Laryngeally echolocating bats produce a rapid succession of echolocation calls just before landing. These landing buzzes exhibit an increase in call rate and a decrease in call peak frequency and duration relative to pre-buzz calls, and resemble the terminal buzz phase calls of an aerial hawking bat's echolocation attack sequence. Sonar strobe groups (SSGs) are clustered sequences of non-buzz calls whose pulse intervals (PIs) are fairly regular and shorter than the PIs both before and after the cluster, but longer than the PIs of buzz calls. Like buzzes, SSGs are thought to indicate increased auditory attention. We recorded the echolocation calls emitted by juvenile big brown bats (Eptesicus fuscus) over postnatal development from birth to 32 days old, when full flight has normally been achieved, and tested the following hypotheses: (i) buzz production precedes the onset of controlled, powered flight; (ii) the emission of SSGs precedes buzzes and coincides with the onset of fluttering behaviour; and (iii) the onset of flight is attained first by young bats with adult-like wing loadings. We found that E. fuscus pups emitted landing buzzes before they achieved powered flight and produced SSGs several days before emitting landing buzzes. Both observations indicate that the onset of adult-like echolocation behaviour occurs prior to adult-like flight behaviour. Pups that achieved flight first were typically those that also first achieved low, adult-like wing loadings. Our results demonstrate that echolocation and flight develop in parallel but may be temporally offset, such that the sensory system precedes the locomotory system during postnatal ontogeny.
Bats are uniquely characterized as the only mammals capable of powered flight. While a minority of today's bat species do not echolocate, the common ancestor of all bats was probably both a laryngeal echolocator and an adept flier (Thiagavel et al., 2018). As such, the first bat species and the majority of chiropteran species alive today reflect an ancient transition (>65 mya) from a non-volant, vision-reliant, nocturnal mammal to one almost exclusively reliant on flight using echolocation to orient and find food (Griffin, 1958; Maor et al., 2017; Thiagavel et al., 2018). The ontogeny of laryngeally produced echolocation calls has been studied in a number of bat species (e.g. Brown et al., 1983; Balcombe, 1990; Jones et al., 1991; de Fanis and Jones, 1995; Kunz and Robson, 1995; Moss et al., 1997; Zhang et al., 2005; Mayberry and Faure, 2015; Mehdizadeh et al., 2018), providing a comprehensive understanding of when specific acoustic features appear in the vocal repertoire of pups. Some features include the ubiquitous nature of isolation calls: relatively long duration, low frequency, multi-harmonic vocalizations made in the first few days of life which facilitate infant–mother communication (Balcombe, 1990; Gelfand and McCracken, 1986; Bohn et al., 2007; Mayberry and Faure, 2015). Adult-like echolocation call designs and the emission of rapid buzzes are present in some bat species at 1 month of age, around the time flight is first achieved (Jones et al., 1991; Moss et al., 1997; de Fanis and Jones, 1995).
Powered flight, laryngeal echolocation and the ability to produce an acoustic buzz (i.e. a series of rapidly produced, short duration vocalizations) were three key innovations that probably allowed the first bats to exploit the, then unrealized, foraging niche of nocturnal flying insects (Griffin, 1958; Ratcliffe et al., 2013). Documenting flight and vocalization ontogenies is crucial for understanding the concurrent development of this integrated sensorimotor system. The ability of individual bats to achieve powered flight is associated with body size and wing morphology. Wing loading [WL=(m×g)/S, N m−2] describes the relationship of body mass (m, kg) multiplied by the net acceleration due to Earth's standard gravity (g=9.81 m s−2) divided by the wing area (S, m2) (Findley et al., 1972; Norberg and Rayner, 1987; Norberg and Fenton, 1988; Adams, 1996; Stern et al., 1997). All else being equal, animals with lower wing loadings produce greater lift that allows slower, more manoeuvrable flight than those with higher wing loadings (Norberg and Rayner, 1987). Wing aspect ratio (henceforth referred to as AR; AR=B2/S) is the ratio of wing span (B) to wing area (S); bats with long, narrow wings have higher ARs than bats with short, wide wings. During development, juvenile bats have higher wing loadings and lower aspect ratios than adults because their wing area is less than that of adult bats. Wing area also initially increases more rapidly than wing span, but both measures reach near-adult values around the time of a pup's first sustained flight, around 24 days after birth (O'Farrell and Studier, 1973; Powers et al., 1991; de Fanis and Jones, 1995; Papadimitriou et al., 1996; McLean and Speakman, 2000). It is not entirely clear how wing development relates to echolocation development and flight transitions.
Echolocation works by comparing temporal and spectral differences between outgoing calls and reflected echoes to infer information about the environment (Neuweiler, 1990; Moss and Surlykke, 2010). Echolocation is an active and dynamic sensing process where bats alter not only the design of individual call emissions (e.g. peak frequency, bandwidth, duration, beam width) but also their temporal patterning to maximize information acquisition for goal-directed behaviour and perception (Moss and Surlykke, 2001; Moss et al., 2006). Bats decrease the pulse interval (PI) between successive vocalizations with decreasing distance to a target during hunting and landing manoeuvres. For example, in free flight, the big brown bat, Eptesicus fuscus (Palisot de Beauvois 1796), emits search phase calls up to 20 ms in duration with PIs ranging from 20 to >100 ms, but during the terminal buzz phase they decrease call duration and PI to <1 ms and <13 ms, respectively (Surlykke and Moss, 2000; Moss and Surlykke, 2001). Most laryngeally echolocating bats also emit a rapid burst of short duration calls with PIs <13 ms (i.e. at rates over 76 calls s−1) when landing on surfaces – the so-called landing buzz (Melcón et al., 2007; Ratcliffe et al., 2013). Landing buzzes closely resemble the initial component of buzzes emitted by bats during aerial hawking attack sequences (Griffin et al., 1960; Tian and Schnitzler, 1997; Melcón et al., 2007) and those buzzes emitted prior to drinking (Greif and Siemers, 2010). It is believed that landing buzzes help bats to accurately and efficiently land on surfaces (Melcón et al., 2007). Among mammals, the ability to emit calls at the rate observed during landing, feeding and drinking buzzes reflects a speed of vocal-motor control unique to laryngeally echolocating bats (Elemans et al., 2011).
