms and a signal duration of 52.9 ms. In this case the duty cycles of the two species were similar despite the much longer duration signals of R. megaphyllus. Hipposideros ater had a high duty cycle as a result of a shorter interpulse interval and higher pulse reputation rate.
Doppler shift compensation
Accurate measurement of DSC requires experimentation on captive animals under controlled conditions. Therefore, data are available only for a small sample: two rhinolophids, and three hipposiderid species. Both R. ferrumequinum and R. rouxi (Schuller 1980) exhibit close to full compensation (100%) of Doppler-shifts in returning echoes. In comparison, H. speoris and H. bicolor exhibit 55–56% compensation (Habersetzer et al. 1984), whereas H. lankadiva had a higher capacity to compensate at 77% (Pillat and Schmidt 1998).
Auditory fovea and stability of the resting frequency
The degree of tuning of the auditory fovea has been measured with respect to tonotopic organisation of brain centres. Audiograms of rhinolophid and hipposiderid species allow comparison of the auditory fovea. A finely tuned auditory fovea was first demonstrated in R. ferrumequinum (Schuller and Pollak 1979). In this species the auditory fovea is considered to consist of a 1.5 kHz band from 83.0 to 84.5 kHz. A total of 16% (96 neurons) of all neurons had best frequencies within this band (Schuller and Pollak 1979). A similarly very sharply tuned fovea has been shown in R. rouxi (Schuller 1980). By comparison, audiograms of three hipposiderid species are not so narrowly tuned and less sensitive to emitted frequencies. The species assessed have been H. speoris and H. bicolor (Neuweiler et al. 1984) and H. lankadiva (Peters 1987 cited in Foeller and Kössl (2000)). Further, in H. speoris and H. bicolor the neurons that process pure tone frequencies occur in a smaller and more confined part of the inferior colliculus (Neuweiler et al. 1984).
Measurements of otoacoustic emissions of the bat cochlea also provide comparative data on the hearing abilities of bats. Otoacoustic emissions are sound waves generated by the cochlea that can be measured in the outer ear canal. If stimulated with two tones (f1 and f2) the cochlea generates distortion-product otoacoustic emissions (DPOAEs). The threshold curves for these DPOAEs, the so-called distortion-product audiogram, gives a close approximation to the neuronal audiogram for a given species (Vater 1998). Using this approach Foeller and Kössl (2000) demonstrated that H. lankadiva has a broader cochlear fovea than R. rouxi. The DPOAE threshold curves showed the threshold increase to CF2 in H. lankadiva amounted to ∼20 dB compared with 40 dB in R. rouxi. Further, H. lankadiva lacked a sharply tuned threshold minimum slightly above CF2 (Foeller and Kössl 2000).
Stability of the resting frequency of calls has been measured in several species of rhinolophid and hipposiderid. Bats at rest do not experience Doppler shift and emit calls at the resting frequency, which can be used as an indirect measure of the auditory fovea (Jacobs et al. 2007). The resting frequency of rhinolophid bats is kept with high accuracy; standard deviation of calls around the resting frequency was 0.20% in R. rouxi (Schuller 1980) and 0.113% in R. ferrumequinum (Zhang et al. 2019). By comparison, resting frequency of hipposiderid species shows more variation. Standard deviation of the resting frequency was 0.50% for H. speoris and 0.75% for H. bicolor (Schuller 1980). However, Zhang et al. (2019) reported a high precision in H. armiger of 0.165%, similar to the results of Schoeppler et al. (2018) for the same species (0.17%).
Pattern of variation in HDC echolocation between rhinolophids and hipposiderids
The data summarised here show differences in signal and auditory characters between the two families. Rhinolophid bats typically call at high duty cycles by giving long duration signals with low pulse repetition rates, have narrowly tuned auditory fovea, and almost full DSC. In contrast, hipposiderid bats typically call at lower duty cycles by producing moderate duration signals with high pulse repetition rates. The auditory fovea of hipposiderids is more broadly tuned than that of rhinolophids and they perform less well at DSC.
Prediction of ecological consequences of differences in HDC echolocation
The evidence presented in the previous section indicates that the HDC echolocation system of rhinolophids is adapted to enable them to exploit insects flying within areas of dense vegetation, a resource that is underexploited. Long duration pure tone signals produced at high duty cycles and a very narrowly tuned auditory fovea facilitated by DSC allow these bats to capture fluttering insects while overcoming environmental clutter. Following the logic for the functioning of the specialised HDC system in rhinolophid bats, the shorter CF signals, more broadly tuned auditory filter, individual variation in CF2 frequency and only partial compensation for Doppler-shifts demonstrate that hipposiderids are not restricted to focus the CF2 echo in a very narrowly tuned auditory fovea (Foeller and Kössl 2000). As a result, if the interpretation of the consequences for foraging of HDC echolocation in rhinolophids is correct, hipposiderids should not be restricted to hunting fluttering insects in dense vegetation. Based this understanding, predictions of expected differences in aspects of foraging ecology between the two families are outlined in Table 1.