Additional evidence that bats actively control the temporal patterning of their calls is seen by sonar strobing. A sonar strobe/sound group (SSG) is a sequence of clustered calls with short PIs bracketed by calls with longer PIs. When an SSG contains ≥3 calls, the PIs within the group are remarkably stable (Moss et al., 2006). Bats emit SSGs during the search and approach phases but not the terminal phase of echolocation, and when faced with challenging tasks, or those that require increased attention. For example, bats commuting from a roost do not emit SSGs, whereas bats foraging in the field and the lab do (Kothari et al., 2014). The use of relatively constant PIs within the SSG is thought to reflect increased auditory attention to sharpen the bat's neural representation of spatial information (Petrites et al., 2009; Hulgard and Ratcliffe, 2016). Similarly, both the frequency of occurrence and number of calls emitted per SSG increases with task complexity (Moss et al., 2006; Hulgard and Ratcliffe, 2016).
The co-development of vocal and motor abilities has been observed in many different species, including humans (Abney et al., 2014). Although the timing of flight and vocal development has, independently, been well studied in juvenile bats, their concurrent maturity with respect to landing buzz emissions, the onset of SSG production and flight milestone transitions has not been systematically explored. Eptesicus fuscus is arguably the world's most thoroughly studied bat species. To our knowledge, the developmental trajectory of buzz call production has yet to be systematically explored in any bat, including E. fuscus (but see Moss et al., 1997, for observations on buzz production in juvenile little brown bats, Myotis lucifugus). Previous authors mentioning landing buzzes in E. fuscus pups reported that they occurred at the offset of every flight (Buchler, 1980; Brown et al., 1983); however, those studies looked exclusively at landing buzzes emitted close to a pup's first flights (i.e. 3–4 weeks after birth). Here, we report on the development of laryngeal echolocation – both call design and the temporal pattern of emission – and the acquisition of controlled, powered flight in the same bat pups over the first 32 days of postnatal life. We sought to determine at what age landing buzzes and SSGs first appear and become indistinguishable from those emitted by adults. We also sought to reveal developmental correlations between flight ability, wing morphology and the production of landing buzzes.
Specifically, we predicted that the development of landing buzzes and adult-like echolocation calls would precede powered flight. We also predicted that the emission of SSGs would precede landing buzzes, with SSGs first observed around the time that bats transition between flopping flight versus fluttering to the ground. We reasoned that echolocation ability would develop more quickly than flight ability, hence the timing of SSG production would correspond to the onset of wing beating behaviour in pups. We also tested two additional and non-mutually exclusive hypotheses by predicting that bats that moved quickly through the various flight phase transitions would develop echolocation ability and reach adult-like (i.e. lower) wing loadings faster than conspecifics that transitioned through the same flight phases more slowly.
MATERIALS AND METHODS
Study animals and trial conditions
Our study was conducted at McMaster University, Hamilton, ON, Canada. All procedures met the guidelines for the care/use of wild animals in research as per the Canadian Council on Animal Care and were approved by the Animal Research Ethics Board of McMaster University. Data were collected from eight male E. fuscus born in captivity to wild-caught mothers in 2017 (March to May). The day of birth was defined as postnatal day zero (PND 0). Because the colony was checked every once 24 h, it is possible that some pup births were first recorded on the day following parturition (but see ‘Morphological data collection and other pertinent observations’, below, for further details). Pup vocalizations were recorded every day or every second day between PND 1 and PND 32. We stopped recording after PND 32 because at this age E. fuscus have reached adult size and are exclusively producing adult-like vocalizations (Kurta and Baker, 1990; Moss et al., 1997; Mayberry and Faure, 2015). Pups not discovered on their date of birth and/or deemed too vulnerable to risk maternal separation were first recorded on PND 2.
To ensure we had at least one matched set of adequate acoustic and video files per animal per recording day, several trials were conducted each day and a single trial with the best matched audio and video recordings was used for analysis. Multiple trials were necessary because bats did not always make flight attempts during recordings or they would fly and land in an area not suitably monitored by our microphones and cameras. When not tested, pups and mothers were housed together in stainless steel wire (1/4 inch mesh) holding cages (28×22×18 cm; l×w×h) in a temperature- and humidity-controlled room and were provided with food (mealworms, Tenebrio molitor) and water ad libitum (Skrinyer et al., 2017).
Acoustic data collection
Acoustic recordings were collected in a flight room (4.9×3.3×3.3 m) in which the ceiling, walls and floor were lined with sound attenuating foam (Sonex® Classic, Pinta Acoustic, Minneapolis, MN, USA). Recordings took place in darkness and two different microphone set-ups were used. Initially, six CM16 microphones [Avisoft Bioacoustics, Glienicke, Germany; frequency sensitivity 2–200 kHz, frequency response approximately flat (±3 dB) from 25 to 140 kHz] were used. Midway through, the set-up was changed to nine CM16 microphones. In both configurations, the microphones were connected to an Avisoft USGH 1216 A/D converter (16-bit). Thus, bat calls were first sampled at 375 kHz per channel and then, after the microphone configuration changed, they were sampled at 250 kHz per channel to accommodate the three additional microphones.
The quality of the audio and video data collected in the set-ups was indistinguishable and therefore data from both configurations were used in our analysis. Microphones were placed throughout the flight room and aimed towards the airspace where bats were expected to fall/fly (Fig. 1). Acoustic files were 11 s in duration (1 s pre-trigger, 10 s hold time) to record all relevant information, as pups would often take a few seconds to initiate a flight attempt. Signals were saved to a ThinkPad X240 laptop computer (Lenovo, Morrisville, NC, USA).
Bats were hand released 60 cm above a table covered with foam. We did this to safely break the fall of young pups that were still unable to fly. This translates to a release point 130 cm above the floor and in line with our highest recording microphone (Fig. 1A,C). The release site was positioned 30 cm from the wall. Bats departed from the researcher's hand of their own volition. Young pups simply fell from the researcher's hand onto the foam-lined table, while older pups often took flight immediately. For each trial, we noted the category of flight exhibited by the pup (see ‘Flight data and analysis’, below, for our flight classification key). Upon landing and cessation of movement, the bat's horizontal distance from the release site was quantified with a measuring tape. We also set up reference points with two parallel ‘hot pack lines’ (two hand warmers per line; Hot Hands, Kobayashi, Dalton, GA, USA) to measure landing distances on images recorded by one or two thermal cameras (T480, FLIR, Wilsonville, OR, USA), depending on camera availability. Hot pack lines ran parallel to the release site, with the heat packs positioned 1 and 2 m from the wall closest to the release site (Fig. 1C).
Flight data and analysis
After every trial we made notes on a pup's flight attempts and movements. Flight ability at different ages was classified into one of four categories using criteria established by Powers et al. (1991): (i) flopping flight: pups fell straight down upon leaving the researcher's hand and exhibited no wing movements or horizontal displacement; (ii) fluttering flight: pups fell straight down upon leaving the researcher's hand and exhibited stretching or flapping wing movements but exhibited no horizontal displacement; (iii) flapping flight: pups exhibited wing flapping and achieved horizontal displacement upon leaving the researcher’s hand but with no control over their descent; and (iv) powered flight: pups exhibited sustained (true) powered flight upon leaving the researcher's hand, with horizontal displacement and clear evidence of path control (e.g. turning).
Trials in which the animal did not leave the researcher's hand or it flew to/landed in areas of the recording room not adequately monitored were not analysed. Of the remaining files, acoustic analysis was conducted on those trials where bats landed in areas covered by our microphone set-up. Because multiple microphones were used in every trial, calls were recorded on multiple channels at different points along the pup's flight trajectory. We typically analysed signals recorded closest to the landing site and/or with the highest signal-to-noise ratio. Clipped signals were not analysed.
Once signals to be analysed were identified for each trial, we assigned calls emitted by pups to one of three phases: (i) pre-flight, defined as the 500 ms period before the pup left the researcher's hand; (ii) in-flight, defined as the time between when the pup left the researcher's hand and when it first contacted the landing surface; and (iii) post-flight, defined as the 500 ms period after the pup first contacted the landing surface. We then used oscillogram, power spectrum and spectrogram displays (BatSound software v.4.2, Pettersson Electronik AB, Uppsala, Sweden) to manually measure temporal and spectral parameters of pup calls. From the oscillograms, we measured signal duration (ms; defined as the time between signal onset and offset) and pulse interval (ms; defined as the time from the onset of one call to the onset of the next call). From the spectrogram, we measured maximum fundamental frequency (kHz; defined as the highest frequency of the fundamental), minimum fundamental frequency (kHz; defined as the lowest frequency of the fundamental) and fundamental bandwidth (kHz; defined as the highest frequency minus the lowest frequency of the fundamental FM signal). From the power spectrum (automatic fast Fourier transform function, size 1024, Hann window) of each call, we measured peak frequency (kHz; defined as the frequency of maximum energy). We also estimated the number of harmonics in each call using information combined from spectrogram and power spectrum displays. Sound analysis settings and displays were kept constant for all analyses to allow for comparison between individuals and developmental milestones.
The presence or absence of landing buzzes and SSGs was noted and, if present, the landing buzz duration and number of calls per SSG were recorded. Landing buzzes are sequences of calls characterized by decreasing signal duration and peak frequency, with PIs <13 ms (i.e. call rates ≥75 Hz). A SSG is a temporal cluster of calls with short, stable PIs bracketed by calls with longer PIs. We used the island criterion and stability criterion for the identification and quantification of SSGs. The island criterion identifies the temporal isolation of a SSG within a continuous stream of biosonar emissions, whereas the stability criterion identifies the nearly constant PIs within a SSG (see fig. 1 in Kothari et al., 2014). When a SSG group consists of ≥3 calls, the individual PIs within the group must be stable (i.e. ≤5% deviation from mean PI within a cluster) and shorter than the PIs flanking the group (i.e. flanking PIs ≥1.2 times mean PI within a cluster; Moss et al., 2006). Note that the stability criterion cannot be used to identify a doublet SSG.