Table 1.
Predicted differences in prey capture behaviour, foraging habitat and diet between rhinolophid and hipposiderid bats
Hipposiderids may have an entirely unique approach to foraging. Recently, Zhang et al. (2019) suggested that the broader auditory fovea and DSC of hipposiderids could provide higher tolerance to changes in Doppler shifts. This, in turn, may enable them to more effectively capture fast evasive prey in relatively open habitats without needing to significantly adjust their vocalisations to overcome Doppler shift changes in echoes (Zhang et al. 2019).
Comparison of prey capture behaviour, foraging habitat and diet in rhinolophids and hipposiderids
Most research on HDC echolocators has occurred in controlled environments, including the laboratory and flight tents/rooms (e.g. Fawcett et al. 2015; Schoeppler et al. 2018; Zhang et al. 2019). Field-based research has been comparatively limited, with most data coming from studies of bat assemblages (e.g. Aldridge and Rautenbach 1987; Rakotoarivelo et al. 2007). Relatively few single-species studies have been carried out on rhinolophid and hipposiderid bats foraging in the wild. However, sufficient information is available to provide a preliminary assessment of the predictions given in Table 1.
Prey perception
The sensory cues involved in prey capture have been assessed for three species each of Rhinolophidae and Hipposideridae (Table 2). Observations were carried out in the laboratory (Link et al. 1986), a field tent (Siemers and Ivanova 2004) or on free-flying bats (Bell and Fenton 1984). Each species was able to detect flying and fluttering insects; however, only H. bicolor captured walking insects (Table 2). Cockroach nymphs were gleaned from the ground by H. bicolor during 24 of 33 trials (Link et al. 1986). These observations support the prediction that hipposiderid bats can use a wider range of sensory cues than rhinolophid bats, including the capture of non-flying prey.
Table 2.
A summary of insect activities that initiated an attack response in three species each of Rhinolophus and Hipposideros
Prey capture behaviour
Quantified observations of prey capture behaviour in wild bats are available for seven species of Rhinolophus and eight hipposiderid species (Table 3). The majority of studies have used radio-telemetry as the method to track bats; however, light tagging and repeated observation of known individuals at foraging sites have also been used.
Table 3.
Prey capture behaviour of rhinolophid and hipposiderid bats in the wild
Species are arranged in decreasing order of size within genera. Data for R. philippinensis, R. hipposideros and H. caffer are the number of prey captures observed, whereas data for the other species are the number of foraging bouts (or bat passes) during which the particular prey capture behaviour was observed. NS, sample size not stated in source; ✓, prey capture method used but not quantified; ×, prey capture method not used
Rhinolophids and hipposiderids both forage using continuous flight and/or perch hunting with continuous flight being the dominant behaviour of each species, except for the two largest hipposiderid species (forearm length >78 mm). Rhinolophids and hipposiderids captured prey by aerial hawking or by gleaning from surfaces, including the ground (Table 3). All species, except for Asellia tridens, captured prey mostly by aerial hawking. Asellia tridens hunted mostly by gleaning from vegetation with occasional aerial hawking (Amichai et al. 2013). Four species from each family captured insects by gleaning from vegetation. Three rhinolophid species and two hipposiderid species captured prey from the ground. In addition, ground gleaning has been observed or inferred from dietary data for R. ferrumequinum (Ransome 1990; Ahmim and Moali 2013), R. euryale (Ahmim and Moali 2013) and R. blasii (Ahmim and Moali 2013). No other method of prey capture has been recorded for these families.
In summary, the available data on prey capture behaviour provide no support for the prediction that hipposiderid bats use a larger range of behaviours than rhinolophids. Although the experimental data presented in Table 2 indicated that only H. bicolor could capture walking prey, species of Rhinolophus are clearly able to glean prey from the ground.