Our analysis primarily focused on calls emitted in-flight because these vocalizations were most relevant to our study. We tested for differences in emitted call parameters for pups at different ages (i.e. between PNDs 2, 12, 22 and 32) and flight ability phases (i.e. between each pup's transition from flopping to fluttering, fluttering to flapping, and flapping to flying phases; see ‘Flight data and analysis’, above).
Morphological data collection and other pertinent observations
We recorded morphological, developmental and behavioural milestones for each pup in every trial: age (PND), forearm length (mm), mass (g), eye status (open/closed) and whether the pup was attached to its mother when removed from the cage (yes/no). These milestones helped to ensure that pup development was healthy, normal and conformed to growth trajectories reported in the literature (Kurta and Baker, 1990; Mayberry and Faure, 2015). Forearm length and mass were also used to corroborate date of birth when pups were born in between daily colony checks (Mayberry and Faure, 2015).
Photos of the left wing of each pup were taken every recording day and analysed using ImageJ (National Institutes of Health, Bethesda, MD, USA) to measure the half wingspan (distance from the tip of the left wing to the midline of the torso) and half wing area (area of the entire left wing, including the left halves of the body and tail membrane). Measurements were taken in triplicate before they were averaged and doubled to estimate the full wingspan (B) and full wing area (S), respectively.
We calculated the aspect ratio and relative wing loading of pups on each recording day. As noted above, aspect ratio is a dimensionless number that describes wing narrowness (AR=B2/S). Wing loading describes lift and flight manoeuvrability [WL=(m×g)/S] (Findley et al., 1972; Norberg and Rayner, 1987; Norberg and Fenton, 1988; Adams, 1996; Stern et al., 1997). Because wing loading increases with body mass in geometrically similar animals, for scaling reasons, larger bats will have higher wing loadings than geometrically similar but smaller bats (Norberg and Fenton, 1988). To correct for scaling effects, we used relative wing loading (RWL=WL/m1/3) because this index is independent of body size (Norberg and Fenton, 1988). As with our acoustic data, we compared ARs and RWLs of pups across developmental ages (i.e. PND 2, 12, 22 and 32) and behavioural milestones (i.e. flight transition phases).
To reveal potential developmental relationships that exist between wing morphology, echolocation behaviour and flight proficiency, we ranked the eight pups with respect to RWL on PND 2, 12, 22 and 32. We then compared the ranks with respect to the age when pups transitioned from flopping to fluttering, fluttering to flapping, and flapping to flying phases, and also to the PND ranks when SSGs and landing buzzes were first observed. When comparing ranks of different variables, we selected variables that temporally corresponded with one another. In other words, we compared the age of milestone achievements with other, subsequent milestones and/or the most reasonable PNDs around which the milestones occurred. For example, bats transitioned from fluttering to flapping flight around PND 17. We compared these flight rankings with RWL score rankings on PND 12 and 22 because these morphological rankings were most temporally similar to the behavioural rankings. For comparisons in which both variables were measured in PNDs (i.e. consecutive flight transitions, or flight transitions versus age at which bats first show landing buzzes and SSGs with ≥3 calls), we used raw ages to explore linear relationships instead of ranked data.
Captive versus wild-born pups
We compared the forearm length, mass and RWL of captive-born pups on PND 32 with those of their mothers (see Mayberry and Faure, 2015) and PND 32 pup forearm length and mass with those of wild-caught, volant adult and juvenile male E. fuscus captured from field sites in southern Ontario. Wing span, AR and RWL were not measured in wild-caught bats. We also compared the age of first flight in our pups with published reports of wild E. fuscus (Kurta and Baker, 1990).
Data are reported as the mean±s.d. Our analysis of pup calls focused on temporal and spectral parameters providing the most insight into SSG and landing buzz development (i.e. minimum PI, maximum and minimum call duration, maximum peak frequency, minimum number of call harmonics, maximum fundamental bandwidth and maximum number of calls per SSG). As previous literature has described the developmental trajectories of echolocation behaviours and morphometrics (e.g. Buchler, 1980; Brown et al., 1983; Mayberry and Faure, 2015), we were able to make reasonable predictions regarding how these variables would change between consecutive time points. As such, we used pre-planned, one-tailed, paired t-tests to compare call parameters measured in-flight and morphological measurements (body mass, wing span, RWL and AR) for selected PNDs (i.e. PND 2 versus PND 12, PND 12 versus PND 22, and PND 22 versus PND 32) and between consecutive flight transitions (i.e. last day of flopping versus last day of fluttering behaviour, and last day of fluttering versus last day of flapping behaviour). We also used one-tailed, paired t-tests to compare the ages at which pups transitioned to new flight behaviours.
For those comparisons in which we had no a priori information, we used one-way between-subjects analyses of variance (ANOVA). Specifically, we used ANOVA to compare the ages at which pups first produced both landing buzzes and SSGs with ≥3 calls across the three phases of pup flight. We used two-sample t-tests to compare parameters between pups and adults. We used Spearman's rank and Pearson's correlation coefficients to identify relationships between individual pup development with respect to flight transitions, in-flight ≥3-call SSG production, in-flight landing buzz production and RWL scores. Pearson's correlation coefficients were used when the variables being compared were both measured in PNDs, whereas Spearman's rank correlation coefficients were used when variables were of different types (i.e. PND and RWL score). Alpha (α) values were adjusted using Bonferroni corrections (Rice, 1989).
We compared in-flight call parameters across sequential pup ages at 10 day intervals (i.e. PND 2 versus 12, PND 12 versus 22, and PND 22 versus 32), and between PND 32 pups and their mothers (Table 1, Fig. 2). All call parameters of PND 2 pups differed from those of PND 12 pups. Minimum call duration, maximum peak frequency, maximum fundamental bandwidth and minimum PI differed between PND 12 and 22 pups. When comparing PND 22 and PND 32 pups, only the maximum fundamental bandwidth differed. PND 2 pups did not emit SSGs or buzzes. Average in-flight call parameters of PND 32 pups and their mothers were indistinguishable from one another.
Very young E. fuscus pups (PND 1 and 2) always exhibited flopping behaviour upon leaving the researcher's hand. On average, pups transitioned from flopping to fluttering by 5.6±2.5 days. Fluttering pups subsequently transitioned to flapping behaviour by 16.4±3.8 days, and pups achieved true powered flight by 23.9±4.3 days. As expected, pups were significantly younger on the last day of flopping behaviour compared with the last day of fluttering behaviour and were also younger on the last day of fluttering behaviour compared with the last day of flapping behaviour (paired t-tests: flopping versus fluttering: t=6.45, P<0.001; fluttering versus flapping: t=4.23, P<0.002). All pups transitioned from flapping to fluttering flight and from fluttering to flapping flight; however, only 7 of 8 pups transitioned from flapping to powered flight in our trials.
To document how calls changed across successive flight classifications, we compared the temporal and spectral parameters of pup calls recorded in-flight on the final day when pups transitioned between the flight ability categories described above (Table 2, Fig. 3). We found differences in the minimum number of call harmonics, maximum fundamental bandwidth, maximum peak frequency, minimum PI and maximum number of calls within an SSG between calls emitted on the final day of flopping flight and those emitted on the final day of fluttering flight. In contrast, only the maximum peak frequency differed between calls emitted on the final day of fluttering flight and the final day of flapping flight.
Very young pups emitted vocalizations with no discernible temporal pattern (i.e. no obvious clusters or groupings). However, over time, pups started emitting bursts of vocalizations clustered into SSGs with a larger number of calls per group (Table 1, Fig. 4). We focused our analysis on the ages when pups began to emit SSGs with ≥3 calls because these SSGs satisfy both the island criterion and stability criterion (see Kothari et al., 2014). There were no age differences when pups first began emitting SSGs with ≥3 calls across the three flight phases (one-way between-subjects ANOVA, F=3.42, P=0.052). Specifically, pups first emitted SSGs with ≥3 calls at 11.75±6.84 days during the pre-flight phase, 6.38±2.56 days during the in-flight phase and 6.38±3.78 days during the post-flight phase.
The average age when pups first began to emit adult-like landing buzzes in-flight was 17.38±4.93 days. By PND 32, landing buzzes emitted by pups were indistinguishable from those emitted by mothers with respect to minimum PI and buzz duration (Table 1, Fig. 4). All pups emitted landing buzzes during the in-flight phase; however, some also emitted pre-flight and post-flight buzzes (i.e. some pups emitted buzzes before leaving the researcher’s hand and/or after landing). Of eight pups examined, four emitted at least one pre-flight buzz at an average age of 20±8.6 PND, and seven emitted at least one post-flight buzz at an average age of 19.9±6.9 PND. There were no age differences when pups first emitted landing buzzes between the different flight phases (one-way between-subjects ANOVA, F=0.35, P=0.708). The timing of in-flight buzzes with respect to landing varied with age; younger pups often produced landing buzzes hundreds of milliseconds prior to landing, while older pups tended to produce in-flight buzzes immediately before landing. Pre- and post-flight buzzes were also indistinguishable from adult landing buzzes with respect to minimum PI and buzz duration. Pre-flight buzzes were first produced at an age on (N=2) or after (N=2) the pup had already emitted an in-flight landing buzz. Of seven pups that emitted post-flight buzzes, five produced them after their first in-flight landing buzz, and two produced them either on the same day (N=1) or 3 days prior to emitting their first in-flight landing buzz (N=1).
Landing buzz duration varied throughout development and across the different flight categories. Fluttering pups emitted landing buzzes with the longest duration (69.1±48.9 ms) compared with the intermediate buzz duration emitted during flapping (58.25±39.7 ms) and powered flight (54.0±40.0 ms). A single PND 10 pup emitted the shortest duration landing buzzes in flopping flight (17.5±6.2 ms), but these buzzes contained only one or two PIs that reached the threshold for defining a landing buzz (i.e. PI <13 ms). Although buzzes emitted in the fluttering, flapping and powered flight categories were longer in duration than those in the flopping flight category (paired t-tests: t=6.87, P<0.001 for flopping versus fluttering; t=8.88, P<0.001 for flopping versus flapping; t=7.58, P<0.001 for flopping versus flying), there were no differences in buzz duration among bats that had achieved the three most advanced flight capabilities (paired t-tests: t=−1.43, P=0.158 for fluttering versus flapping; t=−0.84, P=0.402 for flapping versus flying; t=−1.94, P=0.056 for fluttering versus flying).
Because wing morphology directly affects when a bat first achieves and maintains powered flight, we correlated wingspan, mass, RWL and AR with preselected PND ages (i.e. PND 2, 12, 22 and 32; Table 3) and flight transition days (i.e. last day of flopping, fluttering and flapping flight; Table 4). All morphometric measurements differed between PND 2 and 12 pups, and all but AR differed between PND 12 and 22 pups. Only forearm length continued to increase between PND 22 and 32 pups; all other measures appeared to plateau (Table 3). Wing AR remained constant throughout flight transitions (Table 4), and RWL differences were only observed between the final days of flopping and final fluttering flight. Body mass and wingspan continually increased throughout flight development; we found differences for both measurements between the last day of flopping versus fluttering flight, and the last day of fluttering versus flapping flight (Table 4).