Foraging habitat
Research on foraging by HDC bats in the wild enables a comparison of foraging habitat on the basis of the amount of clutter within foraging habitat (Table 4). For those studies where foraging areas were based on microhabitat use, as defined by Aldridge and Rautenbach (1987), areas within 2 m of vegetation or water surfaces were considered to be high in clutter. These included the following microhabitats of Aldridge and Rautenbach (1987); Zone 2, within 0.5 m of water surface; Zone 4, within stands of vegetation but >0.5 m from foliage; Zone 5, within stands of vegetation but ≤0.5 m from vegetation; Zone 6, surfaces of foliage; and Zone 7, within foliage. The classification of ‘open space’ used by Pavey and Burwell (2000) referred to foraging areas within stands of vegetation that were >0.5 m but <2.0 m from vegetation; therefore, such areas are classified as being high in clutter for the purposes of the current assessment.
Table 4.
The percentage of foraging time spent by rhinolophid and hipposiderid bats in areas with high (within and at edge of stands of vegetation) and low (open) levels of clutter
Ticks and crosses are used to show use or avoidance, respectively, of areas with high or low levels of clutter where there is no quantification. ‘Amount of clutter’ category is based on total foraging time
Of a sample of seven species of Rhinolophus, each species favoured foraging areas that were classified as having high levels of clutter (Table 4). Only three of these species were observed foraging in areas classified as having low levels of clutter. Rhinolophus ferrumequinum showed seasonal variation in use of clutter, avoiding low clutter foraging areas in spring but using them frequently in autumn (Jones and Morton 1992) or summer/autumn (Bontadina et al. 1995).
The same pattern of foraging in areas with high levels of clutter is shown in studies that do not quantify habitat use and in studies that measure activity using bat detectors. Cluttered areas are not necessarily restricted to stands of continuous vegetation, with some Rhinolophus species exploiting edge habitat such as hedgerows, edges of stands of forest and internal forest edges (e.g. Bontadina et al. 2002; Law and Chidel 2002; Russo et al. 2002, 2005; Goiti et al. 2008; Jiang et al. 2008; Lee et al. 2012, 2020; Law et al. 2020).
As with rhinolophid bats, each species in the sample of hipposiderid bats favoured foraging areas that were classified as having high levels of clutter (Table 4). Similar to Rhinolophus, three species were also observed foraging in areas classified as having low levels of clutter.
The preference for areas with high levels of clutter by hipposiderid species is demonstrated by assessments of habitat preference. For example, Hipposideros aff. ruber foraging in a mosaic of agricultural land and remnant forest in Ghana, West Africa, selected seminatural habitats and wooded savannah while avoiding cocoa farms and grass savannah (Nkrumah et al. 2016). The preference for high levels of clutter is also shown by the large species of Macronycteris and Hipposideros that forage solely or mostly by perch hunting (e.g. Pavey 1998b; Razafimanahaka et al. 2016). Macronycteris commersoni foraging in a forest–agricultural mosaic in eastern Madagascar selected sheltered littoral forest over agricultural land and sea-inundated forest (Razafimanahaka et al. 2016).
In summary, the information collected to date does not show a difference in use of habitat by the two families based on the levels of clutter. Hipposiderid bats, including the larger species, which have no similar sized equivalents among Rhinolophus, show a preference for foraging in areas of high clutter as do rhinolophid species. Although Pavey et al. (2001a) suggested that hipposiderid bats may spend more time foraging in edge habitats, at a microhabitat scale they select areas of high clutter within habitat mosaics.
Diet
Studies that assessed the diet of rhinolophid and hipposiderid bats were collated to compare the diets of the two families. I used published research that reported a sample size of at least 20 faecal pellets (or at least five stomachs of individual bats) per species, providing a dietary sample for 15 Rhinolophus species and 12 hipposiderid bats. Some species were covered by multiple studies. For example, I found eight papers on the diet of Rhinolophus hipposideros that matched my criteria (Supplementary Material).
All species assessed captured predominantly flying insects (Fig. 3 and Supplementary Material). Apart from flying insects, several other prey groups were captured by both families but always in low numbers (Fig. 3). Rhinolophus species captured spiders, centipedes and isopods. Four species (R. ferrumequinum, R. euryale, R. blasii, R. hipposideros) captured centipedes, all in the Great Kabylia region of north-east Algeria (Ahmim and Moali 2013). Rhinolophus hipposideros also captured spiders and isopods. Isopod predation was restricted to one study in Britain and Ireland (Williams et al. 2011), whereas spider predation took place in four of eight studies (Supplementary Material) and an additional three studies listed by Mitschunas and Wagner (2015). The other species recorded taking spiders was R. ferrumequinum (Ahmim and Moali 2013).
Fig. 3.