We also wanted to determine whether our oldest pups (PND 32) were morphologically distinct from their mothers as well as from wild-caught juvenile male and wild-caught adult male E. fuscus (Table 3). There was no difference in mass or forearm length between PND 32 pups and wild-caught adult or juvenile males (adult data from Mayberry and Faure, 2015). Moreover, forearm length, AR and RWL of PND 32 pups did not differ from those of their mothers; however, PND 32 pups did have lower masses and shorter wingspans compared with their long-captive mothers (Table 3).
Finally, we explored relationships between when pups accomplish locomotory and vocal developmental milestones by comparing the average ages when pups achieve flight transitions and first emit in-flight SSGs (≥3 calls per SSG) or landing buzzes. We found that pups transition from flopping to fluttering flight at roughly the same age (5.63±2.45 days) when they first emit in-flight SSGs with ≥3 calls per group (6.38±2.56 days; paired t-tests, t=−0.56, P=0.591). Furthermore, the age at which pups transition from fluttering to flapping flight (16.5±3.85 days) did not differ from when pups first produced in-flight landing buzzes (17.38±4.93 days; t=−0.41, P=0.696). We also wanted to determine whether pups that quickly reached flight milestones also reached acoustic milestones more quickly, and whether pups that quickly reached an adult-like morphology also achieved flight and echolocation milestones more quickly. For each comparison, we used either the raw PND or the relative rankings of each pup, with higher ranks representing more advanced progression (i.e. earlier accomplishment of SSG or landing buzzer production, flight ability or adult-like RWL). We found that the age when a pup transitioned from flapping to powered flight was strongly correlated with the RWL score on PND 22 (Spearman's rank correlation, R=0.97, P<0.001). All other correlations were not significant.
The developmental trajectories of echolocation call design and flight ability observed in this study corroborate previous observations collected independently on bat pup vocal development and flight in vespertilionid and other laryngeal echolocating bats (Balcombe, 1990; Jones et al., 1991; De Fanis and Jones, 1995; Moss et al., 1997; Zhang et al., 2005; Mayberry and Faure, 2015). That is, we observed the same changes from isolation call production in very young pups to adult-like echolocation calls over the first 32 days of life (Moss et al., 1997), as well as transitions between historically described flight categories in pups to adult-like powered flight over this same time period (Powers et al., 1991).
By PND 6, pups had transitioned from flopping flight (i.e. falling with no wing movements) to fluttering flight (i.e. wing movements that enable a soft landing). Although fluttering flight does not result in horizontal displacement, the wing movements serve as a precursor to more advanced flying abilities (Powers et al., 1991). Interestingly, pups also started emitting their first SSGs with ≥3 calls at the same time. After reaching fluttering behaviour, pups effectively reduced the number of harmonic elements in their orientation calls and decreased their minimum call duration to adult levels measured in the same flight room (Tables 1 and 2; Mayberry and Faure, 2015). However, the peak frequency and fundamental bandwidth of non-buzz calls only reached adult values around the time that pups transitioned to flapping flight (Tables 1 and 2). Controlled, powered flight was first observed in bats that were ∼24 days old. These results demonstrate that the ability to emit SSGs in E. fuscus pups precedes the ability to produce landing buzzes. Thus, adult-like echolocation call designs and temporal patterns of emission appear to develop before controlled, powered flight. Interestingly, however, pups that emitted SSGs and/or buzzes earlier in development did not also necessarily transition earlier through the various flight categories.
Because our data were obtained from captive-born pups, it is possible that the results may not apply to E. fuscus and other bats raised in the wild. For example, captive bats typically have ad libitum access to food, so they may gain mass more rapidly and have more resources to allocate to laryngeal and/or neural development compared with individuals in nature. Alternatively, laboratory-born pups may have fewer opportunities to listen to the vocalizations of flying and/or landing adult bats so they may acquire echolocation skills more slowly (Prat et al., 2015). However, field and lab developmental studies on other bat species do not differ dramatically with respect to the ontogeny of echolocation call design and buzz emission (Scherrer and Wilkinson, 1993; Moss et al., 1997). Similarly, ab libitum access to nutrition may promote faster flight development, but the confines of captivity could provide fewer opportunities for young pups to practice flying, plus a greater wing loading may slow development. In our study, pups had attained controlled, powered flight by PND ∼24 and this closely matches reports from the wild (PND ∼25; Kurta and Baker, 1990). Although adult E. fuscus kept in long-term captivity can weigh more than age-matched adults from the wild (Mayberry and Faure, 2015), we found no difference in mass or RWL between captive-bred male pups at PND 32 compared with adult or juvenile volant males captured nearby (Mayberry and Faure, 2015).