Number of rhinolophid (n = 15 species) and hipposiderid (n = 12 species) bats that captured each of six prey categories. Summary data in tabular form are given in the Supplementary Material. Data sources: Ahmim and Moali (2013), Amichai et al. (2013), Arlettaz et al. (2000), Beck et al. (1989), Bontadina et al. (2008), Bowie et al. (1999), Eckrich and Neuweiler (1988), Goiti et al. (2004), Jacobs (2000), Jacobs et al. (2007), Jones (1990), Li et al. (2007), Lino et al. (2014), Loumassine et al. (2019), Ma et al. (2008), McAney and Fairley (1989), Milne et al. (2016), Mitschunas and Wagner (2015), Pavey (1999), Pavey and Burwell (1997), Pavey and Burwell (2000), Pavey and Burwell (2004), Pavey et al. (2001a; 2001b), Rakotoarivelo et al. (2009), Salsamendi et al. (2008, 2012), Shi et al. (2009), Thabah et al. (2006), Vestjens and Hall (1977), Weterings et al. (2015), Whitaker and Black (1976), Williams et al. (2011).
Hipposiderid species were recorded consuming insect larvae, spiders and scorpions. Insect larvae were prey of H. caffer in a single study (Whitaker and Black 1976). Likewise, scorpion predation was noted only for Macronycteris commersoni in Madagascar (Rakotoarivelo et al. 2009). However, spider predation was noted for seven of the 12 hipposiderid species (Fig. 3): H. diadema (Pavey and Burwell 1997), H. khasiana (Thabah et al. 2006), H. speoris (Pavey et al. 2001a, 2001b), H. cervinus (Pavey and Burwell 2000), H. caffer (Bowie et al. 1999), H. ater (Milne et al. 2016) and Asellia tridens (Loumassine et al. 2019). One species of Hipposideros also captured birds (Pavey and Burwell 1997).
The dietary information summarised here provides the unexpected result that species of Rhinolophus are not restricted to capturing winged insects. While spiders may be captured aerially while they are ballooning this behaviour is mostly restricted to small individuals (Sutter 1999) and bats are most likely to glean them from webs, foliage or the ground. Centipedes and isopods are terrestrial and flightless and must be gleaned from the ground. There appears to be no alternative situation in which Rhinolophus could encounter these two groups of invertebrates.
Although further investigation is needed, the dietary data summarised here do not indicate that rhinolophid bats have a restricted prey base. Species from both families captured flying and flightless invertebrates. While this range of prey is lower than the extreme diversity shown by, for example, the terrestrial foraging New Zealand lesser short-tailed bat (Mystacina tuberculata) in wet forests (Czenze et al. 2018), both families captured a range of arthropods.
Summary
An important premise in interpretations of the evolutionary ecology of insectivorous bats is that both their echolocation signals and auditory capacities are adapted to the acoustical constraints of their foraging environment (Neuweiler 2000). Based on this relationship, it is expected that the echolocation and neurobiological differences between rhinolophid and hipposiderid bats, identified above, should result in differences in their prey capture behaviour, foraging habitat and diet. The major finding from this review is that there is no discernible difference in prey capture behaviour, foraging habitat and diet based on available information. Rather, the data summarised here indicate that rhinolophids and hipposiderids occupy similar foraging habitat and exploit the same prey base, but each family uses a distinct approach to echolocation to enable it to do so. Further insights on this issue will require more field-based investigations that combine assessment of echolocation and foraging ecology (prey capture, foraging habitat and diet) of species from both families. An understanding of the echolocation and foraging ecology of the little known and enigmatic species in the Rhinonycteridae may also prove fruitful. There is a need to seek further clarification on how laboratory assessments of audition translate into foraging performance in the wild. Evidence of predation on centipedes by four species of rhinolophid demonstrates that these bats do capture non-fluttering prey (Ahmim and Moali 2013) and challenges explanations of the importance of fluttering target detection in prey capture by this family.
Acknowledgements
I am grateful for time spent at various stages of my career in the laboratories of Professor Gerhard Neuweiler (LMU, Munich) and Professor Fritz Geiser (University of New England) while developing the ideas presented in this manuscript. I thank many colleagues for discussing components of this research with me, particularly Gimme Walter, James Fullard, Manfred Kössl, Elisabeth Foeller, Jan Grunwald, Gerhard Neuweiler, David Jacobs, and Mark Brigham. I am grateful for input on this manuscript received from Eric Vanderduys, Damian Milne, Brock Fenton and Kyle Armstrong. Funding was received from the Humboldt Foundation.