Eptesicus fuscus pups reach adult proportions approximately 30 days after birth (Kurta and Baker, 1990; Mayberry and Faure, 2015). Over this time, we observed a transition from intermittently produced, long duration, low frequency, multi-harmonic vocalizations (i.e. isolation and/or rudimentary echolocation calls) to more frequently and regularly produced echolocation calls with shorter duration, broader fundamental bandwidth and fewer harmonics (Fig. 2, Table 1). By PND 32, pups were emitting orientation calls indistinguishable from those of adults (Table 1, Figs 2 and 3). These changes, consistent across trial phases, matched the general trends of previous studies on pup vocal development (Brown et al., 1983; Balcombe, 1990; Jones et al., 1991; de Fanis and Jones, 1995; Moss et al., 1997; Zhang et al., 2005; Monroy et al., 2011; Mayberry and Faure, 2015; Mehdizadeh et al., 2018). Specifically, we observed that maximum and minimum call durations went from being relatively long and highly variable (Tables 1 and 2) on PND 2 through the flop-to-flutter transition (PND ∼6) to adult-like call durations before pups (i) reached PND 12 (Table 1) and (ii) transitioned from fluttering to flapping flight (PND ∼17; Table 2). Maximum peak frequency and maximum fundamental bandwidth only reached adult levels after pups began to emit SSG with ≥3 calls, around the same time they began to emit landing buzzes, and just before transitioning to powered flight (Table 2).
From a biomechanical perspective, the order of the developmental changes makes sense. Similar to other laryngeally echolocating bats, the big brown bat larynx is hypertrophied relative to that of non-echolocating terrestrial mammals, due in large part to its massive cricothyroid musculature (Moss, 1988; Metzner and Schuller, 2010). Calls are produced as the cricothyroid muscle relaxes and the rate of relaxation influences frequency-modulated sweep rates, while the tension achieved during contraction determines the highest frequency of the fundamental element (Elemans et al., 2011; Ratcliffe et al., 2013). Our data suggest that over the first 32 days of life, the cricothyroid musculature obtains the ability to produce calls with adult-like durations before it acquires the ability to produce adult-like peak frequencies and fundamental bandwidths. In other words, the cricothyroid muscle can already relax sufficiently fast enough to produce short duration calls before it can achieve sufficient tension to produce higher peak frequencies and wider fundamental bandwidths. The fact that pups emitted calls with fewer harmonics between PND 6 and 12 demonstrates that the filtering properties of the pup's supralaryngeal vocal tract are present early in development. Functionally, the early decrease in call duration should translate to fewer instances of call–echo overlap, and thus improved echolocation in cluttered conditions. The concurrent decrease in the number of harmonics and increase in call peak frequency and fundamental bandwidth should maintain and/or improve object resolution and frequency-dependent timing.
SSGs were first identified in the echolocation call sequences of flying E. fuscus (Moss et al., 2006) and are produced most frequently when bats face perceptually challenging tasks (e.g. manoeuvring in clutter, capturing moving prey; Moss et al., 2006; Petrites et al., 2009; Kothari et al., 2014; Hulgard and Ratcliffe, 2016). We found that pups began emitting SSGs with ≥3 calls around PND 6, as they transitioned from flopping to fluttering flight. Previous studies have suggested that SSG production is synchronized with respiration and the wing beat cycle (Petrites et al., 2009; Hulgard and Ratcliffe, 2016), and the emergence of SSGs concurrent with the first wing movements supports this idea. Also, pups emitted the maximum number of calls in SSGs around the same time that they transitioned from fluttering to flapping flight, suggesting a link between wing beats and SSG production.
In our pups, SSG production always preceded landing buzzes, leaving open the possibility that SSGs are a precursor to landing buzz production. With respect to individual trajectories, we found no individual correlation between the production of SSGs and the production of first landing buzzes. In other words, pups that produced SSGs earliest in development did not also emit the first landing buzzes, suggesting that the shift from SSGs to buzz production is unidirectional, but not individually conserved. Interestingly, SSGs emitted on PND 12 contained as many calls per group as on PND 32, a period before pup calls had acquired adult-like peak frequencies and fundamental bandwidths (Table 1). Functionally, increasing the number of calls per SSG during the switch from wing movements on PND 6 to horizontal displacement on PND 17 may reflect selection on pups to acquire more information. Indeed, the subsequent increases in call peak frequency and bandwidth would also assist pups in acquiring finer details with echolocation (e.g. object size, texture, velocity).
Adult laryngeally echolocating bats emit buzzes when they are about to intercept airborne prey and land on or drink from surfaces (Griffin et al., 1960; Melcón et al., 2007; Greif and Siemers, 2010). Buzzes are characterized by a decrease in call duration, peak frequency and PI (Britton and Jones, 1999; Surlykke and Moss, 2000; Ratcliffe et al., 2013). Because buzz calls are emitted at higher rates, the returning echoes provide bats with faster updates about their surroundings (Petrites et al., 2009; Ratcliffe et al., 2013). In our study, six out of eight bats began emitting landing buzzes before taking flight, and four bats began to emit landing buzzes prior to transitioning from fluttering to flapping flight. One pup emitted his first landing buzz on the same day it transitioned from flopping to fluttering flight; however, this transition was noticeably later compared to other pups (∼4 days after the average transition time). In general, pups did not begin to emit landing buzzes or call sequences with PIs approaching those of landing buzzes until ∼11 days after they had begun emitting SSGs and achieving adult-like echolocation call designs.
Each echolocation call emitted by a laryngeally echolocating bat is under independent neuromuscular control. Producing calls at landing buzz rates demands that the cricothyroid muscles contract and relax at rates much higher than possible in striated muscles of most vertebrates. In the larynges of vespertilionid bats, the buzz is powered by so-called superfast muscles (Elemans et al., 2011). Given the ubiquity of buzzes in echolocating bats, our data from E. fuscus pups suggest that superfast muscles continue to develop postnatally, reaching their fast rates of contraction near PND 17 (i.e. approximately two-thirds of the way through development).
Controlled, powered flight
As in other vespertilionids, achieving controlled, powered flight was correlated with pups having reached adult-like morphologies, specifically RWL (Table 3; Hughes et al., 1995; Elangovan et al., 2007). Our PND 32 pups did not differ in mass or forearm length from wild adult or juvenile male E. fuscus (Mayberry and Faure, 2015), or from their mothers with respect to forearm length, RWL and orientation call design (Tables 1 and 3). Although our study was laboratory based, the fact that pups first emitted landing buzzes ∼6 days before they began to fly suggests this is also likely to occur in the wild. Indeed, in nature, juvenile E. fuscus achieve controlled, powered flight by PND ∼25 (Kurta and Baker, 1990), around the same time that our captive-bred male pups achieve the same milestone (PND ∼24).
We found no association between the transition to powered flight and any echolocation behaviour milestone, indicating that, in general, pups develop echolocation abilities slightly in advance of flight abilities. Two pups in our study achieved powered flight 1 day prior to the production of their first landing buzz. Because we defined buzzes as having PIs ≤13 ms, these individuals may have been emitting calls at near-buzz rates but without technically meeting our definition. Near-buzzes are likely to provide pups with comparably relevant informative. Indeed, pups that flew before emitting buzzes with PIs ≤13 ms had been emitting multi-call SSGs with PIs <15 and <18 ms, respectively, on or prior to their first day of flight.
A single pup in our study failed to transition from flapping to powered flight. This pup also had the highest and least adult-like RWL score and greatest mass on PND 32 compared with other pups (non-flier RWL=56.94 N m−2, average RWL=45.4±6.7 N m−2; non-flier mass=19.7 g, average mass=17±2.1 g). These data suggest that this pup may have been unable to achieve powered flight because of high transport costs associated with a higher mass and RWL (see below).
The change in mass, wingspan, RWL and AR in E. fuscus closely matches developmental trajectories described in previous studies (de Fanis and Jones, 1995; Hoying and Kunz, 1998; McLean and Speakman, 2000; Mayberry and Faure, 2015). We found that both mass and wingspan increased during development and plateaued between PND 22 and 32, while RWL decreased early in development and reached adult-like values shortly before a pup's first flight. The decrease in RWL results from differences between body mass and wing area growth; wing area increases faster than mass (Hughes et al., 1995; Stern et al., 1997). A decrease in RWL early in development also reflects the lowering of transport costs as pups begin to achieve sustained flight and has been observed in multiple species (O'Farrell and Studier, 1973; Buchler, 1980; Powers et al., 1991; Adams, 1996; Stern et al., 1997; McLean and Speakman, 2000). Transport costs of sustained flight directly relate to wing loading; as RWL decreases, the energy and power required for flight decrease as a result of proportionately larger wing areas relative to body mass (Powers et al., 1991). Transport costs often decrease further at the time of first flight because of a measurable reduction in mass (Hughes et al., 1995).
Wing AR showed a slight increase immediately after PND 2, but plateaued quickly (Table 3). The overall development consistency in AR has been reported (O'Farrell and Studier, 1973; de Fanis and Jones, 1995; Hughes et al., 1995) and reflects the parallel development of wingspan and wing area. As pup wings grow longer, they also increase in surface area. The change in AR immediately after birth may represent an initial divergence of wingspan and wing area; initially, pup wing area may increase faster than wingspan, resulting in a temporarily lower AR.
We observed younger pups emitting landing buzzes at non-optimal times. For example, buzzes emitted during the pre-flight and post-flight phases would not be helpful for acquiring updated environmental information immediately prior to landing. Pre-flight buzzes may reflect an attempt by the pups to acquire information prior to movement due to inexperience with motor procedures involved with landing, while post-flight buzzes could represent vocal practice and/or learning. Because our observations were restricted to a short period of time each day, pups may have emitted buzzes that were not documented.
There was no clear association between the age when pups first emitted buzzes and their flight ability milestones; pups that buzzed earlier did not necessarily acquire powered flight earlier. This suggests that vocal development and flight ability, while clearly related and dependent on one another, are not tightly linked developmentally in a step-wise fashion. Our results provide further information on the development of flight and echolocation abilities in aerially hawking bats. Further research on the ontogeny of terminal feeding buzzes during aerial hawking prey captures could provide additional insight into sensorimotor integration in bats.
We thank Coen Elemans, Andrew Mason and two anonymous reviewers for their constructive comments on the manuscript.
Conceptualization: H.W.M., P.A.F., J.M.R.; Methodology: H.W.M., P.A.F., J.M.R.; Formal analysis: H.W.M., P.A.F., J.M.R.; Investigation: H.W.M.; Resources: P.A.F., J.M.R.; Writing - original draft: H.W.M., P.A.F., J.M.R.; Writing - review & editing: H.W.M., P.A.F., J.M.R.; Visualization: H.W.M., P.A.F., J.M.R.; Supervision: P.A.F., J.M.R.; Project administration: P.A.F., J.M.R.; Funding acquisition: H.W.M., P.A.F., J.M.R.
This research was made possible by discovery grants (P.A.F., J.M.R.) and a post-graduate scholarship (H.W.M.) from the Natural Sciences and Engineering Research Council of Canada (NSERC) and a Canadian Institutes of Health Research (CIHR) grant to P.A.F.
The authors declare no competing or financial interests.